DFSORT

z/OS
DFSORT Application Programming Guide

Version 2 Release 2
SC23-6878-01
Note
Before using this information and the product it supports, be sure to read the general information under Notices on page
915.
Contents
Figures . . . . . . . . . . . . . . . ix
Tables . . . . . . . . . . . . . . . xi
About this document . . . . . . . . xiii
How to use this document . . . . . . .
Required product knowledge . . . . . .
Referenced documents. . . . . . . . .
z/OS information . . . . . . . . . .
Using LookAt to look up message explanations
Notational conventions . . . . . . . .
.
.
. xiii
. xiv
. . xv
. . xv
. . xvi
. . xvi
How to send your comments to IBM

If you have a technical problem .
If you have a technical problem .
.
.
.
.
.
.
.
.
xix
.
.
. xix
. xix
Summary of changes . . . . . . . . xxi

Summary of changes for SC23-6878-01 z/OS
Version 2 Release 2 . . . . . . . . . . . xxi
New information . . . . . . . . . . . xxi
Summary of changes for SC23-6878-00 z/OS
Version 2 Release 1 . . . . . . . . . . . xxii
New information . . . . . . . . . . . xxii
Operational changes that may require user
action. . . . . . . . . . . . . . . xxiv
Chapter 1. Introducing DFSORT . . . . 1

DFSORT overview . . . . . . . . . .
DFSORT on the Web. . . . . . . . . .
DFSORT FTP site . . . . . . . . . . .
Invoking DFSORT . . . . . . . . . .
How DFSORT works . . . . . . . . .
Operating systems . . . . . . . . .
Control fields and collating sequences . . .
Cultural environment considerations . . .
DFSORT processing . . . . . . . . .
Input data setsSORTIN and SORTINnn . .
Output data setsSORTOUT and OUTFIL . .
Data set considerations . . . . . . . .
Sorting or copying records . . . . . .
Merging records . . . . . . . . . .
Data set notes and limitations . . . . .
XTIOT, uncaptured UCBs and DSAB above 16
megabytes . . . . . . . . . . . . .
z/OS file system considerations . . . . .
Installation defaults. . . . . . . . . .
Migrating to DFSORT from other sort products
DFSORT messages and return codes . . . .
Use Blockset whenever possible . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
4
4
4
5
5
5
7
7
11
12
12
12
13
13
.
.
.
.
.
.
.
.
.
.
.
.
16
16
17
24
25
26
Chapter 2. Invoking DFSORT with Job

Control Language . . . . . . . . . . 27
Using the JCL . . . .
Using the JOB statement .
.
.
Copyright IBM Corp. 1973, 2015
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 27
. 29
Using SET and PROC symbols in DFSORT control

statements . . . . . . . . . . . . . . .
Using the EXEC statement . . . . . . . . .
Specifying EXEC statement cataloged procedures
Specifying EXEC/DFSPARM PARM options . .
Aliases for PARM options . . . . . . . .
Using DD statements . . . . . . . . . . .
Duplicate ddnames . . . . . . . . . . .
Shared tape units . . . . . . . . . . .
System DD statements. . . . . . . . . .
Program DD statements . . . . . . . . .
29
29
30
32
60
61
63
63
63
65
Chapter 3. Using DFSORT program

control statements . . . . . . . . . 81
Using program control statements . . . . . . . 81
Control statement summary . . . . . . . . . 81
Describing the primary task . . . . . . . . 81
Including or omitting records . . . . . . . 82
Reformatting and editing records . . . . . . 82
Producing multiple output and reports and
converting records . . . . . . . . . . . 82
Joining two files . . . . . . . . . . . . 82
Invoking additional functions and options . . . 83
Using symbols . . . . . . . . . . . . 83
General coding rules . . . . . . . . . . . 83
Continuation lines . . . . . . . . . . . 85
Inserting comment statements . . . . . . . 87
Coding restrictions . . . . . . . . . . . 88
ALTSEQ control statement . . . . . . . . . 88
Altering EBCDIC collating sequenceexamples
89
DEBUG control statement . . . . . . . . . 91
Specifying diagnostic optionsexamples . . . 95
END control statement . . . . . . . . . . 95
Discontinue reading control
statementsexamples . . . . . . . . . . 95
INCLUDE control statement . . . . . . . . . 96
Relational condition . . . . . . . . . . 99
Comparisons . . . . . . . . . . . . . 99
Including records in the output data
setcomparison examples . . . . . . . . 108
Substring comparison tests . . . . . . . . 110
setsubstring comparison example . . . . . 112
Bit logic tests . . . . . . . . . . . . 112
Method 1: Bit operator tests . . . . . . . 112
Padding and truncation . . . . . . . . . 114
Including records in the output data setbit
operator test examples . . . . . . . . . 114
Method 2: Bit comparison tests . . . . . . 115
Including records in the output data setbit
comparison test examples . . . . . . . . 116
Date comparisons . . . . . . . . . . . 117
Including records in the output data setdate
comparisons . . . . . . . . . . . . . 120
Numeric tests . . . . . . . . . . . . 120
iii

set--numeric tests . . . . . . . . .
Alphanumeric tests . . . . . . . .
set--alphanumeric tests . . . . . . .
INCLUDE/OMIT statement notes . . .
INREC control statement . . . . . . .
INREC statement notes . . . . . . .
Reformatting records before processing
examples . . . . . . . . . . . .
JOINKEYS control statement . . . . . .
JOIN control statement . . . . . . . .
MERGE control statement . . . . . . .
Specifying a MERGE or COPYexamples .
MODS control statement . . . . . . .
Identifying user exit routinesexamples .
OMIT control statement . . . . . . . .
Omitting records from the output data
setexample . . . . . . . . . .
OPTION control statement . . . . . . .
Aliases for OPTION statement options . .
Specifying DFSORT options or
COPYexamples . . . . . . . . .
OUTFIL control statements . . . . . . .
OUTFIL statements notes . . . . . .
OUTFIL featuresexamples . . . . .
OUTREC control statement . . . . . . .
OUTREC statement notes . . . . . .
Reformatting records after processing
examples . . . . . . . . . . . .
RECORD control statement . . . . . . .
Describing the record format and
lengthexamples . . . . . . . . .
REFORMAT control statement. . . . . .
SORT control statement . . . . . . . .
SORT/MERGE statement notes . . . .
Specifying a SORT or COPYexamples .
SUM control statement . . . . . . . .
SUM statement notes . . . . . . . .
Adding summary fieldsexamples . . .
.
.
. 122
. 122
.
.
.
.
.
.
.
.
124
124
125
148
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
150
162
162
162
165
166
168
169
.
.
.
. 171
. 173
. 219
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 426
. 438
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
220
223
374
377
402
424
441
442
442
449
449
451
453
454
Chapter 4. Using a JOINKEYS

application for joining two files. . . . 457
Overview. . . . . . . . . . . . . . .
JOINKEYS application processing . . . . .
Sample JOINKEYS applications . . . . . .
JCL for a JOINKEYS application . . . . . . .
JOINKEYS statements . . . . . . . . . .
JOIN statement . . . . . . . . . . . . .
REFORMAT statement . . . . . . . . . .
JOINKEYS application notes . . . . . . . .
JOINKEYS application examples . . . . . . .
Example 1 - Paired F1/F2 records without
duplicates . . . . . . . . . . . . .
Example 2 - Paired F1/F2 records with
duplicates (cartesian) . . . . . . . . . .
Example 3 - Paired F1 records . . . . . . .
Example 4 - Unpaired F2 records . . . . . .
Example 5 - Paired and unpaired F1/F2 records
(indicator method). . . . . . . . . . .
iv
457
458
459
460
462
467
468
471
473
473
475
477
479
482
z/OS V2R2 DFSORT Application Programming Guide
Example 6 - Paired and unpaired F1/F2 records

(FILL method) . . . . . . . . . . . . 485
Chapter 5. Using your own user exit

routines . . . . . . . . . . . . . . 487
User exit routine overview . . . . . . . . .
DFSORT program phases . . . . . . . . .
Functions of routines at user exits . . . . . .
DFSORT input/user exit/output logic examples
Opening and initializing data sets . . . . .
Modifying control fields . . . . . . . . .
Inserting, deleting, and altering records . . .
Summing records . . . . . . . . . . .
Handling special I/O. . . . . . . . . .
VSAM user exit functions . . . . . . . .
Determining action when intermediate storage
is insufficient . . . . . . . . . . . .
Closing data sets . . . . . . . . . . .
Terminating DFSORT . . . . . . . . . .
32-bit and 64-bit parameter lists . . . . . . .
64-bit address terminology . . . . . . . .
Addressing and residence modes for user exits . .
How user exit routines affect DFSORT performance
Summary of rules for user exit routines . . . .
Loading user exit routines . . . . . . . .
User exit linkage conventions . . . . . . .
Dynamically binding or link-editing user exit
routines . . . . . . . . . . . . . .
Assembler user exit routines (input phase user
exits) . . . . . . . . . . . . . . . .
E11 user exit: opening data sets/initializing
routines . . . . . . . . . . . . . .
E15 user exit: passing or changing records for
sort and copy applications . . . . . . . .
E16 user exit: handling intermediate storage
miscalculation . . . . . . . . . . . .
E17 user exit: closing data sets . . . . . .
E18 user exit: handling input data sets . . . .
E19 user exit: handling output to work data sets
E61 user exit: modifying control fields . . . .
Assembler user exit routines (output phase user
exits) . . . . . . . . . . . . . . . .
E31 user exit: opening data sets/initializing
routines . . . . . . . . . . . . . .
E32 user exit: handling input to a merge only
E35 user exit: changing records . . . . . .
E37 user exit: closing data sets . . . . . .
E38 user exit: handling input data sets . . . .
E39 user exit: handling output data sets . . .
Sample E15 and E35 routines using the 64-bit
parameter lists . . . . . . . . . . . . .
Sample routines written in assembler using the
32-bit parameter lists . . . . . . . . . . .
E15 user exit: altering record length . . . . .
E16 user exit: sorting current records when
NMAX is exceeded . . . . . . . . . .
E35 user exit: altering record length . . . . .
E61 user exit: altering control fields . . . . .
COBOL user exit routines . . . . . . . . .
COBOL user exit requirements . . . . . .
COBOL user exit routines (input phase user exit)
487
488
490
490
492
492
492
492
492
493
493
493
493
493
494
494
495
495
496
496
498
498
499
499
503
503
504
507
508
510
510
510
512
517
517
517
518
518
518
519
519
520
521
521
523
COBOL E15 user exit: passing or changing

records for sort . . . . . . . . . . .
COBOL user exit routines (output phase user exit)
COBOL E35 user exit: changing records . .
Sample routines written in COBOL . . . . .
COBOL E15 user exit: altering records . . .
COBOL E35 user exit: inserting records. . .
E15/E35 return codes and EXITCK . . . . .
. 523
529
. 529
. 535
. 535
. 536
. 537
Chapter 6. Invoking DFSORT from a

program . . . . . . . . . . . . . 541
Invoking DFSORT dynamically . . . . . .
What are system macro instructions? . . . .
Using system macro instructions . . . . . .
Using JCL DD statements . . . . . . . .
Overriding DFSORT control statements from
programs . . . . . . . . . . . . . .
Invoking DFSORT with the 24-bit parameter list
Entry point name . . . . . . . . . .
Providing program control statements . . .
Invoking DFSORT with the extended (31-bit)
parameter list . . . . . . . . . . . .
Entry point name . . . . . . . . . .
Entry point name . . . . . . . . . .
64-bit address terminology . . . . . . .
Writing the macro instruction . . . . . . .
Parameter list examples . . . . . . . .
Restrictions for dynamic invocation . . . . .
Merge restriction . . . . . . . . . .
Copy restrictions . . . . . . . . . .
.
.
.
.
541
541
541
542
. 542
543
. 543
. 543
. 550
. 550
. 550
553
. 553
. 553
. 554
. 559
. 560
. 561
. 561
. 561
Chapter 7. Using ICETOOL. . . . . . 563

Overview. . . . . . . . . . . .
ICETOOL/DFSORT relationship . . .
ICETOOL JCL summary . . . . . .
ICETOOL operator summary . . . .
Complete ICETOOL examples . . . .
Using symbols . . . . . . . . .
Using SET and PROC symbols in control
statements . . . . . . . . . .
Invoking ICETOOL . . . . . . .
Putting ICETOOL to use. . . . . .
Job control language for ICETOOL . . .
JCL restrictions . . . . . . . . .
ICETOOL statements . . . . . . . .
General coding rules . . . . . . .
COPY operator . . . . . . . . . .
Operand descriptions. . . . . . .
Copy examples . . . . . . . . .
COPY operator with JOINKEYS example
COUNT operator . . . . . . . . .
COUNT examples . . . . . . . .
DATASORT operator . . . . . . . .
DATASORT examples . . . . . .
DEFAULTS operator . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
563
563
564
565
566
567
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
567
567
568
571
573
574
574
575
575
577
578
579
580
585
585
587
588
590
Operand descriptions. . . . . . . . .
DEFAULTS example . . . . . . . . .
DISPLAY operator . . . . . . . . . . .
Simple report . . . . . . . . . . .
Tailored report . . . . . . . . . . .
Sectioned report . . . . . . . . . .
MERGE operator . . . . . . . . . . .
MERGE examples . . . . . . . . . .
MODE operator . . . . . . . . . . .
MODE example . . . . . . . . . .
OCCUR operator . . . . . . . . . . .
Simple report . . . . . . . . . . .
Tailored report . . . . . . . . . . .
OCCUR examples . . . . . . . . . .
RANGE operator . . . . . . . . . . .
RANGE example . . . . . . . . . .
RESIZE operator . . . . . . . . . . .
RESIZE examples . . . . . . . . . .
SELECT operator . . . . . . . . . . .
SELECT examples . . . . . . . . . .
SORT operator . . . . . . . . . . . .
Sort examples . . . . . . . . . . .
SORT operator with JOINKEYS example . .
SPLICE operator . . . . . . . . . . .
SPLICE examples . . . . . . . . . .
STATS operator. . . . . . . . . . . .
STATS example. . . . . . . . . . .
SUBSET operator . . . . . . . . . . .
SUBSET examples . . . . . . . . . .
UNIQUE operator . . . . . . . . . . .
UNIQUE example . . . . . . . . . .
VERIFY operator . . . . . . . . . . .
VERIFY example . . . . . . . . . .
Calling ICETOOL from a program . . . . .
TOOLIN interface . . . . . . . . . .
Parameter list interface . . . . . . . .
ICETOOL notes and restrictions . . . . . .
Notes on using JOINKEYS with COPY and SORT
ICETOOL return codes . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
591
592
594
595
596
597
598
642
643
644
644
645
645
646
647
648
649
659
662
662
664
664
666
666
668
670
673
677
678
679
681
681
688
691
708
709
710
710
712
716
718
718
719
720
720
721
722
722
722
727
728
. 728
Chapter 8. Using symbols for fields

and constants . . . . . . . . . . . 731
Field and constant symbols overview
DFSORT example . . . . . .
SYMNAMES DD statement. . . .
SYMNOUT DD statement . . . .
SYMNAMES statements . . . . .
Comment and blank statements .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
731
732
733
734
734
734
Contents
Symbol statements . . . . . . . . .
Keyword statements . . . . . . . . .
Using SYMNOUT to check your SYMNAMES
statements . . . . . . . . . . . .
Using symbols in DFSORT statements . . . .
SORT and MERGE . . . . . . . . .
SUM . . . . . . . . . . . . . .
INCLUDE and OMIT. . . . . . . . .
INREC and OUTREC. . . . . . . . .
OUTFIL . . . . . . . . . . . . .
JOINKEYS . . . . . . . . . . . .
REFORMAT . . . . . . . . . . . .
OPTION . . . . . . . . . . . . .
Using symbols in ICETOOL operators . . . .
COUNT . . . . . . . . . . . . .
DATASORT . . . . . . . . . . . .
DISPLAY . . . . . . . . . . . . .
OCCUR . . . . . . . . . . . . .
RANGE . . . . . . . . . . . . .
SELECT . . . . . . . . . . . . .
SPLICE . . . . . . . . . . . . .
STATS, UNIQUE and VERIFY . . . . . .
SUBSET . . . . . . . . . . . . .
ICETOOL Example . . . . . . . . .
Using SET and PROC symbols in DFSORT and
ICETOOL statements . . . . . . . . . .
Using JPn parameters in EXEC PARM for
DFSORT . . . . . . . . . . . . .
Using JPn parameters in EXEC PARM for
ICETOOL . . . . . . . . . . . .
Description of JPn"string" . . . . . . .
Example 1 . . . . . . . . . . . .
Example 2 . . . . . . . . . . . .
Notes for symbols . . . . . . . . . . .
. 734
. 744
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
747
747
748
749
749
750
752
756
756
756
757
757
757
757
758
758
758
758
758
758
758
. 760
. 761
.
.
.
.
.
761
762
762
763
765
Chapter 9. Using extended function

support . . . . . . . . . . . . . . 767
Using EFS . . . . . . . . . . . . . .
Addressing and residence mode of the EFS
program . . . . . . . . . . . . . . .
How EFS works . . . . . . . . . . . .
DFSORT program phases . . . . . . . .
DFSORT calls to your EFS program . . . . .
What you can do with EFS . . . . . . . . .
Opening and initializing data sets . . . . .
Examining, altering, or ignoring control
statements . . . . . . . . . . . . .
Processing user-defined data types with EFS
program user exit routines . . . . . . . .
Supplying messages for printing to the message
data set . . . . . . . . . . . . . .
Terminating DFSORT . . . . . . . . . .
Closing data sets and housekeeping . . . . .
Structure of the EFS interface parameter list . . .
Action codes . . . . . . . . . . . .
Control statement request list . . . . . . .
Control statement string sent to the EFS
program . . . . . . . . . . . . . .
Control statement string returned by the EFS
program . . . . . . . . . . . . . .
vi
767
768
768
768
769
773
774
774
776
776
776
776
776
779
779
779
781
EFS formats for SORT, MERGE, INCLUDE, and

OMIT control statements . . . . . . . .
D1 format on FIELDS operand . . . . . .
D2 format on COND operand . . . . . . .
Length of original control statement . . . . .
Length of the altered control statement . . . .
EFS program context area . . . . . . . .
Extract buffer offsets list . . . . . . . . .
Record lengths list. . . . . . . . . . .
Information flags . . . . . . . . . . .
Message list . . . . . . . . . . . . .
EFS program exit routines . . . . . . . . .
EFS01 and EFS02 function description . . . .
EFS01 user exit routine . . . . . . . . .
EFS02 user exit routine . . . . . . . . .
Addressing and residence mode of EFS program
user exit routines . . . . . . . . . . .
EFS program return codes you must supply . . .
Record processing order . . . . . . . . . .
How to request a SNAP dump . . . . . . .
EFS program example . . . . . . . . . .
DFSORT initialization phase: . . . . . . .
DFSORT termination phase. . . . . . . .
783
783
784
785
785
785
785
786
786
788
788
789
789
790
792
793
793
795
796
796
798
Chapter 10. Improving efficiency . . . 799

Improving performance . . . . . . . . . .
Design your applications to maximize performance
Directly invoke DFSORT processing . . . . .
Plan ahead when designing new applications
Specify efficient sort/merge techniques . . . .
Specify input/output data set characteristics
accurately . . . . . . . . . . . . .
Use extended format data sets. . . . . . .
Use DFSMSrmm-managed tapes, or ICETPEX
Specify devices that improve elapsed time. . .
Use options that enhance performance . . . .
Use DFSORT's fast, efficient productivity
features . . . . . . . . . . . . . .
Avoid options that degrade performance . . .
Use main storage efficiently . . . . . . .
Allocate temporary work space efficiently . . .
Use Hipersorting . . . . . . . . . . .
Sort with data space . . . . . . . . . .
Use memory object sorting . . . . . . . .
Use ICEGENER instead of IEBGENER . . . .
ICEGENER return codes. . . . . . . . .
Use DFSORT's performance booster for The SAS
System . . . . . . . . . . . . . . .
Use DFSORT's BLDINDEX support . . . . . .
799
799
799
800
800
801
802
802
802
802
805
806
807
810
811
812
813
813
816
817
817
Chapter 11. Examples of DFSORT job

streams . . . . . . . . . . . . . . 819
Summary of examples . . . . .
Storage administrator examples . .
REXX examples . . . . . . .
CLIST examples . . . . . . .
Sort examples . . . . . . . .
Example 1. Sort with ALTSEQ . .
Example 2. Sort with OMIT, SUM,
DYNALLOC and ZDPRINT . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
OUTREC,
. . . .
.
.
.
.
.
.
819
819
820
821
821
821
. 822
Example 3. Sort with ASCII tapes . . . . .

Example 4. Sort with E15, E35, FILSZ,
AVGRLEN and DYNALLOC . . . . . . .
Example 5. Called sort with SORTCNTL,
CHALT, DYNALLOC and FILSZ . . . . . .
Example 6. Sort with VSAM input/output,
DFSPARM and option override . . . . . .
Example 7. Sort with COBOL E15, EXEC PARM
and MSGDDN . . . . . . . . . . . .
Example 8. Sort with dynamic link-editing of
exits . . . . . . . . . . . . . . .
Example 9. Sort with the extended parameter
list interface . . . . . . . . . . . . .
Example 10. Sort with OUTFIL . . . . . .
Example 11. Sort with Pipes and OUTFIL SPLIT
Example 12. Sort with INCLUDE and LOCALE
Example 13: Sort with z/OS UNIX files . . .
Example 14. Sort with IFTHEN . . . . . .
Example 15. Sort with 64-bit parameter lists,
E15, E35 and OUTFIL . . . . . . . . .
MERGE examples . . . . . . . . . . . .
Example 1. Merge with EQUALS . . . . . .
Example 2. Merge with LOCALE and OUTFIL
Copy examples . . . . . . . . . . . . .
Example 1. Copy with EXEC PARMs, SKIPREC,
MSGPRT and ABEND . . . . . . . . .
Example 2. Copy with INCLUDE and VLSHRT
Example 3. Copy with OUTREC, PARSE and
BUILD . . . . . . . . . . . . . .
ICEGENER example . . . . . . . . . . .
ICETOOL example . . . . . . . . . . .
824
825
826
827
828
829
831
833
835
836
837
838
839
844
844
845
846
847
847
848
849
850
Appendix A. Using work space . . . . 853

Introduction . . . . . . . . . . .
Central storage . . . . . . . . . .
Work data set devices . . . . . . .
Disk and tape devices . . . . . .
Number of devices . . . . . . .
Non-synchronous storage subsystems .
Allocation of work data sets . . . . .
Dynamic allocation of work data sets .
Dynamic over-allocation of work space .
JCL allocation of work data sets . . .
Disk capacity considerations . . . . .
Exceeding disk work space capacity . .
Tape capacity considerations . . . . .
Exceeding tape work space capacity . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
853
853
854
854
854
855
855
856
858
858
859
860
860
861
Appendix B. Specification/override of
DFSORT options . . . . . . . . . . 863
Main features of sources of DFSORT options .
DFSPARM data set . . . . . . . .

EXEC statement PARM options . . . .
SORTCNTL data set . . . . . . . .
SYSIN data set . . . . . . . . . .
Parameter lists . . . . . . . . . .
Override tables . . . . . . . . . .
Directly invoked DFSORT . . . . . . .
Notes to directly invoked DFSORT table .
Program invoked DFSORT with the extended
parameter list . . . . . . . . . . .
Notes to extended parameter list table . .
Program invoked DFSORT with the 24-bit
parameter list . . . . . . . . . . .
Notes to 24-bit list table . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
864
865
865
865
865
865
866
874
.
.
. 874
. 882
.
.
. 882
. 890
Appendix C. Data format descriptions
891
DFSORT data formats . . . . . .

Where DFSORT formats can be used .
DFSORT formats for COBOL data types
. 891
. 897
. 900
.
.
.
.
.
.
.
.
.
Appendix D. EBCDIC and ASCII

collating sequences . . . . . . . . 901
EBCDIC .
ASCII . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 901
. 903
Appendix E. DFSORT abend

processing . . . . . . . . . . . . 907
Checkpoint/restart . . . . . . . . . . . 907
DFSORT abend categories . . . . . . . . . 908
Abend recovery processing for unexpected abends 908
Processing of error abends with A-type messages
909
CTRx abend processing . . . . . . . . . . 909
Appendix F. Accessibility . . . . . . 911

Accessibility features . . . . . . .
Consult assistive technologies . . . .
Keyboard navigation of the user interface
Dotted decimal syntax diagrams . . .
Using assistive technologies . . . .
z/OS information . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
911
911
911
911
913
913
914
Notices . . . . . . . . . . . . . . 915
Programming interface information .
Trademarks . . . . . . . . .
.
.
.
.
.
.
.
.
. 916
. 916
Index . . . . . . . . . . . . . . . 917
. 864
Contents
vii
viii
Figures
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
Control Fields . . . . . . . . . . . . 6
Record Processing Order . . . . . . . . 9
Using ICETOOL to List Installation Defaults
18
Control Statement Format . . . . . . . . 84
Continuation Line Format. . . . . . . . 86
Sample Records . . . . . . . . . . . 110
OUTFIL Processing Order . . . . . . . 230
Examples of Notation for Binary Fields
444
JOINKEYS Application Processing . . . . 458
Examples of DFSORT Input/User
Exit/Output Logic . . . . . . . . . . 489
E18 User Exit Example . . . . . . . . 507
E15 DFSORT Interface with COBOL . . . . 525
LINKAGE SECTION Code Example for E15
(Fixed-Length Records) . . . . . . . . 526
(Variable-Length Record). . . . . . . . 526
E35 Interface with COBOL . . . . . . . 531
(Fixed-Length Records) . . . . . . . . 532
(Variable-Length Records) . . . . . . . 532
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
COBOL E15 Routine Example (FLR) . . . .

COBOL E35 Routine Example (VLR)
The 24-Bit Parameter List . . . . . . .
The Extended Parameter List . . . . . .
The 64-Bit Parameter List . . . . . . .
Coding a 24-Bit Parameter List. . . . . .
Coding an Extended Parameter List . . . .
Simple ICETOOL Job . . . . . . . . .
Parameter List for Parameter List Interface
ICETOOL Parameter List Interface Example
JCL for Parameter List Interface Program
Example . . . . . . . . . . . . .
Relationship Between DFSORT and an EFS
Program . . . . . . . . . . . . .
EFS Program Calls for a Sort . . . . . .
EFS Program Calls for a Merge or Copy
Control Statement Processing Sequence
EFS Interface Parameter List . . . . . .
Information Flags . . . . . . . . . .
DFSORT Register Convention . . . . . .
Calling Sequence to EFS02 by DFSORT
EFS Record Processing Sequence for a Sort or
Merge . . . . . . . . . . . . . .
EFS Record Processing Sequence for a Copy
Faster Sorting with COBOL . . . . . . .
536
537
545
551
555
560
561
567
722
726
727
768
769
770
775
778
787
789
791
794
795
804
ix
Tables
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
Related documents . . . . . . . . .
Referenced documents . . . . . . . .
Options That Can Ease Migration . . . .
FILSZ Variations Summary. . . . . . .
Aliases for MSGPRT/MSGCON Options
Aliases for PARM Options . . . . . .
DD Statement Parameters Used by DFSORT
DCB Subparameters Used by DFSORT . .
Compare Field Formats and Lengths
Permissable Field-to-Field Comparisons for
INCLUDE/OMIT (Group 1) . . . . .
Permissable Field-to-Field Comparisons for
INCLUDE/OMIT (Group 2) . . . . .
Permissible Field-to-Constant Comparisons
for INCLUDE/OMIT . . . . . . . .
Valid and Invalid Decimal Constants
Decimal Numbers for Current Date . . .
Decimal Numbers for Future Dates . . .
Decimal Numbers for Past Dates . . . .
Valid and Invalid Character String Constants
Valid and Invalid Strings with Double-Byte
Data . . . . . . . . . . . . .
Character Strings for Current Date . . .
Character Strings for Future Dates . . .
Character Strings for Past Dates . . . .
Valid and Invalid Hexadecimal Constants
Bit Comparison Example 2: Results for
Selected Field Values . . . . . . . .
Permissible Comparisons for Dates . . .
Logic Table for INCLUDE/OMIT. . . . .
Examples of Valid and Invalid Column
Alignment . . . . . . . . . . .
Examples of Valid and Invalid Blank
Separation . . . . . . . . . . .
Examples of Valid and Invalid Binary Zero
Separation . . . . . . . . . . .
Examples of Valid and Invalid Character
String Separation . . . . . . . . .
Examples of Valid and Invalid Hexadecimal
String Separation . . . . . . . . .
Example of DYNSPC Primary Space . . .
FILSZ Variations Summary . . . . . .
SIZE Variations Summary . . . . . .
SDB=LARGE Block Sizes for Tape Output
Data Sets . . . . . . . . . . . .
Aliases for OPTION Statement Options
Current date constants . . . . . . .
Future Date Constants . . . . . . .
Past Date Constants . . . . . . . .
Current time constants . . . . . . .
. xiv
. xv
. 24
. 42
47
. 60
61
. 62
100
. 101
. 101
. 102
103
. 103
. 104
. 104
104
.
.
.
.
105
105
106
107
107
. 114
. 115
. 116
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
. 117
. 119
. 124
74.
. 130
75.
. 131
76.
. 131
77.
. 131
78.
.
.
.
.
132
184
189
189
79.
80.
81.
82.
. 204
219
. 258
. 259
. 260
. 261
83.
84.
85.
86.
TRAN=ATOE ASCII-to-EBCDIC translation

TRAN=ETOA ASCII-to-EBCDIC translation
Edit Field Formats and Lengths . . . . .
Edit Mask Patterns. . . . . . . . . .
Edit Mask Signs . . . . . . . . . .
Digits Needed for Numeric Fields . . . .
Edit Mask Output Field Lengths . . . . .
To Output Field Lengths . . . . . . . .
Input and result fields for Yxx date editing
Input fields for p,m,Yxx date conversion
TOJUL and TOGREG output date fields
Output for weekdays . . . . . . . . .
Output for weeknum . . . . . . . . .
p,m,Yxx fields for date arithmetic . . . . .
Input and result fields for Yxx(s) date editing
Input and result fields for Yxx(s) date editing
Digits for TOTAL Fields . . . . . . . .
Control Field Formats and Lengths . . . .
Summary Field Formats and Lengths
Functions of Routines at Program User Exits
(Sort) . . . . . . . . . . . . . .
Functions of Routines at Program User Exits
(Copy and Merge) . . . . . . . . . .
Summary of 64-bit Terminology for Exits
Addressing Mode Flags for Exits in 64-bit
Parameter List . . . . . . . . . . .
32-bit E15 User Exit Parameter List . . . .
E15 Without a SORTIN Data Set . . . . .
E15 With a SORTIN Data Set Before End of
Input . . . . . . . . . . . . . .
E15 With a SORTIN Data Set After End of
Input . . . . . . . . . . . . . .
E35 With a SORTOUT or OUTFIL Data Set
Before End of Input . . . . . . . . .
E35 Without a SORTOUT or OUTFIL Data Set
Before End of Input . . . . . . . . .
E35 With a SORTOUT or OUTFIL Data Set
After End of Input . . . . . . . . . .
E35 without a SORTOUT or OUTFIL Data Set
After End of Input . . . . . . . . . .
Aliases for Message Option . . . . . . .
Summary of 64-bit Terminology for Programs
Obtaining Various Statistics . . . . . . .
Creating Multiple Versions/Combinations of
Data Sets . . . . . . . . . . . . .
JCL Statements for ICETOOL . . . . . .
Attributes of Edit Masks . . . . . . . .
Edit Mask Patterns. . . . . . . . . .
Return Area Lengths/Operation-Specific
Values . . . . . . . . . . . . . .
266
267
269
271
273
274
274
280
284
286
287
287
288
291
295
296
348
445
452
491
491
494
495
501
501
511
511
514
514
538
538
538
538
539
539
539
548
553
568
569
571
601
602
724
xi
87. Functions of an Extended Function Support

(EFS) Program . . . . . . . . . . .
88. D1 Format Returned by an EFS Program
89. Correlator Identifier and D2 Format Returned
by an EFS Program . . . . . . . . .
90. Original and Altered Control Statements
91. Number of Tracks per Cylinder for Disk
Devices . . . . . . . . . . . . .
92. Minimum Storage Required for Various File
Sizes . . . . . . . . . . . . . .
93. Work Space Requirements for Various Input
Characteristics . . . . . . . . . . .
94. Number of Tracks per Cylinder for Disk
Devices . . . . . . . . . . . . .
xii
773
784
785
797
801
855
859
860
95. Work Space Requirements of the Various Tape

Techniques . . . . . . . . . . . .
96. Directly Invoked DFSORT Option
Specification/Override . . . . . . . .
97. Extended Parameter List DFSORT Option
98. 24-Bit List DFSORT Option
99. Allowed with Frequently Used Data Types
100. Allowed with Other Data Types . . . . .
101. Equivalent DFSORT formats for various
COBOL data types . . . . . . . . . .
102. EBCDIC Collating Sequence . . . . . .
103. ASCII Collating Sequence . . . . . . .
860
867
875
883
897
898
900
901
903
About this document

This document is intended to help you to sort, merge, and copy data sets using
DFSORT. This document is not designed to teach you how to use DFSORT, but is
for programmers who already have a basic understanding of DFSORT, and need a
task-oriented guide and reference to its functions and options. If you are a new
user, then you should read z/OS DFSORT: Getting Started first. z/OS DFSORT:
Getting Started is a self-study guide that tells you what you need to know to begin
using DFSORT quickly, with step-by-step examples and illustrations.
How to use this document

The various sections of this document present related information grouped
according to tasks you want to do. The first three chapters of the document explain
what you need to know to invoke and use DFSORT's primary record-processing
functions. The remaining chapters explain more specialized features. The
appendixes provide specific information about various topics.
v Chapter 1, Introducing DFSORT, on page 1, presents an overview of DFSORT,
explaining what you can do with DFSORT and how you invoke DFSORT
processing. It describes how DFSORT works, discusses data set formats and
limitations, and explains how installation defaults can affect your DFSORT
applications.
v Chapter 2, Invoking DFSORT with Job Control Language, on page 27, explains
how to use job control language (JCL) to run your DFSORT jobs. It explains how
to code JOB, EXEC, and DD statements, and how you can use cataloged
procedures and EXEC PARM options to simplify your JCL and override
installation defaults.
v Chapter 3, Using DFSORT program control statements, on page 81, presents
the DFSORT control statements you use to sort, merge, and copy data. It
explains how to filter your data so you work only with the records you need,
how to edit data by reformatting and summing records, and how to produce
multiple output data sets and reports. It explains how to write statements that
direct DFSORT to use your own routines during processing.
v Chapter 4, Using a JOINKEYS application for joining two files, on page 457
explains how you can perform various types of "join" applications on two files
(F1 and F2) by one or more keys with DFSORT. It explains the JCL and control
statements you can use for JOINKEYS applications.
v Chapter 5, Using your own user exit routines, on page 487, describes how to
use DFSORT's program exits to call your own routines during program
processing. You can write routines to delete, insert, alter, and summarize records,
and you can incorporate your own error-recovery routines.
v Chapter 6, Invoking DFSORT from a program, on page 541, describes how you
use a system macro instruction to initiate DFSORT processing from your own
assembler program. It also lists specific restrictions on invoking DFSORT from
PL/I and COBOL.
v Chapter 7, Using ICETOOL, on page 563, describes how to use the
multi-purpose DFSORT utility ICETOOL. It explains the JCL and operators you
can use to perform a variety of tasks with ICETOOL.
v Chapter 8, Using symbols for fields and constants, on page 731, explains how
to define symbols and use them in DFSORT control statements and ICETOOL
operators.
xiii
v Chapter 9, Using extended function support, on page 767, explains how to use
the Extended Function Support (EFS) interface to tailor control statements, to
handle user-defined data types and collating sequences, and to have DFSORT
issue customized informational messages during processing.
v Chapter 10, Improving efficiency, on page 799, recommends ways with which
you can maximize DFSORT processing efficiency. This chapter covers a wide
spectrum of improvements you can make, from designing individual
applications for efficient processing at your site to using DFSORT features such
as Hipersorting, dataspace sorting, and ICEGENER.
v Chapter 11, Examples of DFSORT job streams, on page 819, contains annotated
example job streams for sorting, merging, and copying records.
v Appendix A, Using work space, on page 853, explains main storage
considerations and how to estimate the amount of intermediate storage you
might require when sorting data.
v Appendix B, Specification/override of DFSORT options, on page 863, contains
a series of tables you can use to find the order of override for similar options
that are specified in different sources.
v Appendix C, Data format descriptions, on page 891, gives examples of the
assembled data formats.
v Appendix D, EBCDIC and ASCII collating sequences, on page 901, lists the
collating sequences from low to high order for EBCDIC and ASCII characters.
v Appendix E, DFSORT abend processing, on page 907, describes the ESTAE
recovery routine for processing abends, and the Checkpoint/Restart facility.
v Notices on page 915, includes the notices, Programming Interface information,
and the trademark list.
Required product knowledge

To use this document effectively, you should be familiar with the following
information:
v Job control language (JCL)
v Data management
v Tape and disk hardware
You should also be familiar with the information presented in the following related
documents:
Table 1. Related documents
Document Order Number
xiv
z/OS DFSORT Messages, Codes and Diagnosis

Guide
SC23-6879
z/OS MVS JCL Reference
SA23-1385
z/OS MVS JCL User's Guide
SA23-1386
z/OS DFSMS Using Data Sets
SC23-6855
z/OS DFSMS Using Magnetic Tapes
SC23-6858
Referenced documents
This document refers to the following documents:
Table 2. Referenced documents
Order number
z/OS DFSMSdfp Checkpoint/Restart
SC23-6862
z/OS DFSMS Macro Instructions for Data Sets
SC23-6852
z/OS DFSMS Using Data Sets
SC23-6855
z/OS MVS JCL Reference
SA23-1385
z/OS MVS JCL User's Guide
SA23-1386
z/OS MVS Programming: Assembler Services

Reference IAR-XCT
SA23-1370
z/OS MVS Programming: Assembler Services

Guide
SA23-1368
z/OS Install/Migration page

(http://www.ibm.com/systems/z/os/zos/
installation/)
z/OS Install/Migration page

(http://www.ibm.com/systems/z/os/zos/
installation/)
z/OS V2R2.0 UNIX System Services User's

Guide
SA23-2279
The z/OS DFSORT Application Programming Guide is a part of a more extensive

DFSORT library. These documents can help you work with DFSORT more
effectively.
Publication Title
Order Number
Planning For and

Customizing DFSORT
z/OS DFSORT Installation and SC23-6881

Customization
Learning to Use DFSORT
z/OS DFSORT: Getting Started SC23-6880
Diagnosing Failures and

Interpreting Messages
z/OS DFSORT Messages,

Codes and Diagnosis Guide
SC23-6879
Tuning DFSORT
z/OS DFSORT Tuning Guide
SC23-6882
z/OS information
This information explains how z/OS references information in other documents
and on the web.
When possible, this information uses cross document links that go directly to the
topic in reference using shortened versions of the document title. For complete
titles and order numbers of the documents for all products that are part of z/OS,
see z/OS Information Roadmap.
To find the complete z/OS library, go to IBM Knowledge Center
(http://www.ibm.com/support/knowledgecenter/SSLTBW/welcome).
About this document
xv
Using LookAt to look up message explanations

LookAt is an online facility that lets you look up explanations for most of the IBM
messages you encounter, as well as for some system abends and codes. Using
LookAt to find information is faster than a conventional search because in most
cases LookAt goes directly to the message explanation.
You can use LookAt from these locations to find IBM message explanations for
z/OS elements and features, z/VM, z/VSE, and Clusters for AIX and Linux:
v The Internet. You can access IBM message explanations directly from the LookAt
Web site at www.ibm.com/servers/eserver/zseries/zos/bkserv/lookat/.
v Your z/OS TSO/E host system. You can install code on your z/OS systems to
access IBM message explanations using LookAt from a TSO/E command line
(for example: TSO/E prompt, ISPF, or z/OS UNIX System Services).
v Your Microsoft Windows workstation. You can install LookAt directly from the
z/OS and Software Products DVD Collection (SK3T-4271) and use it from the
resulting Windows graphical user interface (GUI). The command prompt (also
known as the DOS > command line) version can still be used from the directory
in which you install the Windows version of LookAt.
v Your wireless handheld device. You can use the LookAt Mobile Edition from
www.ibm.com/servers/eserver/zseries/zos/bkserv/lookat/lookatm.html with a
handheld device that has wireless access and an Internet browser.
You can obtain code to install LookAt on your host system or Microsoft Windows
workstation from:
v The z/OS and Software Products DVD Collection (SK3T-4271).
v The LookAt Web site (click Download and then select the platform, release,
collection, and location that suit your needs). More information is available in
the LOOKAT.ME files available during the download process.
Notational conventions
The syntax diagrams in this document are designed to make coding DFSORT
program control statements simple and unambiguous. The lines and arrows
represent a path or flowchart that connects operators, parameters, and delimiters in
the order and syntax in which they must appear in your completed statement.
Construct a statement by tracing a path through the appropriate diagram that
includes all the parameters you need, and code them in the order that the diagram
requires you to follow. Any path through the diagram gives you a correctly coded
statement, if you observe these conventions:
v Read the syntax diagrams from left to right and from top to bottom.
v Begin coding your statement at the spot marked with the double arrowhead.
v A single arrowhead at the end of a line indicates that the diagram continues on
the next line or at an indicated spot.
xvi
Notational Conventions
A continuation line begins with a single arrowhead.
v Strings in upper-case letters, and punctuation (parentheses, apostrophes, and so

on), must be coded exactly as shown.
Semicolons are interchangeable with commas in program control statements
and the EXEC PARM string. For clarity, only commas are shown in this
document.
v Strings in all lowercase letters represent information that you supply.
v Required parameters appear on the same horizontal line (the main path) as the
operator, while optional parameters appear in a branch below the main path.
Required

Optional
v Where you can make one choice between two or more parameters, the
alternatives are stacked vertically.
Operator
Required Choice 1
Required Choice 2
Required Choice 3

Optional Choice 1
Optional Choice 2
If one choice within the stack lies on the main path (as in this example, left), you
must specify one of the alternatives. If the stack is placed below the main path
(as in this example, right), then selections are optional, and you can choose
either one or none of them.
v The repeat symbol shows where you can return to an earlier position in the
syntax diagram to specify a parameter more than once (see the first example
later in this section), to specify more than one choice at a time from the same
stack (see the second example later in this section), or to nest parentheses (see
the third example later in this section).
a,b,c
Choice-1
Choice-2
Choice-3
Do not interpret a repeat symbol to mean that you can specify incompatible
parameters. For instance, do not specify both ABEND and NOABEND in the
same EXEC statement, or attempt to nest parentheses incorrectly.
Use any punctuation or delimiters that appear within the repeat symbol to
separate repeated items.
v A double arrowhead at the end of a line indicates the end of the syntax diagram.
About this document
xvii
Notational Conventions
xviii
How to send your comments to IBM

We appreciate your input on this publication. Feel free to comment on the clarity,
accuracy, and completeness of the information or provide any other feedback that
you have.
Use one of the following methods to send your comments:
1. Send an email to mhvrcfs@us.ibm.com.
2. Send an email from the "Contact us" web page for z/OS (http://
www.ibm.com/systems/z/os/zos/webqs.html).
Include the following information:
v Your name and address.
v Your email address.
v Your telephone or fax number.
v The publication title and order number:
SC23-6878-01
v The topic and page number that is related to your comment.
v The text of your comment.
When you send comments to IBM, you grant IBM a nonexclusive right to use or
distribute the comments in any way appropriate without incurring any obligation
to you.
IBM or any other organizations use the personal information that you supply to
contact you only about the issues that you submit.
If you have a technical problem

Do not use the feedback methods that are listed for sending comments. Instead,
take one of the following actions:
v Contact your IBM service representative.
v Call IBM technical support.
v Visit the IBM Support Portal at z/OS Support Portal (http://www-947.ibm.com/
systems/support/z/zos/).
If you have a technical problem

Do not use the feedback methods listed above. Instead, do one of the following:
v Contact your IBM service representative
v Call IBM technical support
v Visit the IBM support portal at z/OS Support Portal (http://www-947.ibm.com/
systems/support/z/zos/)
xix
xx
Summary of changes
Summary of changes for SC23-6878-01 z/OS Version 2 Release 2
This document contains information that was previously presented in z/OS
DFSORT Application Programming Guide, SC28-6878-00, which supported z/OS
Version 2 Release 1.
The following sections summarize the changes to that information.
The message ICE099A is enhanced to list the failing member name in case of an
error. Any automated actions based on the presence of this message should be
evaluated.
The text for existing DFSORT messages ICE000I and ICE288I has been changed.
Any automated actions based on the presence of these messages should be
evaluated.
New information
This edition includes the following new enhancements:
High Performance Ficon (HPF) Exploitation

A program that invokes DFSORT, ICETOOL or ICEGENER will now be able to
exploit the High Performance Ficon (zHPF) hardware. When zHPF is available,
DFSORT SORT or MERGE applications will use BSAM for SORTIN, SORTOUT,
and OUTFIL data sets.
Date functions
Two new date functions to calculate the week number of a given date and Age as
date duration are added.
WEEKNUM function converts a given Julian/Gregorian date to number of week.
There are 2 versions of this function, the standard USA format and the ISO format.
WEEKNUM=USA function returns an integer in the range of 1 to 54 that
represents the week of the year. The week starts with Sunday, and January 1 is
always in the first week.
WEEKNUM=ISO function returns an integer in the range of 1 to 53 that represents
the week of the year. The week starts with Monday and includes 7 days. Week 1 is
the first week of the year to contain a Thursday, which is equivalent to the first
week containing January 4.
AGE function returns a date duration that specifies the number of years, months,
and days between an input date and current date.
Age=YMD produces an 8-byte result which has duration in years (0-9999), months
(00-12), and days (00-31).
Age=YM produces a 6-byte result which has duration in years (0-9999), months
(00-12).
xxi
Age=YD produces a 7-byte result which has duration in years (0-9999), days
(00-366).
Message changes
The text for existing DFSORT messages has been changed. Any automated actions
based on the presence of these messages should be evaluated.
ICE000I
ICE099A
ICE121A
ICE288I
Summary of changes for SC23-6878-00 z/OS Version 2 Release 1

This document contains information that was previously presented in z/OS
DFSORT Application Programming Guide, SC26-7523-07, which supported z/OS
Version 1.
The following sections summarize the changes to that information.
New information
This edition includes the following new enhancements:
64-bit support
Eligible user programs and exits can now be written to:
v Call DFSORT from a program in 64-bit addressing mode (AMODE 64) using a
new 64-bit parameter list and the entry name ICEMAN64 or SORT64
v Use DFSORT E15, E35 and E32 exits running in 64-bit addressing mode
(AMODE 64)
v Pass 64-bit addressed records to DFSORT using new 64-bit parameter lists for
E15, E32 and E35 exits.
Performance and resource usage improvements

DFSORT has been enhanced to improve use of central storage in relation to system
activity. The existing EXPMAX, EXPOLD and EXPRES installation options will now
be evaluated in conjunction with available resources at run-time which can give
you better control over how DFSORT uses available central storage resources. The
likelihood of over committed central storage resources and excessive paging has
been reduced, which can provide improved reliability and performance for all
workloads including DFSORT.
The shipped default for EXPOLD is changed from MAX to 50%. The shipped
default for EXPRES is changed from 0 to 10%.
DFSORT has been enhanced to allocate storage in smaller increments and then
check resource availability before each additional increment is reserved. This can
allow concurrent sort applications to better share available storage and allows sorts
to stop allocating additional storage if resources are no longer available.
Additionally, DFSORT's dynamic allocation of work data sets will be adjusted to
reduce the likelihood of failure when a sort application is unable to obtain all of
the expected central storage and must then use more disk work space than
expected.
xxii
A new TUNE installation default allows you to specify whether DFSORT should
allocate storage in increments with additional disk work space to minimize the risk
of failure, or to allocate all storage at initialization so disk work space allocation
can be reduced.
Blockset merge applications can now use storage above 16 MB virtual with more
functions (such as E61, INREC, OUTREC, INCLUDE, OMIT and SUM), providing
improved performance and virtual storage constraint relief.
XTIOT, uncaptured UCBs and DSAB above 16 megabytes virtual

Work data sets dynamically allocated by DFSORT will use options for XTIOT,
uncaptured UCBs, and DSAB above 16 megabyte virtual to the extent that z/OS
supports them.
Extended address volumes

DFSORT support for EAS-eligible data set types on Extended Address Volumes has
been enhanced to increase the maximum size of a DFSORT work data set. With full
track blocking, the maximum number of tracks that can be used for a single work
data set has been increased from 1048576 to 16777215. In situations where DFSORT
must use a reduced block size for the work data sets, less than 16777215 tracks can
be used.
Alphanumeric comparison tests

New UC, LC, MC, UN, LN and MN keywords (similar to the previously available
NUM keyword) now allow you to test a field for various combinations of
alphanumeric characters or non-alphanumeric characters using binary (BI) format.
The new keywords allow you to select any of the following alphanumeric
character sets:
v UC: Uppercase characters (A-Z)
v
v
v
v
v
LC: Lowercase characters (a-z)

MC: Mixed case characters (A-Z, a-z)
UN: Uppercase and numeric characters (A-Z, 0-9)
LN: Lowercase and numeric characters (a-z, 0-9)
MN: Mixed case and numeric characters (A-Z, a-z, 0-9)
You can use these new keywords in the following comparison operands: COND,
INCLUDE, OMIT, BEGIN, END, WHEN and TRLID.
More PARSE fields

You can now use up to 1000 parsed fields (%0-%999) with the PARSE function; the
previous limit was 100 parsed fields (%0-%99).
Repeating PARSE fields

REPEAT=v is a new PARSE option that can be used to repeat a particular parse
field definition multiple times.
REPEAT=v can be used with % to specify v identically defined consecutive parsed
fields to be ignored.
REPEAT=v can be used with %n, %nn or %nnn to specify v identically defined
consecutive parsed fields for which data is to be extracted. The parsed fields will
start with the %n, %nn or %nnn field you select and be incremented by one for
each repeated parsed field.
Summary of changes
xxiii
Alphanumeric tests for PARSE fields

STARTAFT=an, STARTAT=an, ENDBEFR=an and ENDAT=an can now be used
with the PARSE function to start or end when a character from any of various
alphanumeric character sets is found. New keywords for an allow you to select
any of the following alphanumeric character sets:
v LC: Lowercase characters (a-z)
v MC: Mixed case characters (A-Z, a-z)
v
v
v
v

NUM: Numeric characters (0-9)
Insert a string at the end of variable-length records

VLTRAIL=string is a new OUTFIL option that allows you to insert a character
string (C'string') or hexadecimal string (X'yy...yy') at the end of each variable-length
OUTFIL output record.
Symbol enhancements
DFSORT Symbols can now be used for many more operands. The following
operands of the form KEYWORD=n can now be specified as KEYWORD=symbol
where symbol represents an equivalent number (for example, if you have
New_Length,25 in SYMNAMES, you can use LENGTH=New_Length wherever
you can use LENGTH=25): ABSPOS, ACCEPT, ADDPOS, AVGRLEN, DO,
ENDPOS, ENDREC, FIXLEN, ID, IFOUTLEN, INCR, LENGTH, LINES, MAXLEN,
RECORDS, REPEAT, SAMPLE, SEQ, SKIPREC, SPLIT1R, SPLITBY, START,
STARTPOS, STARTREC, STOPAFT and SUBPOS.
A symbol for a number can now be used for the length of the output field with the
CHANGE operand, and for the length of the sequence number with the SEQNUM
operand.
A symbol can be used for any of the new %000-%999 parsed fields.
A symbol can be used for string with the new VLTRAIL=string operand.
Improved diagnostics
DFSORT now provides specific reason codes and associated documentation to aid
in diagnosing and correcting errors associated with messages ICE083A and
ICE141A.
The text for message ICE118I has been changed to recommend the use of
FILSZ=En.
The ICE236I options-in-effect message now includes values for TUNE, EXPMAX,
EXPOLD and EXPRES.
The text for message ICE285A has been changed for clarification.
Operational changes that may require user action

The following are operational changes that may require user action for existing
DFSORT/ICETOOL applications that use certain functions as specified:
xxiv
Central storage controls

The IBM-supplied default for the existing EXPOLD installation option has been
changed to EXPOLD=50%. If you want DFSORT to use EXPOLD, as in previous
releases, you can set EXPOLD=MAX.
The IBM-supplied default for the existing EXPRES installation option has been
changed to EXPRES=10%. If you want DFSORT to use EXPRES, as in previous
releases, you can set EXPRES=0.
The IBM-supplied default for the new TUNE installation option is TUNE=STOR
which specifies allocation of available central storage as needed in increments sized
to balance resource usage for concurrent sorts. If you want DFSORT to allocate
available central storage using fixed sized increments, as in previous releases, you
can set TUNE=OLD.
Message changes
The text for existing DFSORT messages ICE000I, ICE083A, ICE118I, ICE141A,
ICE236I and ICE285A has been changed. Any automated actions based on the
presence of these messages should be evaluated.
New Reserved Words for Symbols

The following are new DFSORT/ICETOOL reserved words which are no longer
allowed as symbols: LC, LN, MC, MN, UC and UN.
If you used any of these words as a symbol previously, you must change them. For
example, if you used MC, you can change it to mc.
Summary of changes
xxv
xxvi
Chapter 1. Introducing DFSORT

DFSORT overview
This chapter introduces IBM z/OS DFSORT Licensed Program 5650-ZOS.
DFSORT is intended to run in problem state and in a user key ( that is, key 8 or
higher).
DFSORT is a program you use to sort, merge, and copy information.
v When you sort records, you arrange them in a particular sequence, choosing an
order more useful to you than the original one.
v When you merge records, you combine the contents of two or more previously
sorted data sets into one.
v When you copy records, you make an exact duplicate of each record in your data
set.
Merging records first requires that the input data sets are identically sorted for the
information you will use to merge them and that they are in the same order
required for output. You can merge up to 100 different data sets at a time.
In addition to the three basic functions, you can perform other processing
simultaneously:
You can control which records to keep in the final output data set of a DFSORT
run by using INCLUDE and OMIT statements in your application. These
statements work like filters, testing each record against criteria that you supply
and retaining only the ones you want for the output data set. For example, you
might choose to work only with records that have a value of Kuala Lumpur in
the field reserved for office location. Or perhaps you want to leave out any record
dated after 2001 if it also contains a value greater than 20 for the number of
employees.
You can parse, edit, and reformat your records before or after other processing by
using INREC and OUTREC statements. INREC and OUTREC statements support a
wide variety of reformatting tasks including:
v The use of fixed position/length fields or variable position/length fields. For
fixed fields, you specify the starting position and length of the field directly. For
variable fields, such as delimited fields, comma separated values (CSV), tab
separated values, blank separated values, keyword separated fields,
null-terminated strings (and many other types), you define rules that allow
DFSORT to extract the relevant data into fixed parsed fields, and then use the
parsed fields as you would use fixed fields.
v Insertion of blanks, zeros, strings, current date, future date, past date, current
time, sequence numbers, decimal constants, and the results of arithmetic
expressions before, between, and after the input fields in the reformatted
records.
v Sophisticated conversion capabilities, such as find and replace, hexadecimal
display, bit display, translation of EBCDIC letters from lowercase to uppercase or
uppercase to lowercase, translation of characters from EBCDIC to ASCII and
from ASCII to EBCDIC, translation of characters using the ALTSEQ translation
table, conversion of numeric values from one format to another, left-justify or
DFSORT Overview
v
v
left-squeeze (remove leading blanks or all blanks and shift left), and right-justify
or right-squeeze (remove trailing blanks or all blanks and shift right).
Sophisticated editing capabilities, such as control of the way numeric fields are
presented with respect to length, leading or suppressed zeros, thousands
separators, decimal points, leading and trailing positive and negative signs, and
so on.
Twenty-seven pre-defined editing masks are available for commonly used
numeric editing patterns, encompassing many of the numeric notations used
throughout the world. In addition, a virtually unlimited number of numeric
editing patterns are available via user-defined editing masks.
Transformation of SMF, TOD, and ETOD date and time values to more usable
forms.
Conversion of input date fields of one type (CH, ZD, PD, 2-digit year, 4-digit
year, Julian, Gregorian) to corresponding output date fields of another type or to
a corresponding day of the week.
Various types of arithmetic operations for input date fields.
v Selection of a character constant, hexadecimal constant, or input field from a

lookup table for output, based on a character, hexadecimal, or bit string as input
(that is, lookup and change).
You can create the reformatted INREC or OUTREC records in one of the following
ways:
v By building the entire record one item at a time.
v By only overlaying specific columns.
v By performing find and replace operations.
v By using sophisticated conditional logic or group operations to choose how
different records are reformatted.
You can sum numeric information from many records into one record with the
SUM statement. For example, if you want to know the total amount of a yearly
payroll, you can add the values for a field containing salaries from the records of
all your employees.
You can create one or more output data sets for a sort, copy, or merge application
from a single pass over one or more input data sets by using OUTFIL control
statements. You can use multiple OUTFIL statements, with each statement
specifying the OUTFIL processing to be performed for one or more output data
sets. OUTFIL processing begins after all other processing ends (that is, after
processing for exits, options, and other control statements). OUTFIL statements
support a wide variety of output data set tasks, including:
v Creation of multiple output data sets containing unedited or edited records from
a single pass over one or more input data sets.
v Creation of multiple output data sets containing different ranges or subsets of
records from a single pass over one or more input data sets. In addition, records
that are not selected for any subset can be saved in a separate output data set.
v Conversion of variable-length record data sets to fixed-length record data sets.
v Conversion of fixed-length record data sets to variable-length record data sets.
v A wide variety of parsing, editing and reformatting tasks including:
The use of fixed position/length fields or variable position/length fields. For
fixed fields, you specify the starting position and length of the field directly.
For variable fields, such as delimited fields, comma separated values (CSV),
tab separated values, blank separated values, keyword separated fields,
DFSORT Overview
Insertion of blanks, zeros, strings, current date, future date, past date, current
records.
Sophisticated conversion capabilities, such as find and replace, hexadecimal
display, bit display, translation of EBCDIC letters from lowercase to uppercase
or uppercase to lowercase, translation of characters from EBCDIC to ASCII or
from ASCII to EBCDIC, translation of characters using the ALTSEQ
translation table, conversion of numeric values from one format to another,
left-justify or left-squeeze (remove leading blanks or all blanks and shift left),
and right-justify or right-squeeze (remove trailing blanks or all blanks and
shift right).
Sophisticated editing capabilities, such as control of the way numeric fields
are presented with respect to length, leading or suppressed zeros, thousands
separators, decimal points, leading and trailing positive and negative signs,
and so on.
forms.
year, Julian, Gregorian) to corresponding output date fields of another type or
to a corresponding day of the week.
Selection of a character constant, hexadecimal constant, or input field from a
lookup table for output, based on a character, hexadecimal, or bit string as
input (that is, lookup and change).
v Creation of the reformatted records in one of the following ways:
By building the entire record one item at a time.
v
v
v
By only overlaying specific columns.

By performing find and replace operations.
By using sophisticated conditional logic or group operations to choose how
Highly detailed three-level (report, page, and section) reports containing a
variety of report elements you can specify (for example, current date, current
time, edited or converted page numbers, character strings, and blank lines) or
derive from the input records (for example, character fields; unedited, edited, or
converted numeric input fields; edited or converted record counts; and edited or
converted totals, maximums, minimums, and averages for numeric input fields).
Creation of multiple output records from each input record, with or without
intervening blank output records.
Repetition and sampling of data records.
Splitting of data records in rotation among a set of output data sets.
You can perform various "join" operations on two files by one or more keys. A
JOINKEYS application allows you to create joined records in a variety of ways
including inner join, full outer join, left outer join, right outer join and unpaired
DFSORT Overview
combinations. The two input files can be of different types (fixed, variable, VSAM,
and so on) and have keys in different locations. The records from the two input
files can be processed in a variety of ways before and after they are joined.
You can control DFSORT functions with other control statements by specifying
alternate collating sequences, invoking user exit routines, overriding installation
defaults, and so on.
You can direct DFSORT to pass control during run time to routines you design
and write yourself. For example, you can write user exit routines to summarize,
insert, delete, shorten, or otherwise alter records during processing. However, keep
in mind that the extensive editing capabilities provided by the INCLUDE, OMIT,
INREC, OUTREC, SUM, and OUTFIL statements can eliminate the need to write
user exit routines. You can write your own routines to correct I/O errors that
DFSORT does not handle, or to perform any necessary abnormal end-of-task
operation before DFSORT terminates.
You can write an EFS (Extended Function Support) program to intercept DFSORT
control statements and PARM options for modification prior to use by DFSORT or
to provide alternate sequence support for user-defined data.
You can define and use a symbol for any field, constant, or output column that is
recognized in a DFSORT control statement or ICETOOL operator. This makes it
easy to create and reuse collections of symbols (that is, mappings) representing
information associated with various record layouts. You can use system symbols
(for example, &JOBNAME.) in your symbol constants. You can use SET and PROC
symbols in your symbol constants. See Chapter 8, Using symbols for fields and
constants, on page 731.
DFSORT on the Web

For articles, online documents, news, tips, techniques, examples, and more, visit
the DFSORT home page at:
http://www.ibm.com/storage/dfsort
DFSORT FTP site

You can obtain DFSORT articles and examples via anonymous FTP to:
ftp.software.ibm.com/storage/dfsort/mvs/
Invoking DFSORT
You can invoke DFSORT processing in the following ways:
v With an EXEC job control statement in the input stream using the name of the
program (for example, PGM=ICEMAN or PGM=SORT) or the name of a
cataloged procedure (for example, SORTD). See Chapter 2, Invoking DFSORT
with Job Control Language, on page 27.
TSO users can allocate the needed ddnames (for example, SYSOUT, SORTIN,
SORTOUT and SYSIN), and invoke DFSORT using a calling method equivalent
to PGM=ICEMAN. For example:
call *(iceman)
Restriction: TSO users cannot invoke DFSORT using:

iceman
or any other alias for DFSORT (for example, SORT) in this form.
Invoking DFSORT
See Chapter 11, Examples of DFSORT job streams, on page 819 for examples of
invoking DFSORT from REXX and CLISTs.
v With a program written in basic assembler language using a system macro
instruction. See Chapter 6, Invoking DFSORT from a program, on page 541.
v With programs written in either COBOL or PL/I with a special facility of the
language. See the programmer's guide describing the compiler version available
at your location.
v With the ICETOOL utility. See Chapter 7, Using ICETOOL, on page 563.
In this document, the term directly invoked means that DFSORT is not initiated from
another program. The term program invoked means that DFSORT is initiated from
another program.
How DFSORT works

This section contains a list of the operating systems supported by DFSORT and an
explanation of how DFSORT uses control fields and collating sequences to sort,
merge, and copy the records of a data set.
The Blockset technique is DFSORT's most efficient technique for sorting, merging
and copying data sets. DFSORT uses the Blockset technique whenever possible to
take advantage of its highly optimized internal algorithms and efficient utilization
of IBM hardware. If Blockset cannot be used, DFSORT uses another of its
techniques Peerage/Vale or Conventional.
Operating systems
DFSORT runs under control of your z/OS operating system and must be initiated
according to the appropriate conventions.
Additionally, DFSORT runs under z/OS when it is running as a guest under
z/VM.
DFSORT is compatible with all of the IBM processors supported by z/OS. In
addition to any device supported by z/OS for program residence, DFSORT also
operates with any device QSAM or VSAM uses for input or output.
Control fields and collating sequences

You define control fields to identify the information you want DFSORT to sort or
merge. When thinking of the contents of your data sets, you probably think of
names, dates, account numbers, or similar pieces of useful information. For
example, when sorting your data sets, you might choose to arrange your records in
alphabetical order, by family name. By using the byte position and length (in bytes)
of the portion of each record containing a family name, you can define it as a
control field to manipulate with DFSORT.
DFSORT uses the control fields you define as keys in processing. A key is a
concept, such as family name, that you have in mind when you design a record
processing strategy for a particular application. A control field, on the other hand,
is a discrete portion of a record that contains the text or symbols corresponding to
that information in a form that can be used by DFSORT to identify and sort, or
merge the records. For all practical purposes, you can think of keys as equivalent
to the control fields DFSORT uses in processing.
How DFSORT Works

To arrange your records in a specific order, identify one or more control fields of
your records to use as keys. The sequence in which you list the control fields
becomes the order of priority DFSORT uses to arrange your records. The first
control field you specify is called the major control field. Subsequent control fields
are called minor control fields, as in first, second, third minor control fields, and so
on.
If two or more records have identical values for the first control field, they are
arranged according to the values in the second. Records with identical values for
the first and second are arranged according to the third, and so on, until a
difference is found or no more control fields are available.
Records with identical values for all the control fields specified retain their original
input order or are arranged randomly, depending upon which of the two options,
EQUALS or NOEQUALS, is in effect. You can direct DFSORT to retain the original
input order for records with identical values for all control fields by specifying
EQUALS.
Control fields may overlap, or be contained within other control fields (such as a
three-digit area code, within a 10-digit telephone number). They do not need to be
contiguous but must be located within the first 32752 bytes of the record (see
Figure 1).
Record
Control
field 3
Control
field 4
Control field 1
(major)
Control
field 2
Figure 1. Control Fields. Control fields may overlap, or be contained within other control fields.
DFSORT offers several standard collating sequences. You can choose to arrange your
records according to these standard collating sequences or according to a collating
sequence defined in the active locale. Conceptually, a collating sequence is a
specific arrangement of character priority used to determine which of two values
in the same control field of two different records should come first. DFSORT uses
EBCDIC, the standard IBM collating sequence, or the ASCII collating sequence
when sorting or merging records. If locale processing is in effect, DFSORT will use
the collating sequence defined in the active locale.
The collating sequence for character data and binary data is absolute; character and
binary fields are not interpreted as having signs. For packed decimal, zoned
decimal, fixed-point, normalized floating-point, and the signed numeric data
formats, collating is algebraic; each quantity is interpreted as having an algebraic
sign.
You can modify the standard EBCDIC sequence to collate differently if, for
example, you want to allow alphabetic collating of national characters. An alternate
collating sequence can be defined with the ALTSEQ installation option, or you can
define it yourself at run-time with the ALTSEQ program control statement. You can
also specify a modified collating sequence with an E61 user exit or with an EFS
program.
How DFSORT Works

You can specify the LOCALE installation or run-time option to use an active
locale's collating rules.
Cultural environment considerations

DFSORT's collating behavior can be modified according to your cultural
environment. Your cultural environment is defined to DFSORT using the X/Open**
locale model. A locale is a collection of data grouped into categories that describes
the information about your cultural environment.
The collate category of a locale is a collection of sequence declarations that defines
the relative order between collating elements (single character and multi-character
collating elements). The sequence declarations define the collating rules.
The cultural environment is established by selecting the active locale. The active
locale affects the behavior of locale-sensitive functions. In particular, the active
locale's collating rules affect DFSORT's SORT, MERGE, INCLUDE, and OMIT
processing as follows:
v Sort and Merge
DFSORT produces sorted or merged records for output according to the collating
rules defined in the active locale. This provides sorting and merging for singleor multi-byte character data based on defined collating rules that retain the
cultural and local characteristics of a language.
v Include and Omit
DFSORT includes or omits records for output according to the collating rules
defined in the active locale. This provides inclusion or omission for single- or
multi-byte character data based on defined collating rules that retain the cultural
and local characteristics of a language.
Note: Locale processing is not used for IFTRAIL TRLID or IFTHEN WHEN,
BEGIN or END constants or compare fields.
The DFSORT option LOCALE specifies whether locale processing is to be used
and, if so, designates the active locale. Only one locale can be active at a time for
any DFSORT application.
DFSORT processing
You must prepare job control language (JCL) statements and DFSORT program
control statements to invoke DFSORT processing. JCL statements (see Chapter 6,
Invoking DFSORT from a program, on page 541) are processed by your
operating system. They describe your data sets to the operating system and initiate
DFSORT processing. DFSORT program control statements (see Chapter 3, Using
DFSORT program control statements, on page 81) are processed by the DFSORT
program. They describe the functions you want to perform and invoke the
processing you request.
A sort application usually requires intermediate storage as working space during
the program run. This storage can be one of the following:
1. Hiperspace, using DFSORT's Hipersorting feature.
2. Work data setseither allocated dynamically by DFSORT's DYNALLOC facility
or specified by the user, using JCL DD statements. If specified by the user, the
intermediate storage devices and the amount of work space must be indicated.
Methods for determining the amount of work space to allocate are explained in
Appendix A, Using work space, on page 853.
DFSORT Processing
3. A combination of Hiperspace and work data sets.
Merge and copy applications do not require intermediate storage.
Figure 2 on page 9 illustrates the processing order for record handling, exits,
statements, and options. Use this diagram with the text following it to understand
the order DFSORT uses to run your job.
DFSORT Processing
Figure 2. Record Processing Order
As shown in Figure 2, DFSORT processing follows this order:
DFSORT Processing
1. DFSORT first checks whether you supplied a SORTIN data set for SORT and
COPY jobs or SORTINnn data sets for MERGE jobs. If so, DFSORT reads the
input records from them.
v If no SORTIN data set is present for a SORT or COPY job, you must use an
E15 user exit to insert all the records. (This is also true if you invoke
DFSORT from a program with the address of an E15 user exit in the
parameter list, because SORTIN will be ignored.) DFSORT can use a
COBOL E15 routine if you specified the E15 user exit in the MODS
statement.
v If no SORTINnn data sets are present for a MERGE job, you must use an
E32 user exit to insert all the records.
2. If input records for SORT or COPY jobs are read from a SORTIN data set,
DFSORT performs processing specified with the SKIPREC option. DFSORT
deletes records until the SKIPREC count is satisfied. Eliminating records
before a SORT or COPY gives better performance.
3. If the input records for a SORT or COPY job are read from a SORTIN data set,
DFSORT checks whether you specified an E15 user exit. If so, DFSORT
transfers control to the user exit routine. You can use a COBOL E15 routine if
the E15 user exit is specified in the MODS statement. The E15 routine can
insert, delete, or reformat records.
4. DFSORT performs processing specified on an INCLUDE or OMIT statement.
If you used an E15 user exit routine to reformat the records, the
INCLUDE/OMIT fields you specify must apply to the current format rather
than to the original format. If you use the INCLUDE or OMIT statements to
delete unnecessary records before SORT, MERGE, or COPY processing, your
jobs run more efficiently.
5. For SORT or COPY jobs, DFSORT performs processing specified with the
STOPAFT option. Record input stops after the maximum number of records
(n) you specify have been accepted. DFSORT accepts records for processing if
they are:
v Read from SORTIN or inserted by E15
v Not deleted by SKIPREC
v Not deleted by E15
v Not deleted by an INCLUDE or OMIT statement.
6. DFSORT performs processing specified in an INREC statement. Data records
are parsed, edited and reformatted according to the options specified. If you
reformatted the records before this step, the INREC fields you specify must
apply to the current format rather than to the original format.
7. DFSORT performs processing specified in the SORT, MERGE, or OPTION
COPY statement.
v For SORT, all input records are processed before any output record is
processed.
v For COPY or MERGE, an output record is processed after an input record is
processed.
v For SORT or MERGE, if a SUM statement is present, DFSORT processes it
during the SORT or MERGE processing. DFSORT summarizes the records
and deletes duplicates. If you reformatted the records before this step, the
SORT or MERGE and SUM fields you specify must apply to the current
format rather than to the original format.
8. DFSORT performs processing specified in an OUTREC statement. Data
records are parsed, edited and reformatted according to the options specified.
10
DFSORT Processing
If you reformatted the records before this step, the OUTREC fields you specify
must apply to the current format rather than to the original format.
9. If an E35 user exit is present, DFSORT transfers control to your user exit
routine after all statement processing (except OUTFIL) is completed. If you
reformatted the records, the E35 user exit receives the records in the current
format rather than in the original format. You can use a COBOL E35 routine if
you specify the E35 user exit in the MODS statement. You can use the E35 exit
routine to add, delete, or reformat records.
If SORTOUT and OUTFIL data sets are not present, the E35 exit must dispose
of all the records because DFSORT treats these records as deleted. (This is also
true if you do not specify OUTFIL data sets and DFSORT is invoked with the
address of an E35 user exit in the parameter list, because SORTOUT will be
ignored.)
10. DFSORT writes your records to the SORTOUT data set, if present.
11. DFSORT performs processing specified in one or more OUTFIL statements, if
present:
v DFSORT performs processing specified with the STARTREC, SAMPLE, and
ENDREC options. Record input for the OUTFIL data sets starts with the
record indicated by STARTREC, ends with the record indicated by
ENDREC, and is sampled according to the records indicated by SAMPLE.
v DFSORT performs processing specified with the INCLUDE, OMIT, or SAVE
option. Records are included or omitted from the OUTFIL data sets
according to the criteria specified.
v DFSORT performs processing specified with the ACCEPT option. Record
processing ends when the ACCEPT limit is reached.
v DFSORT performs processing specified with the PARSE, OUTREC (or
BUILD), OVERLAY, FINDREP, IFTHEN, FTOV, VTOF (or CONVERT),
VLFILL, VLTRIM, VLTRAIL and REPEAT options. Data records are parsed,
edited, reformatted, converted and repeated according to the options
specified.
v DFSORT performs processing specified with the LINES, HEADER1,
TRAILER1, HEADER2, TRAILER2, SECTIONS, NODETAIL, BLKCCH1,
BLKCCH2, BLKCCT1 and REMOVECC options. Data records are
reformatted and report records are generated for the OUTFIL data sets.
v DFSORT performs SPLIT, SPLITBY, or SPLIT1R processing. Records are
distributed among the OUTFIL data sets as evenly as possible.
v DFSORT writes your OUTFIL records to the appropriate OUTFIL data sets.
Input data setsSORTIN and SORTINnn

DFSORT processes two types of input data sets, referred to as the SORTIN data set
(or just SORTIN) and the SORTINnn data sets (or just SORTINnn).
The SORTIN DD statement specifies the input data set (or concatenated input data
sets) for a sort or copy application. If a SORTIN DD statement is present, it will be
used by default for a sort or copy application unless you invoke DFSORT from a
program with the address of an E15 user exit in the parameter list.
The SORTINnn DD statements (where nn can be 00 to 99) specify the data sets for
a merge application. If a SORTINnn DD statement is present, it will be used by
default for a merge application unless you invoke DFSORT from a program with
the address of an E32 user exit in the parameter list.
11
Input Data SetsSORTIN and SORTINnn

Data set considerations contains general information about input data sets. For
specific information about the SORTIN data set, see SORTIN DD statement on
page 67. For specific information about the SORTINnn data sets, see SORTINnn
DD statement on page 69.
Output data setsSORTOUT and OUTFIL

DFSORT processes two types of output data sets, referred to as the SORTOUT data
set (or just SORTOUT) and the OUTFIL data sets.
The SORTOUT DD statement specifies the single non-OUTFIL output data set for a
sort, copy, or merge application. OUTFIL processing does not apply to SORTOUT.
If a SORTOUT DD statement is present, it will be used by default for a sort, copy,
or merge application unless you invoke DFSORT from a program with the address
of an E35 user exit in the parameter list.
The FNAMES and FILES parameters of one or more OUTFIL statements specify
the ddnames of the OUTFIL data sets for a sort, copy, or merge application. The
parameters specified for each OUTFIL statement define the OUTFIL processing to
be performed for the OUTFIL data sets associated with that statement. Each
ddname specified must have a corresponding DD statement.
Although the ddname SORTOUT can actually be used for an OUTFIL data set, the
term SORTOUT will be used to denote the single non-OUTFIL output data set.
Data set considerations contains general information about output data sets. For
specific information about the SORTOUT data set, see SORTOUT and OUTFIL DD
statements on page 73. For specific information about the OUTFIL data sets, see
SORTOUT and OUTFIL DD statements on page 73 and OUTFIL control
statements on page 223.
Data set considerations

You must define any data sets you provide for DFSORT according to the
conventions your operating system requires. You can use the label checking
facilities of the operating system during DFSORT processing. See Application
Development Guide for details.
You must describe all data sets (except those allocated with the DYNALLOC
parameter) in DD statements. You must place the DD statements in the operating
system input stream with the job step that allocates data sets for DFSORT
processing.
Sorting or copying records

Input to a sort or copy application can be a blocked or unblocked QSAM or VSAM
data set containing fixed- or variable-length records. QSAM input data sets can be
concatenated even if they are on dissimilar devices. See SORTIN DD statement
on page 67 for the restrictions that apply.
Output from a sort or copy application can be blocked or unblocked QSAM or
VSAM data sets, regardless of whether the input is QSAM or VSAM. Unless
OUTFIL is used to convert variable input to fixed output, or fixed input to variable
output, an output data set must be the same type (fixed or variable) as the input
data set.
12
Data Set Considerations

Files in a z/OS file system are supported as input and output for sort and copy
applications.
Merging records
Input to a merge application can be up to 100 blocked or unblocked QSAM or
VSAM data sets containing fixed- or variable-length records. The input data sets
can be either QSAM or VSAM, but not both. The records in all input data sets
must already be sorted in the same order as that required for output.
Output from a merge application can be blocked or unblocked QSAM or VSAM
data sets, regardless of whether the input is QSAM or VSAM. Unless OUTFIL is
used to convert variable input to fixed output, or fixed output to variable output,
an output data set must be the same type (fixed or variable) as the input data set.
Files in a z/OS file system are supported as input and output for merge
applications.
Data set notes and limitations

There are some considerations and limitations that you need to be aware of. These
are described in the following sections.
For more information about specific DFSORT data sets, see Using DD statements
on page 61.
General considerations
Variable-length records are processed with a record descriptor word (RDW) in
positions 1-4, so the data starts in position 5. Fixed-length records are processed
without an RDW, so the data starts in position 1. Control statement positions
should be specified accordingly.
Your records can be EBCDIC, ASCII, Japanese, and data types you define yourself.
To process Japanese data types with DFSORT, you can use the IBM Double Byte
Character Set Ordering Support Program (DBCS Ordering), Licensed Program
5665-360, Release 2.0, or you can use locale processing with the appropriate locale.
Input and output data sets must be on devices that can be used with QSAM or
VSAM.
Standard system data management rules apply to all data set processing. In
particular:
v Be aware that when using fixed standard record format for input data sets, the
first short block is treated like an End of Volume. See z/OS DFSMS Using Data
Sets for more details.
v Be aware that, in some cases, if a DD statement specifies a data set for output
that is extended to a second or subsequent volume, and another DD statement
within the same step requests the same data set, only the records on the first
volume will be read, and incorrect output will result.
Specifically, when a new output data set is allocated with a unit count and
volume count greater than one, but specifies no volume serial numbers, one
volume is allocated. If a second or succeeding DD statement within the same
step requests the same data set, the same volume is allocated to it. If this job
step extends the output data set to more volumes, this new volume information
is not available to the second or succeeding DD statement.
13

Thus, you should not use different DDs for a data set to be used for output and
then input in the same step, unless the data set cannot be extended to a second
or subsequent volume, or is allocated with the guaranteed space attribute in the
storage class. See z/OS MVS JCL Reference, SA23-1385 and z/OS DFSMS Using
Data Sets, SC23-6855 for more details.
The maximum record length DFSORT can handle is subject to the following
limitations:
v Record length can never exceed the maximum record length you specify.
v Variable-length records are limited to 32756 bytes.
v VSAM variable-length records are limited to 32752 bytes.
v Fixed-length records are limited to 32760 bytes.
v Variable block-spanned records are limited to 32767 bytes.
v For a tape work data set sort, the maximum record length is limited to 32752
bytes with NOEQUALS in effect and to 32748 bytes with EQUALS in effect.
Note: If AQ format is specified, or CH format is specified and the CHALT option
is in effect, the maximum record length for variable-length records is 32767 bytes,
less the length of the control fields.
The number of records that can be sorted using a given amount of storage is
reduced by:
v Processing control fields of different formats
v Large numbers of control fields
v Large numbers of intermediate data sets.
Providing an Extended Function Support program with an EFS01 routine can limit
the record length that can be used when processing variable-length records.
The minimum block length for tape work data sets is 18 bytes; the minimum
record length is 14 bytes.
Padding and truncation

You can control the action that DFSORT takes when the SORTOUT LRECL is
smaller than the SORTIN/SORTINnn LRECL with the TRUNC option as described
in OPTION control statement on page 173.
DFSORT truncates fixed-length records on the right when the SORTOUT LRECL is
smaller than the SORTIN/SORTINnn LRECL provided that:
v The application is not a conventional merge or tape work data set sort.
v TRUNC=RC16 is not in effect.
You can control the action that DFSORT takes when a variable-length output
record is longer than the LRECL of the SORTOUT or OUTFIL data set to which it
is to be written by using the VLLONG or NOVLLONG option as described in
OPTION control statement on page 173.
You can control the action that DFSORT takes when the SORTOUT LRECL is larger
than the SORTIN/SORTINnn LRECL with the PAD option as described in
DFSORT pads fixed-length records with binary zeros on the right when the
SORTOUT LRECL is larger than the SORTIN LRECL provided that:
14

v The Blockset technique is selected.
v The application is a sort or copy.
v PAD=RC16 is not in effect.
DFSORT does not pad or truncate records returned from an E15 or E35 user exit
since it expects the exit to pad or truncate each record appropriately.
You can use INREC, OUTREC, and OUTFIL to pad, truncate, and reformat records.
See INREC control statement on page 125 and OUTREC control statement on
page 402 for details.
See Use ICEGENER instead of IEBGENER on page 813 for information about
padding and truncating with ICEGENER.
For more information about Blockset and other DFSORT techniques, see Specify
efficient sort/merge techniques on page 800.
QSAM considerations
v If you use DSN=NULLFILE on your DD statement for an input data set, a
system restriction prevents DFSORT from using the EXCP access method.
v Empty input data sets can be used.
v If any of the input data sets are on tape without standard labels, DCB
parameters must be specified on their DD statements.
v ISO/ANSI Version 1 tape files can only be used as inputnever as output.
v DFSORT sets appropriate BUFNO values for the input and output data sets;
specifying BUFNO in the DD statements for these data sets has no effect.
See SORTIN DD statement on page 67 for additional considerations.
VSAM considerations
v You can have DFSORT process VSAM records as fixed-length (F) or
variable-length (V). When you use VSAM input, DFSORT selects fixed-length
processing if you specify RECORD TYPE=F or variable-length processing if you
specify RECORD TYPE=V. If you do not specify RECORD TYPE=x, DFSORT
selects the record type to use according to the "rules" described in the discussion
of the TYPE operand in RECORD control statement on page 438. The record
type selected affects how the records are treated, and how control statement
positions should be specified, as follows:
Variable-length processing: An RRDS, KSDS, ESDS or VRRDS can always be
processed as variable-length. For VSAM input, DFSORT reads each record
and prepends a record descriptor word (RDW) to it. For VSAM output,
DFSORT removes the RDW before writing each record. Since DFSORT uses
an RDW in positions 1-4 to process variable-length records, the data starts in
position 5. Control statement positions should be specified accordingly.
Fixed-length processing: An RRDS can always be processed as fixed-length.
A KSDS, ESDS or VRRDS used for input should only be processed as
fixed-length if all of its records have a length equal to the maximum record
size defined for the cluster. Otherwise, input records which are shorter than
the maximum record size are padded with bytes that may or may not be
zeros (that is, "garbage" bytes). DFSORT does not use an RDW to process
fixed-length records, so the data starts in position 1. Control statement
positions should be specified accordingly.
v If a data set is password protected, passwords can be entered at the console or
(with some restrictions) through routines at user exits E18, E38, and E39.
15
v
v
v
v
v
v
Note: Passwords cannot be handled in this way for OUTFIL data sets.
If VSAMIO and RESET are in effect, a data set defined with REUSE can be used
for both input and output for a sort; that is, the data set can be sorted in-place.
A data set used for input or output must have been previously defined.
If VSAMEMT is in effect, an empty input data set is processed as having zero
records.
VSAM data sets must not be concatenated (system restriction).
VSAM and non-VSAM input data sets must not be specified together for a sort,
merge or copy application.
If output is a VSAM key-sequenced data set (KSDS), the key must be the first
control field (or the key fields must be in the same order as the first control
field). VSAM does not allow you to store records with duplicate primary keys.
Any VSAM exit function available for input data sets can be used except
EODAD. See the description of E18 use with VSAM in Chapter 5, Using your
own user exit routines, on page 487.
You must build the VSAM exit list with the VSAM EXLST macro instruction
giving the addresses of your routines that handle VSAM exit functions.
v
v When processing variable-length records with VSAM input and non-VSAM
output, the output LRECL must be at least 4 bytes greater than the maximum
record size defined for the cluster. Non-VSAM variable-length records have a
record descriptor word (RDW) field 4 bytes long at the beginning of each record,
but VSAM records do not. The record size defined for the VSAM cluster is
therefore 4 bytes less than the non-VSAM LRECL.
v An output data set defined without REUSE is processed as MOD.
v If RESET is in effect, an output data set defined with REUSE is processed as
NEW. If NORESET is in effect, an output data set defined with REUSE is
processed as MOD.
v DFSORT cannot access VSAM data sets in RLS mode, that is, RLS=CR and
RLS=NRI are not supported for VSAM input and output data sets.
XTIOT, uncaptured UCBs and DSAB above 16 megabytes

A program that invokes DFSORT, ICETOOL or ICEGENER can dynamically
allocate input, output and work data sets using the options for XTIOT, uncaptured
UCBs, and DSAB above 16 megabyte virtual. These data sets are supported to the
extent that z/OS supports them.
Note: A program that invokes DFSORT, ICETOOL or ICEGENER should not
allocate other data sets such as message, control, list, count, or symbol data sets
using the options for XTIOT, uncaptured UCBs, and DSAB above 16 megabyte
virtual. These data set options are not supported and the data set will not be
recognized.
z/OS file system considerations

Files in a z/OS file system can be used for input and output, but are only
supported by the Blockset technique. If Blockset is not selected for a DFSORT
application that uses z/OS UNIX files, DFSORT will issue an error message and
terminate.
You should be familiar with the information found in z/OS V2R2.0 UNIX System
Services User's Guide regarding z/OS UNIX files if you use them. DFSORT uses
16
z/OS File System Considerations

BSAM to access z/OS UNIX files and is thus subject to all of the capabilities and
restrictions that entails, as described in z/OS DFSMS Using Data Sets.
Installation defaults
When your system programmers installed DFSORT, they selected separate sets of
installation defaults for the following eight installation environments:
ICEAM1 (JCL)
is the batch direct invocation environment. This set of installation defaults
is used at run time when DFSORT is invoked directly (that is, not through
programs) by batch jobs, provided that an enabled time-of-day installation
environment (ICETDx) is not activated.
ICEAM2 (INV)
is the batch program invocation environment. This set of installation
defaults is used at run time when DFSORT is invoked through batch
programs, provided that an enabled time-of-day installation environment
(ICETDx) is not activated.
ICEAM3 (TSO)
is the TSO direct invocation environment. This set of installation defaults is
used at run time when DFSORT is invoked directly (that is, not through
programs) by foreground TSO users, provided that an enabled time-of-day
installation environment (ICETDx) is not activated.
ICEAM4 (TSOINV)
is the TSO program invocation environment. This set of installation
defaults is used at run time when DFSORT is invoked through programs
by foreground TSO users, provided that an enabled time-of-day installation
environment (ICETDx) is not activated.
ICETD1 (TD1)
is the first time-of-day installation environment. This set of installation
defaults is used at run time when it is activated for the time-of-day of the
run, provided it is enabled by the installation environment (ICEAMx) in
effect.
ICETD2 (TD2)
is the second time-of-day installation environment. This set of installation
effect.
ICETD3 (TD3)
is the third time-of-day installation environment. This set of installation
effect.
ICETD4 (TD4)
is the fourth time-of-day installation environment. This set of installation
effect.
The selected installation defaults can affect the way your applications run, and in
many cases can be overridden by specifying the appropriate run-time parameters
(see Appendix B, Specification/override of DFSORT options, on page 863 for full
17
Installation Defaults
override details). This document assumes that DFSORT is running with the
installation defaults it was delivered with (that is, with the IBM-supplied
installation defaults).
You can use an ICETOOL job similar to the following one to display a report of the
installation defaults to be used at run-time.
//DFRUN JOB A402,PROGRAMMER
//LISTDEF EXEC PGM=ICETOOL
//TOOLMSG DD SYSOUT=A
//DFSMSG DD SYSOUT=A
//SHOWDEF DD SYSOUT=A
//TOOLIN DD *
DEFAULTS LIST(SHOWDEF)
/*
Figure 3. Using ICETOOL to List Installation Defaults
See Chapter 7, Using ICETOOL, on page 563 and DEFAULTS operator on page
590 for more information on using ICETOOL and the DEFAULTS operator.
The functions of the available installation options are summarized later in this
section. z/OS DFSORT Installation and Customization contains complete descriptions
of the available installation options, as well as planning considerations and general
information about installing DFSORT. Step-by-step installation procedures are
listed in the z/OS Install/Migration page (http://www.ibm.com/systems/z/os/
zos/installation/).
Parameter
Function
INV|JCL|TSO|TSOINV|TD1|TD2|TD3|TD4
Specifies the invocation installation environment (ICEAMx) or time-of-day
installation environment (ICETDx) for which this set of installation defaults
is to be used.
ENABLE
Specifies whether ICETDx installation defaults are to be used if activated
for this ICEAMx installation environment.
day
Specifies the time ranges for each day of the week when this ICETDx
installation environment is to be activated.
ABCODE
Specifies the ABEND code used when DFSORT abends for a critical error.
ALTSEQ
Specifies changes to the ALTSEQ translation table.
ARESALL
Specifies the number of bytes reserved above 16MB virtual for system use.
ARESINV
Specifies the number of bytes reserved above 16MB virtual for the
invoking program when DFSORT is program invoked.
CFW
Specifies whether DFSORT can use cache fast write when processing work
data sets.
18
CHALT
Translates format CH as well as format AQ, or translates format AQ only.
CHECK
Specifies whether record count checking is suppressed for applications that
use an E35 user exit routine without an output data set.
CINV
Specifies whether DFSORT can use control interval access for VSAM data
sets.
COBEXIT
Specifies the library for COBOL E15 and E35 routines.
DIAGSIM
Specifies whether a SORTDIAG DD statement is to be simulated for
DFSORT applications.
DSA
Specifies the maximum amount of storage available to DFSORT for
dynamic storage adjustment of Blockset sort applications.
DSPSIZE
Specifies the maximum amount of data space to use for dataspace sorting.
DYNALOC
Specifies the default values for device name and number of work data sets
to be dynamically allocated. These default values are used in conjunction
with the DYNAUTO installation option and the DYNALLOC run-time
option.
DYNAPCT
Specifies additional work data sets to be dynamically allocated and only
used when needed.
DYNAUTO
Specifies whether work data sets are dynamically allocated automatically.
DYNSPC
Specifies the total default primary space allocation for all of the
dynamically allocated work data sets when the file size is unknown.
EFS
Specifies the name of a user-written Extended Function Support program
to be called by DFSORT.
EQUALS
Specifies whether the order of records that collate identically is preserved
from input to output.
ERET
Specifies the action taken if DFSORT encounters a critical error.
ESTAE
19
Specifies whether DFSORT deletes its ESTAE recovery routine early or uses
it for the entire run.
EXITCK
Specifies whether DFSORT terminates or continues when it receives certain
invalid return codes from E15 or E35 user exit routines.
EXPMAX
Specifies the maximum total amount of available storage to be used at any
one time by all Hipersorting, memory object sorting and dataspace sorting
applications.
EXPOLD
Specifies the maximum total amount of old storage to be used at any one
time by all Hipersorting, memory object sorting and dataspace sorting
applications.
EXPRES
Specifies the minimum amount of available storage to be reserved for use
by non-Hipersorting, non-memory object sorting and non-dataspace sorting
applications.
FSZEST
Specifies whether DFSORT treats run-time options FILSZ=n and SIZE=n as
exact or estimated file sizes.
GENER
Specifies the name that ICEGENER is to use to transfer control to the
IEBGENER system utility. (ICEGENER is DFSORT's facility for IEBGENER
jobs.)
GNPAD
Specifies the action to be taken by ICEGENER for LRECL padding.
GNTRUNC
Specifies the action to be taken by ICEGENER for LRECL truncation.
HIPRMAX
Specifies the maximum amount of Hiperspace to use for Hipersorting.
IDRCPCT
Specifies a percentage which represents the approximate amount of data
compaction achieved by using the Improved Data Recording Capability
feature of IBM tape devices that support compaction.
IEXIT
Specifies whether DFSORT passes control to your site's ICEIEXIT routine.
IGNCKPT
Specifies whether the checkpoint/restart facility is ignored if it is requested
at run-time and the Blockset technique (which does not support the
checkpoint/restart facility) can be used.
IOMAXBF
Specifies an upper limit to the amount of buffer space to be used for
SORTIN, SORTINnn and SORTOUT data sets.
20
LIST
Specifies whether DFSORT prints control statements.
LISTX
Specifies whether DFSORT prints control statements returned by an
Extended Function Support program.
LOCALE
Specifies whether locale processing is to be used and, if so, designates the
active locale.
MAXLIM
Specifies an upper limit to the amount of main storage available to
DFSORT below 16MB virtual.
MINLIM
Specifies a lower limit to the amount of main storage available to DFSORT.
MOSIZE
Specifies the maximum amount of memory object storage to use for
memory object sorting.
MOWRK
Specifies whether DFSORT can use memory object storage as intermediate
work space.
MSGCON
Specifies the class of program messages DFSORT writes to the master
console.
MSGDDN
Specifies an alternate name for the message data set.
MSGPRT
Specifies the class of program messages DFSORT writes to the message
data set.
NOMSGDD
Specifies whether DFSORT terminates or continues when the message data
set is required but is not available.
NULLOFL
Specifies the action to be taken by DFSORT when there are no data records
for an OUTFIL data set.
NULLOUT
Specifies the action to be taken by DFSORT when there are no records for
the SORTOUT data set.
ODMAXBF
Specifies an upper limit to the amount of buffer space to be used for each
OUTFIL data set.
OUTREL
Specifies whether unused temporary output data set space is released.
21
OUTSEC
Specifies whether DFSORT uses automatic secondary allocation for output
data sets that are temporary or new.
OVERRGN
Specifies the amount of main storage above the REGION value available to
Blockset.
OVFLO
Specifies the action to be taken by DFSORT when BI, FI, PD or ZD
summary fields overflow.
PAD
Specifies the action to be taken by DFSORT for LRECL padding.
PARMDDN
Specifies an alternate ddname for the DFSORT DFSPARM data set.
RESALL
Reserves storage for system and application use when
SIZE/MAINSIZE=MAX is in effect.
RESET
Specifies whether DFSORT processes a VSAM output data set defined with
REUSE as a NEW or MOD data set.
RESINV
Reserves storage for programs invoking DFSORT when
SIZE/MAINSIZE=MAX is in effect.
SDB
Specifies whether DFSORT should use the system-determined optimum
block size for output data sets when the block size is zero.
SDBMSG
Specifies whether DFSORT and ICETOOL should use the
system-determined optimum block size for message and list data sets
when the block size is zero.
SIZE
Specifies the maximum amount of main storage available to DFSORT.
SMF
Specifies whether DFSORT produces SMF type-16 records.
SOLRF
Specifies whether DFSORT uses the reformatted record length for the
SORTOUT LRECL.
SORTLIB
Specifies whether DFSORT searches a system or private library for the
modules used with a tape work data set sort or Conventional merge.
SPANINC
22
Specifies the action to be taken by DFSORT when incomplete spanned
records are detected.
SVC
Specifies a user SVC number for DFSORT.
SZERO
Specifies whether DFSORT treats numeric -0 and +0 values as signed (that
is, different) or unsigned (that is, the same).
TEXIT
Specifies whether DFSORT passes control to your site's ICETEXIT routine.
TMAXLIM
Specifies an upper limit to the total amount of main storage above and
below 16MB virtual available to DFSORT when SIZE/MAINSIZE=MAX is
in effect.
TUNE
Specifies whether DFSORT should favor optimization of central storage or
disk work space for sort applications.
TRUNC
Specifies the action to be taken by DFSORT for LRECL truncation.
VERIFY
Specifies whether the sequence of output records is verified.
VIO
Specifies whether virtual allocation of work data sets is accepted.
VLLONG
Specifies whether DFSORT truncates long variable-length output records.
VLSCMP
Specifies whether DFSORT pads short variable-length compare fields.
VLSHRT
Specifies whether DFSORT continues processing if a short variable-length
control field, compare field or summary field is found.
VSAMBSP
Specifies the number of VSAM buffers DFSORT can use.
VSAMEMT
Specifies whether DFSORT accepts an empty VSAM input data set.
VSAMIO
Specifies whether DFSORT allows a VSAM data set defined with REUSE to
be sorted in-place.
WRKREL
Specifies whether unused temporary work data set space is released.
WRKSEC
23
Specifies whether DFSORT uses automatic secondary allocation for
temporary work data sets.
Y2PAST
Specifies the sliding or fixed century window.
ZDPRINT
Specifies whether DFSORT produces printable numbers from positive ZD
fields that result from summarization.
Tables showing all the possible sources of specification and order of override for
each option are shown in Appendix B, Specification/override of DFSORT
options, on page 863.
Migrating to DFSORT from other sort products

If you are migrating to DFSORT, you should run an ICETOOL DEFAULTS report
to review the installation defaults set at your site. In particular, the options shown
in the table that follows can make DFSORT operate more like other sort products,
thus making migration easier. The installation options, described in z/OS DFSORT
Installation and Customization, change the way DFSORT works globally by default.
The run-time options, described in Chapter 2, Invoking DFSORT with Job Control
Language, on page 27 and Chapter 3, Using DFSORT program control
statements, on page 81, can be used to override the installation defaults for
specific jobs.
Table 3. Options That Can Ease Migration
Run-Time
ABCODE=MSG/n
DYNALOC=(d,n)
DYNALLOC=(d,n)
DYNAPCT=x/OLD
DYNAPCT=x/OLD
DYNAUTO=YES/IGNWKDD/NO
DYNALLOC=(d,n)
DYNSPC=n
DYNSPC=n
EQUALS=YES/NO/VBLKSET
EQUALS/NOEQUALS
EXITCK=STRONG/WEAK
EXITCK=STRONG/WEAK
FSZEST=YES/NO
FILSZ=n/En/Un
NOMSGDD=QUIT/ALL/CRITICAL/NONE
PARMDDN=ddname
RESET=YES/NO
RESET/NORESET
SORTLIB=SYSTEM/PRIVATE
24
SZERO=YES/NO
SZERO/NOSZERO
VLLONG=YES/NO
VLLONG/NOVLLONG
VLSCMP=YES/NO
VLSCMP/NOVLSCMP
VSAMEMT=YES/NO
VSAMEMT/NVSAMEMT
VSAMIO=YES/NO
VSAMIO/NOVSAMIO
ZDPRINT=YES/NO
ZDPRINT/NZDPRINT
DFSORT Messages and Return Codes
DFSORT messages and return codes

You can determine, during installation or run-time, whether DFSORT writes
messages to the message data set, to the master console, or to both. You can also
direct an Extended Function Support program to write messages to the message
data set.
Messages written to the message data set can be either critical error messages,
informational error messages, or diagnostic messages, as determined during
installation or run-time.
Messages written to the master console can be either critical error messages or
informational error messages, as determined during installation.
See z/OS DFSORT Messages, Codes and Diagnosis Guide for complete information
about DFSORT messages.
For successful completion, DFSORT passes back a return code of 0 or 4 to the
operating system or the invoking program.
For unsuccessful completion due to an unsupported operating system, DFSORT
passes back a return code of 24 to the operating system or the invoking program.
For unsuccessful completion with NOABEND in effect, DFSORT passes back a
return code of 16, 20 or 28 to the operating system or the invoking program.
For unsuccessful completion with ABEND in effect, DFSORT issues a user abend
with the appropriate code as specified by the ABCODE installation option (either
the error message number or a number between 1 and 99).
The meanings of the return codes that DFSORT passes back (in register 15) are:
0
Successful completion. DFSORT completed successfully.
Successful completion. DFSORT completed successfully, and:

v OVFLO=RC4 was in effect and summary fields overflowed, or
v PAD=RC4 was in effect and the SORTOUT LRECL was larger than the
SORTIN/SORTINnn LRECL (LRECL padding), or
v TRUNC=RC4 was in effect and the SORTOUT LRECL was smaller than
the SORTIN/SORTINnn LRECL (LRECL truncation), or
v SPANINC=RC4 was in effect and one or more incomplete spanned
records was detected, or
v NULLOUT=RC4 was in effect and there were no records for the
SORTOUT data set, or
v NULLOFL=RC4 was in effect and there were no data records for an
OUTFIL data set.
16
Unsuccessful completion. DFSORT detected an error that prevented it

from completing successfully.
20
Message data set missing. Installation option NOMSGDD=QUIT was in

effect and neither a message data set DD statement nor a SYSOUT DD
statement was provided.
24
Unsupported operating system. This release of DFSORT does not support

this operating system.
25
DFSORT Messages and Return Codes

28
Wrong entry name. DFSORT detected one of the following errors related to
the program entry name that prevented it from completing successfully:
v DFSORT was invoked directly with PGM=ICEMAN64 or PGM=SORT64
(instead of with PGM=ICEMAN or PGM=SORT).
v DFSORT was invoked from a program using entry name ICEMAN64 or
SORT64, and an extended invocation parameter list, or a 24-bit
invocation parameter list (instead of a 64-bit invocation parameter list).
v DFSORT was invoked from a program using an entry name other than
ICEMAN64 or SORT64 (for example, ICEMAN or SORT) and a 64-bit
invocation parameter list.
Use Blockset whenever possible

Blockset is DFSORT's most efficient technique. It supports many features not
supported by DFSORT's less efficient Peerage/Vale and Conventional techniques
(see ICE189A in z/OS DFSORT Messages, Codes and Diagnosis Guide for a list of
these features). DFSORT always selects Blockset for a copy application. DFSORT
selects Blockset for a sort or merge application unless something prevents it from
doing so (see ICE800I in z/OS DFSORT Messages, Codes and Diagnosis Guide for a
list of reasons Blockset cannot be used).
Note: Blockset cannot be used to process BDAM data sets.
Message ICE143I indicates whether Blockset or a less efficient technique was
selected for a particular run. If Blockset was not selected for a sort or merge
application, check the reason code in message ICE800I, which indicates the reason
Blockset could not be used. If you did not get message ICE800I, add the following
DD statements to your application and rerun it:
//SORTDIAG DD DUMMY
//SYSOUT DD SYSOUT=*
If possible and appropriate, remove the obstacle that is causing Blockset not to be
selected.
26
Chapter 2. Invoking DFSORT with Job Control Language

Using the JCL
Your operating system uses the job control language (JCL) you supply with your
DFSORT program control statements to:
v Identify you as an authorized user
v Allocate the necessary resources to run your job
v Run your job
v Return information to you about the results
v Terminate your job.
You must supply JCL statements with every DFSORT job you submit.
Required JCL includes a JOB statement, an EXEC statement, and several DD
statements. The statements you need and their exact form depend upon whether
you:
v Use an EXEC statement in the input job stream or a system macro instruction
within another program to invoke DFSORT
v Use EXEC statement cataloged procedures to invoke DFSORT
v Specify various DFSORT control statements or PARM options
v Want to use program exits to activate routines of your own
v Use dynamic binding or link-editing
v Want to see diagnostic messages.
The JCL statements and their functions are listed in the following. Details on
coding the individual statements are presented in subsequent sections.
JCL Statement
Description
//JOBLIB DD
Defines your program link library if it is not already known to the system
//STEPLIB DD
Same as //JOBLIB DD
//SORTLIB DD
Defines the data set that contains special load modules if it is not already
known to the system
//SYSOUT DD1
Defines the message data set
//SYMNAMES DD
Defines the SYMNAMES data set containing statements to be used for
symbol processing
//SYMNOUT DD
27
Using the JCL

Defines the data set in which SYMNAMES statements and the symbol
table are to be listed
//SORTIN DD1
Defines the input data set for a sort or copy
//SORTINnn DD1
Defines the input data sets for a merge
//SORTOUT DD1
Defines the SORTOUT output data set for a sort, merge, or copy
//outfil DD
Defines an OUTFIL output data set for a sort, merge, or copy
//SORTWKdd DD1
Defines intermediate storage data sets for a sort
//DFSPARM DD1
Contains DFSORT PARM options and program control statements
//SYSIN DD
Contains DFSORT program control statements
//SORTCNTL DD1
Same as //SYSIN DD
//SORTDIAG DD
Specifies that all messages and program control statements be printed
//SORTCKPT DD
Defines the data set for checkpoint records
//SYSUDUMP DD
Defines the data set for output from a system ABEND dump routine
//SYSMDUMP DD
Same as //SYSUDUMP DD
//SYSABEND DD
Same as //SYSUDUMP DD
//SORTSNAP DD
Defines the snap dump data set dynamically allocated by DFSORT
//ddname
Defines the data set containing exit routines (as specified in the MODS
program control statement).
The following DD statements are necessary only for dynamic binding or
link-editing of exit routines
//SYSPRINT DD
Defines the message data set for the linkage editor
//SYSUT1 DD
Defines the intermediate storage data set for the linkage editor
28
Using the JCL

//SYSLIN DD
Defines the data set for control information for the linkage editor
//SYSLMOD DD
Defines the data set for output from the linkage editor
//SORTMODS DD
Defines the temporary partitioned data set for user exit routines from
SYSIN.
1
These are the default ddnames with which DFSORT was delivered.
SYSOUT and DFSPARM may have been changed during DFSORT
installation. You can change all of the indicated ddnames at run time. For
override information, see Appendix B, Specification/override of DFSORT
Using the JOB statement

The JOB statement is the first JCL statement of your job. It must contain a valid job
name in the name field and the word JOB in the operation field. All parameters in
the operand field are optional, although your site may have made information
such as account number and the name of the programmer mandatory:
//jobname JOB accounting information, programmers name, etc.
Using SET and PROC symbols in DFSORT control statements

For jobs that directly invoke DFSORT (PGM=SORT or PGM=ICEMAN), you can
construct DFSORT symbols that incorporate JCL PROC or SET symbols as well as
text and system symbols. You can then use those constructed DFSORT symbols in
DFSORT control statements in the same way you use regular DFSORT symbols.
For complete information, see Chapter 8, Using symbols for fields and constants,
on page 731.
Using the EXEC statement

The EXEC statement is the first JCL statement of each job step or of each procedure
step in a cataloged procedure. It identifies DFSORT to the operating system. You
can also specify DFSORT options on the EXEC statement.
The format of the EXEC statement is:
//stepname EXEC
PGM= SORT
ICEMAN
PROC= SORT
SORTD
SORT
SORTD
,
, PARM =
options
, other parameters
29
Using The EXEC Statement

If you use a cataloged procedure (discussed in detail later in this section), specify
PROC=SORT or PROC=SORTD. You can omit PROC= and simply specify SORT or
SORTD. However, PROC= can remind you that a cataloged procedure is being
used.
If you do not use a cataloged procedure, use PGM= either with the actual name of
the sort module (ICEMAN) or with one of its aliases: SORT, IERRCO00, or
IGHRCO00. Be sure that the alias has not been changed at your site.
Specifying EXEC statement cataloged procedures

A cataloged procedure is a set of JCL statements, including DD statements, that has
been assigned a name and placed in a partitioned data set called the procedure
library. Two cataloged procedures are supplied with the program: SORT and
SORTD. Specify them in the first parameter of the EXEC statement by
PROC=SORT, PROC=SORTD, or simply SORT or SORTD.
SORT cataloged procedure

You can use the supplied SORT cataloged procedure when you include user
routines that require binding or link-editing. Using this procedure without using
bound or link-edited user routines is inefficient because the SORT cataloged
procedure allocates linkage editor data sets whether or not you include user
routines.
When you specify EXEC PROC=SORT or EXEC SORT, the following JCL
statements are generated:
//SORT
//STEPLIB
//SORTLIB
//SYSOUT
//SYSPRINT
//SYSLMOD
//SYSLIN
//SYSUT1
//
30
EXEC
DD
DD
DD
DD
DD
DD
DD
PGM=ICEMAN
DSNAME=yyy,DISP=SHR
DSNAME=xxx,DISP=SHR
SYSOUT=A
DUMMY
DSNAME=&GOSET,UNIT=SYSDA,SPACE=(3600,(20,20,1))
DSNAME=&LOADSET,UNIT=SYSDA,SPACE=(80,(10,10))
DSNAME=&SYSUT1,SPACE=(1024,(60,20)),
UNIT=(SYSDA,SEP=(SORTLIB,SYSLMOD,SYSLIN))
00
10
20
30
40
50
60
70
80
Line
Explanation
00
The stepname of the procedure is SORT. This EXEC statement initiates the
program, which is named ICEMAN.
10
The STEPLIB DD statement defines the data set containing the DFSORT
program modules. If DFSORT was installed as part of the normal system
link libraries, the STEPLIB DD statement is unnecessary. It is needed only
if DFSORT resides in a separate link library which is not part of the link
list. (Your installation's system programmers can give you this
information.) The STEPLIB DD statement shown assumes that the data set
name represented by yyy is cataloged.
20
The SORTLIB DD statement defines a private data set containing the

modules needed for a sort using tape work files or a merge using the
Conventional technique. The data set is cataloged, and the data set name
represented by xxx was specified at installation time; it can be
SYS1.SORTLIB.

If the modules were installed in a system library and installation option
SORTLIB=SYSTEM is used, the SORTLIB DD statement is unnecessary and
is ignored unless dynamic link of user exits is used.
30
Defines an output data set for system use (messages). It is directed to

system output class A.
40
Defines SYSPRINT as a dummy data set because linkage editor diagnostic

output is not required.
50
Defines a data set for linkage editor output. Any system disk device is
acceptable for the output. Space for 20 records with an average length of
3600 bytes is requested; this is the primary allocation. Space for 20 more
records is requested if the primary space allocation is not sufficient; this is
the secondary allocation, which is requested each time primary space is
exhausted. The last value is space for a directory, which is required
because SYSLMOD is a new partitioned data set.
60
The SYSLIN data set is used by the program for linkage editor control
statements. It is created on any system disk device, and it has space for 10
records with an average length of 80 bytes. If the primary space allocation
is exhausted, additional space is requested in blocks large enough to
contain 10 records. No directory space is necessary.
70/80
The SYSUT1 DD statement defines a work data set for the linkage editor.
SORTD cataloged procedure

You can use the supplied SORTD cataloged procedure when you do not include
user routines or when you include user routines that do not require binding or
link-editing.
When you specify EXEC PROC=SORTD or EXEC SORTD, the following JCL
statements are generated:
//SORT EXEC PGM=ICEMAN
//STEPLIB DD DSNAME=yyy,DISP=SHR
//SORTLIB DD DSNAME=xxx,DISP=SHR
//SYSOUT DD SYSOUT=A
00
10
20
30
Line
Explanation
00
The stepname of the SORTD procedure is SORT
10
The STEPLIB DD statement defines the data set containing the DFSORT
program modules. If DFSORT was installed as part of the normal system
link libraries, the STEPLIB DD statement is unnecessary. It is needed only
if DFSORT resides in a separate link library which is not part of the link
list. (Your installation's system programmers can give you this
information.) The STEPLIB DD statement shown assumes that the data set
name represented by yyy is cataloged.
20
The SORTLIB DD statement defines a private data set that contains the
modules needed for a sort using tape work files or a merge that uses the
Conventional technique. The data set name of the program subroutine
library, represented by xxx, is specified at installation time; it can be
SYS1.SORTLIB.
31

If the modules were installed in a system library and installation option
SORTLIB=SYSTEM is used, then the SORTLIB DD statement is
unnecessary and is ignored unless dynamic link edit of user exits is used.
30
Directs messages to system output class A
Specifying EXEC/DFSPARM PARM options

When you invoke DFSORT with JCL, you can specify some DFSORT options on
the PARM parameter of the EXEC statement as illustrated on the following page.
These options include EFS, LIST, NOLIST, LISTX, NOLISTX, MSGPRT, and
MSGDDN, which are ignored if specified in an OPTION statement in SYSIN. Full
override and applicability details are listed in Appendix B, Specification/override
of DFSORT options, on page 863.
If you use the DFSPARM DD statement instead, you can specify both EXEC PARM
options and DFSORT control statements in a single source data set that overrides
all other sources. See DFSPARM DD statement on page 76.
Details of aliases for PARM options are given under the description of individual
options. Aliases for PARM options on page 60 summarizes the available aliases.
DFSORT accepts but does not process the following EXEC/DFSPARM PARM
options: BALANCE, BALN, BIAS=value, BMSG, CASCADE, CMP=value, CPU,
CRCX, DEBUG, DIAG, ELAP, EXCPVR=value, IO, INCOR=value, INCORE=value,
LRGSORT, L6=value, L7=value, NOCOMMAREA, NOINC, NOIOERR,
NOSTIMER, OPT=value, OSCL, PEER, POLY, PRINT121 and STIMER.
Note: If DEBUG is specified as the first value in a DFSPARM statement, it will be
interpreted as a DEBUG control statement rather than as a DFSPARM PARM
option.
An option shown in the form parameter=value can also be specified in the
equivalent form parameter(value) or parameter=(value). For example, the following
are equivalent:
v DYNALLOC=SYSDA
v DYNALLOC(SYSDA)
v DYNALLOC=(SYSDA)
An option shown in the form parameter=(list) can also be specified in the
equivalent form parameter(list). For example, the following are equivalent:
v DYNALLOC=(SYSDA,5)
v DYNALLOC(SYSDA,5)
Syntax Diagram for EXEC PARM
32

,
,PARM='
ABEND
NOABEND
ARESALL=
n
nK
nM
AVGRLEN=n
BSAM
CINV
NOCINV
COBEXIT=
COB1
COB2
DSA=n
DSPSIZE=
MAX
n
DYNALLOC
=
d
(,n)
(d,n)
OFF
DYNAPCT=
x
OLD
DYNSPC=n
EFS=
name
NONE
EQUALS
NOEQUALS
E15=COB
E35=COB
FILSZ=
x
Ex
Ux
HIPRMAX=
OPTIMAL
n
p%
LIST
NOLIST
LISTX
NOLISTX
LOCALE=
name
CURRENT
NONE
MSGDDN=ddname
MOSIZE=
MAX
n
%p
MOWRK
NOMOWRK
MSGPRT=
ALL
CRITICAL
NONE
NULLOUT=
RC0
RC4
RC16
ODMAXBF=
n
nK
nM
Additional
33

OUTREL
NOOUTREL
OVFLO=
RC0
RC4
RC16
PAD=
RC0
RC4
RC16
RESALL=
n
nK
nM
RESET
NORESET
SDB=
LARGE
YES
INPUT
NO
SIZE=
n
nK
nM
MAX
MAX-m
MAX-mK
MAX-mM
SKIPREC=z
SOLRF
NOSOLRF
SPANINC=
RC0
RC4
RC16
STOPAFT=n
SZERO
NOSZERO
TRUNC=
RC0
RC4
RC16
VERIFY
NOVERIFY
VLLONG
NOVLLONG
VLSCMP
NOVLSCMP
VLSHRT
NOVLSHRT
VSAMEMT
NOVSAMENT
VSAMIO
NOVSAMIO
WRKREL
NOWRKREL
WRKSEC
NOWRKSEC
Y2PAST=
s
f
ZDPRINT
NZDPRINT
ABEND or NOABEND
34
ABEND
NOABEND

Temporarily overrides the ERET installation option, which specifies whether
DFSORT abends or terminates with a return code of 16 if your sort, copy, or
merge is unsuccessful.
ABEND
specifies that if your sort, copy, or merge is unsuccessful, DFSORT abends
with a user completion code equal to the appropriate message number or
with a user-defined number between 1 and 99, as set during installation
with installation option ABCODE=n.
When DEBUG ABEND is in effect, a user abend code of zero may be
issued when a tape work data set sort or Conventional merge is
unsuccessful.
NOABEND
specifies that an unsuccessful sort, copy, or merge terminates with a return
code of 16.
Note: RC16=ABE and NORC16 can be used instead of ABEND and
NOABEND, respectively.
Default: Usually the installation default. See Appendix B, Specification/
override of DFSORT options, on page 863 for full override details.
Applicable Functions: See Appendix B, Specification/override of DFSORT
options, on page 863
ARESALL
ARESALL=
n
nK
nM
Temporarily overrides the ARESALL installation option, which specifies the

number of bytes to be reserved above 16MB virtual for system use. For more
information, see the discussion of the ARESALL option in OPTION control
statement on page 173.
n
specifies that n bytes of storage are to be reserved.

Limit: 8 digits.
nK specifies that n times 1024 bytes of storage are to be reserved.

Limit: 5 digits.
nM specifies that n times 1048576 bytes of storage are to be reserved.
Limit: 2 digits.
Note: RESERVEX=value can be used instead of ARESALL=value.
AVGRLEN
AVGRLEN=n
Specifies the average input record length in bytes for variable-length record
sort applications. For more information, see the discussion of the AVGRLEN
option in OPTION control statement on page 173.
35

n
specifies the average input record length. The value for n must be between
4 and 32767 and must include the 4 byte record descriptor word (RDW).
Note: L5=n can be used instead of AVGRLEN=n.

Default: If AVGRLEN=n is not specified, DFSORT will use one-half of the
maximum record length as the average record length. See Appendix B,
Specification/override of DFSORT options, on page 863 for full override
details.
BSAM
BSAM
Temporarily bypasses the EXCP access method normally used for input and
output data sets. BSAM is ignored for VSAM input and output data sets. Note
that if Blockset is not selected and BSAM processing is used with concatenated
SORTIN input and both null and non-null data sets are specified, all null data
sets must precede all non-null data sets; otherwise, the results are
unpredictable.
Attention: This option can degrade performance.
Default: None; optional. See Appendix B, Specification/override of DFSORT
options, on page 863 for full override details.
CINV or NOCINV
CINV
NOCINV
Temporarily overrides the CINV installation option, which specifies whether

DFSORT can use control interval access for VSAM data sets. For more
information, see the explanation of the CINV option in OPTION control
CINV
directs DFSORT to use control interval access when possible for VSAM
data sets.
NOCINV
directs DFSORT not to use control interval access.
COBEXIT
COBEXIT=
COB1
COB2
Temporarily overrides the COBEXIT installation option, which specifies the

library for COBOL E15 and E35 routines.
36

COB1
specifies that COBOL E15 and E35 routines are run with the OS/VS
COBOL run-time library or, in some cases, with no COBOL run-time
library.
COBEXIT=COB1 is obsolete, but is still available for compatibility reasons.
Note that Language Environment is the only run-time library for COBOL
supported by IBM service.
COB2
specifies that COBOL E15 and E35 routines are run with either the VS
COBOL II run-time library or the Language Environment run-time library.
Note: The DFSORT documents only discuss the Language Environment
run-time library, and assume that COBEXIT=COB2 is in effect. Although it is
possible to run with older run-time libraries, and with COBEXIT=COB1, these
are not recommended or discussed, and are not supported by IBM service.
DSA
DSA=n
Temporarily overrides the DSA installation option, which specifies the

maximum amount of storage available to DFSORT for dynamic storage
adjustment of a Blockset sort application when SIZE/MAINSIZE=MAX is in
effect. For more information, see the discussion of the DSA option in OPTION
control statement on page 173".
n
specifies that DFSORT can dynamically adjust storage to improve

performance, subject to a limit of n MB. n must be n value between 0 and
2000. If n is less than or equal to the TMAXLIM value in effect, n is set to 0
to indicate that storage will not be dynamically adjusted.

DSPSIZE
DSPSIZE=
MAX
n
Temporarily overrides the DSPSIZE installation option which specifies the

maximum amount of data space to be used for dataspace sorting. For more
information, see the discussion of the DSPSIZE option in OPTION control
MAX
specifies that DFSORT dynamically determines the maximum amount of
data space to be used for dataspace sorting. In this case, DFSORT bases its
data space usage on the size of the file being sorted and the paging activity
of the system.
37

n
specifies the maximum amount, in megabytes, of data space to be used for

dataspace sorting. n must be a value between 0 and 9999. The actual
amount of data space used does not exceed n, but may be less depending
on the size of the file being sorted and the paging activity of the system.
If n is zero, dataspace sorting is not used.

DYNALLOC
DYNALLOC

=
d
(,n)
(d,n)
Specifies that DFSORT dynamically allocates needed work space. You do not
need to calculate and use JCL to specify the amount of work space needed by
the program.
For more information, see the discussion of the DYNALLOC option in
OPTION control statement on page 173 and Appendix A, Using work
space, on page 853
d
specifies the device name. You can specify any IBM disk or tape device
supported by your operating system in the same way you would specify it
in the JCL UNIT parameter. You can also specify a group name, such as
DISK or SYSDA.
specifies the maximum number of requested work data sets. If you specify
more than 255, a maximum of 255 data sets is used. If you specify 1 and
the Blockset technique is selected, a maximum of 2 data sets is used. If you
specify more than 32 and the Blockset technique is not selected, a
maximum of 32 data sets is used.
Note: For optimum allocation of resources such as virtual storage, avoid
specifying a large number of work data sets unnecessarily.

DYNALLOC=OFF
DYNALLOC=OFF
Directs DFSORT not to allocate intermediate workspace dynamically. It

overrides installation option DYNAUTO=YES or the DYNALLOC parameter
(without OFF) specified at run-time. For more information, see the discussion
of the DYNALLOC option in OPTION control statement on page 173.
OFF
directs DFSORT not to allocate intermediate workspace dynamically.
38

DYNAPCT
DYNAPCT=
x
OLD
specifies additional work data sets to be dynamically allocated with zero space.
DFSORT only extends these data sets when necessary to complete a sort
application. The availability of additional work data sets can help avoid out of
space ABENDs.
x
specifies the number of additional work data sets (y) as a percentage of the
maximum number of dynamically allocated work data sets
(DYNALLOC/DYNALOC n value) in effect. y will be set to n * x%. The
total number of dynamically allocated work data sets will be n + y. For
example, if DYNALLOC=(SYSDA,20) and DYNAPCT=20 are in effect, 4
additional work data sets will be allocated for a total of 24.
The value x must be between 0 and 254. The minimum value for y is 1 and
the maximum value for y is 254. The maximum value for n + y is 255; if x
results in a value for n + y greater than 255, y will be set to 255-n.
OLD
specifies additional work data sets should only be allocated when DFSORT
cannot determine the file size. When DFSORT is able to determine the file
size, additional work data sets will not be allocated (y=0), and the total
number of work data sets will be n.
Note: When message ICE118I is issued indicating that DFSORT cannot
determine the file size, y is set as follows:
v For DYNAPCT=OLD, y is set to n * 50%
v For DYNAPCT=x with x <= 50, y is set to n * 50%
v For DYNAPCT=x with x > 50, y is set to n * x%
DYNSPC
DYNSPC=n
Temporarily overrides the DYNSPC installation option, which specifies the

total default primary space allocation for all of the dynamically allocated work
data sets when the input file size is unknown. That is, when DFSORT cannot
determine the input file size for a sort application and the number of records is
not supplied by a FILSZ or SIZE value. For more information, see the
discussion of the DYNSPC option in OPTION control statement on page 173.
n
specifies the total default primary space, in megabytes, to be allocated for

all dynamically allocated work data sets (n is not the primary space for
each data set). n must be a value between 1 and 65535.
Do not specify a value which exceeds the available disk space, because this
causes dynamic allocation to fail for sort applications that use this value.
39

EFS
EFS=
name
NONE
Temporarily overrides the EFS installation option, which specifies whether

DFSORT passes control to an EFS program. See Chapter 9, Using extended
function support, on page 767 for more information on EFS.
name
specifies the name of the EFS program that will be called to interface with
DFSORT.
NONE
means no call will be made to the EFS program.
Note: If you use locale processing for SORT, MERGE, INCLUDE, or OMIT
fields, you must not use an EFS program. DFSORT's locale processing may
eliminate the need for an EFS program. See OPTION control statement on
page 173 for information related to locale processing.
EQUALS or NOEQUALS
EQUALS
NOEQUALS
Temporarily overrides the EQUALS installation option, which specifies

whether the original sequence of records that collate identically for a sort or a
merge should be preserved from input to output. For more information, see
the discussion of the EQUALS and NOEQUALS options in OPTION control
EQUALS
specifies that the original sequence must be preserved.
NOEQUALS
specifies that the original sequence need not be preserved.
E15=COB
E15=COB
Specifies that your E15 routine is written in COBOL and temporarily overrides
the MODS statement for E15. If you specify E15=COB but do not identify an
E15 module with a MODS statement, the E15=COB is ignored.
40

E35=COB
E35=COB
Specifies that your E35 routine is written in COBOL and temporarily overrides
the MODS statement for E35. If you specify E35=COB but do not identify an
E35 module with a MODS statement, the E35=COB is ignored.
FILSZ
FILSZ=
x
Ex
Ux
Specifies either the exact number of records to be sorted or merged, or an

estimate of the number of records to be sorted. This record count is used by
DFSORT for two purposes:
1. To check that the actual number of records sorted or merged is equal to the
exact number of records expected. FILSZ=x causes this check to be
performed and results in termination with message ICE047A if the check
fails.
2. To determine the input file size for a sort application. DFSORT performs
calculations based on the user supplied record count and other parameters
(such as AVGRLEN) to estimate the total number of bytes to be sorted. This
value is important for sort applications, since it is used for several internal
optimizations as well as for dynamic work data set allocation (see OPTION
DYNALLOC). If no input record count (or only an estimate) is supplied for
the sort application, DFSORT attempts to automatically compute the file
size to be used for the optimizations and allocations.
The type of FILSZ value specified (x, Ex, Ux, or none) controls the way
DFSORT performs the previous two functions, and can have a significant effect
on performance and work data set allocation. See Specify input/output data
set characteristics accurately on page 801 and Allocation of work data sets
on page 855 for more information on file size considerations.
x
specifies the exact number of records to be sorted or merged. This value is
always used for both the record check and file size calculations. FILSZ=x
can be used to force DFSORT to perform file size calculations based on x,
and to cause DFSORT to terminate the sort or merge application if x is not
exact.
If installation option FSZEST=NO is in effect and FILSZ=x is specified,
DFSORT terminates if the actual number of records is different from the
specified value (x), the actual number of records placed in the IN field of
message ICE047A (or message ICE054I) before termination. However, if
installation option FSZEST=YES is in effect, DFSORT treats FILSZ=x like
FILSZ=Ex; it does not terminate when the actual number of records does
not equal x.
The specified value (x) must take into account the number of records in the
41

input data sets, records to be inserted or deleted by exit E15 or E32, and
records to be deleted by the INCLUDE/OMIT statement, SKIPREC, and
STOPAFT. x must be changed whenever the number of records to be sorted
or merged changes in any way.
FILSZ=0 causes Hipersorting, dataspace sorting, and dynamic allocation of
work space not to be used, and results in termination with the message
ICE047A unless the number of records sorted or merged is 0.
Limit: 28 digits (15 significant digits)
Ex
specifies an estimated number of records to be sorted. This value is not
used for the record check. It is used for file size calculations, but only if
DFSORT could not automatically compute the file size. In all other cases,
this value is ignored by DFSORT. See Dynamic allocation of work data
sets on page 856 for details on exactly when FILSZ=Ex is used or ignored
by DFSORT.
The specified value (x) should take into account the number of records in
the input data sets, records to be inserted or deleted by exit E15, and
STOPAFT. x should be changed whenever the number of records to be
sorted changes significantly.
FILSZ=E0 will always be ignored.
Ux
specifies the number of records to be sorted. This value is not used for the
record check, but is always used for file size calculations. FILSZ=Ux can be
used to force DFSORT to perform file size calculations based on x, while
avoiding termination if x is not exact.
The FSZEST installation option has no effect on FILSZ=Ux processing.
The specified value (x) should take into account the number of records in
the input data sets, records to be inserted or deleted by exit E15, and
STOPAFT. x should be changed whenever the number of records to be
sorted changes significantly.
FILSZ=U0 causes Hipersorting, dataspace sorting, and dynamic allocation
of work space not to be used, and can cause degraded performance or
termination with the message ICE046A, if the actual number of records to
be sorted is significantly larger than 0.
Table 4 summarizes: the differences for the three FILSZ variations. Note that
FILSZ=n is equivalent to FILSZ=En if installation option FSZEST=YES is
specified.
Table 4. FILSZ Variations Summary.
Conditions
FILSZ=n
FILSZ=Un
FILSZ=En
Number of records
Exact
Estimate
Estimate
Applications
Sort, merge
Sort
Sort
Terminate if wrong?
Yes
No
No
Use for file size calculation?
Yes
Yes
When DFSORT
cannot compute
file size
n includes records:
42

Table 4. FILSZ Variations Summary. (continued)
Conditions
FILSZ=n
FILSZ=Un
FILSZ=En
In input data sets
Yes
Yes
Yes
Inserted/deleted by E15
Yes
Yes
Yes
Inserted by E32
Yes
No
No
Yes
Yes
Yes
Deleted by SKIPREC
Yes
Yes
Yes
Deleted by STOPAFT
Yes
Yes
Yes
Update n when number of records

changes:
In any way
Significantly
Significantly
Effects of n=0
Hipersorting and Hipersorting and None

DYNALLOC not DYNALLOC not
used
used
Deleted by INCLUDE/OMIT
statement
Note: Using the FILSZ parameter to supply inaccurate information to DFSORT

can negatively affect DFSORT's performance, and when work space is
dynamically allocated, can result in wasted disk space or termination with
message ICE083A or ICE046A. Therefore, it is important to update the record
count value whenever the number of records to be sorted changes significantly.
HIPRMAX
HIPRMAX=
OPTIMAL
n
p%
Temporarily overrides the HIPRMAX installation option, which specifies the

maximum amount of Hiperspace to be used for Hipersorting. For more
information, see the discussion of the HIPRMAX option in OPTION control
OPTIMAL
specifies that DFSORT determines dynamically the maximum amount of
Hiperspace to be used for Hipersorting.
n

Hiperspace to be used for Hipersorting, subject to a limit of nMB. n must
be a value between 0 and 32767. If n is 0, Hipersorting is not used.
%p specifies that DFSORT determines dynamically the maximum amount of

hiperspace to be used for Hipersorting, subject to a limit of p percent of an
appropriate portion of central storage. p must be a value between 0 and
100. If p is 0, Hipersorting is not used. The value calculated for p% is
limited to 32767MB, and is rounded down to the nearest MB.
43

LIST or NOLIST
LIST
NOLIST
Temporarily overrides the LIST installation option, which specifies whether

DFSORT program control statements should be written to the message data
set. See z/OS DFSORT Messages, Codes and Diagnosis Guide for full details on
use of the message data set.
LIST
specifies that all DFSORT control statements are printed on the message
data set.
NOLIST
specifies that DFSORT control statements are not printed.
LISTX or NOLISTX
LISTX
NOLISTX
Temporarily overrides the LISTX installation option, which specifies whether

DFSORT writes to the message data set the program control statements
returned by an EFS program. See z/OS DFSORT Messages, Codes and Diagnosis
Guide for full details on use of the message data set.
LISTX
specifies that control statements returned by an EFS program are printed to
the message data set.
NOLISTX
specifies that control statements returned by an EFS program are not
printed to the message data set.
Note:
1. If EFS=NONE is in effect after final override rules have been applied,
NOLISTX will be set in effect.
2. LISTX and NOLISTX can be used independently of LIST and NOLIST.
3. For more information on printing EFS control statements, see z/OS DFSORT
Messages, Codes and Diagnosis Guide.
LOCALE
44

LOCALE=
name
CURRENT
NONE
Temporarily overrides the LOCALE installation option, which specifies

whether locale processing is to be used and, if so, designates the active locale.
For more information, see the discussion of the LOCALE option in OPTION
control statement on page 173.
name
specifies that locale processing is to be used and designates the name

of the locale to be made active during DFSORT processing.
The locales are designated using a descriptive name. For example, to
set the active locale to represent the French language and the cultural
conventions of Canada, specify LOCALE=FR_CA. You can specify up
to 32 characters for the descriptive locale name. The locale names
themselves are not case-sensitive. See Using Locales for complete locale
naming conventions.
You can use IBM-supplied and user-defined locales.
The state of the active locale prior to DFSORT being entered will be
restored on DFSORT's completion.
CURRENT
specifies that locale processing is to be used, and the current locale
active when DFSORT is entered will remain the active locale during
DFSORT processing.
NONE
specifies that locale processing is not to be used. DFSORT will use the
binary encoding of the code page defined for your data for collating
and comparing.
MOSIZE
MOSIZE=
MAX
n
%p
Temporarily overrides the MOSIZE installation option, which specifies the

maximum amount of memory object storage to be used for memory object
sorting. For more information, see the discussion of the MOSIZE option in
MAX
memory object storage to be used for memory object sorting.
n

memory object storage to be used for memory object sorting, subject to a
limit of nMB. n must be a value between 0 and 2147483646. If n is 0,
memory object sorting is not used.
%p specifies that DFSORT determines dynamically the maximum amount of

memory object storage to be used for memory object sorting, subject to a
limit of p percent of the available central storage. p must be a value
45

between 0 and 100. If p is 0, memory object sorting is not used. The value
calculated for p% is limited to 2147483646MB, and is rounded down to the
nearest MB.
MOWRK or NOMOWRK
MOWRK
NOMOWRK
Temporarily overrides the MOWRK installation option, which specifies

whether the memory object storage available to DFSORT for memory object
sorting can be used as intermediate work space. DFSORT has the capability of
using memory object storage as intermediate work space (similar to the way
Hiperspace is used but more efficient), or as an extension of main storage.
Using memory object storage as intermediate work space is the preferred and
recommended choice, but can be disabled, if appropriate.
MOWRK
specifies that memory object storage can be used as intermediate work
space, or as an extension of main storage, as appropriate.
NOMOWRK
specifies that memory object storage can only be used as an extension of
main storage.
MSGDDN
MSGDDN=ddname
Temporarily overrides the MSGDDN installation option, which specifies an

alternate ddname for the message data set. For more information, see the
discussion of the MSGDDN option in OPTION control statement on page
173.
The ddname can be any 1- through 8-character name, but must be unique
within the job step; do not use a name that is used by DFSORT (for example,
SORTIN). If the ddname specified is not available at run-time, SYSOUT is used
instead. For details on using the message data set, see z/OS DFSORT Messages,
Codes and Diagnosis Guide.
Note: MSGDD=ddname can be used instead of MSGDDN=ddname.
MSGPRT
46

MSGPRT=
ALL
CRITICAL
NONE
Temporarily overrides the MSGPRT installation option, which specifies the

class of messages to be written to the message data set. See z/OS DFSORT
Messages, Codes and Diagnosis Guide for full details on use of the message data
set.
ALL
specifies that all messages except diagnostic messages ICE800I to ICE999I
are printed on the message data set. Control statements are printed only if
LIST is in effect.
CRITICAL
specifies that only critical messages are printed on the message data set.
Control statements are printed only if LIST is in effect.
NONE
specifies that no messages or control statements are printed.
Note: The forms FLAG(I)|FLAG(U)|NOFLAG, and
MSG={NO|NOF|AB|AP|AC|CB|CC|CP|PC|SC|SP} are also accepted. The
following table lists the equivalent MSGPRT/MSGCON specifications for these
alternate forms:
Table 5. Aliases for MSGPRT/MSGCON Options
Option
MSGPRT
MSGCON
NO
NONE
NONE
NOF
NONE
NONE
AB
ALL
ALL
AP
ALL
CRITICAL
AC
NONE
ALL
CB
CRITICAL
CRITICAL
CC
NONE
CRITICAL
CP
CRITICAL
CRITICAL
PC
ALL
ALL
SC
ALL
CRITICAL
SP
CRITICAL
ALL
NOFLAG
NONE
CRITICAL
FLAG(I)
ALL
CRITICAL
FLAG(U)
CRITICAL
CRITICAL

NULLOUT
NULLOUT=
RC0
RC4
RC16
47

Temporarily overrides the NULLOUT installation option, which specifies the
action to be taken by DFSORT when there are no records for the SORTOUT
data set. For more information, see the discussion of the NULLOUT option in
RC0
specifies that DFSORT should issue message ICE173I, set a return code of
0, and continue processing when there are no records for the SORTOUT
data set.
RC4
data set.
RC16
specifies that DFSORT should issue message ICE206A, terminate, and give
a return code of 16 when there are no records for the SORTOUT data set.
ODMAXBF
ODMAXBF=
n
nK
nM
Temporarily overrides the ODMAXBF installation option, which specifies the

maximum buffer space DFSORT can use for each OUTFIL data set. For more
information, see the discussion of the ODMAXBF option in OPTION control
n
specifies that a maximum of n bytes of buffer space is to be used for each

OUTFIL data set. If you specify less than 262144, 262144 is used. If you
specify more than 16777216, 16777216 is used.
Limit: 8 digits
nK specifies that a maximum of n times 1024 bytes of buffer space is to be

used for each OUTFIL data set. If you specify less than 256K, 256K is used.
If you specify more than 16384K, 16384K is used.
Limit: 5 digits
nM specifies that a maximum of n times 1048576 bytes of buffer space is to be
used for each OUTFIL data set. If you specify 0M, 256K is used. If you
specify more than 16M, 16M is used.
Limit: 2 digits
OUTREL or NOOUTREL
48

OUTREL
NOOUTREL
Temporarily overrides the OUTREL installation option, which specifies whether

unused temporary output data set space is to be released.
OUTREL
specifies that unused temporary output data set space is released.
NOOUTREL
specifies that unused temporary output data set space is not released.
Note: RLSOUT and NORLSOUT can be used instead of OUTREL and
NOOUTREL, respectively.
OVFLO
OVFLO=
RC0
RC4
RC16
Temporarily overrides the OVFLO installation option, which specifies the

action to be taken by DFSORT when BI, FI, PD or ZD summary fields
overflow. For more information, see the discussion of the OVFLO option in
RC0
specifies that DFSORT should issue message ICE152I (once), set a return
code of 0 and continue processing when summary fields overflow.
RC4
code of 4 and continue processing when summary fields overflow.
RC16
specifies that DFSORT should issue message ICE195A, terminate and give
a return code of 16 when summary fields overflow.
PAD
PAD=
RC0
RC4
RC16
Temporarily overrides the PAD installation option, which specifies the action to
be taken by DFSORT when the SORTOUT LRECL is larger than the
SORTIN/SORTINnn LRECL, for the cases where DFSORT allows LRECL
padding. For more information, see the discussion of the PAD option in
49

RC0
code of 0 and continue processing when the SORTOUT LRECL is larger
than the SORTIN/SORTINnn LRECL.
RC4
specifies that DFSORT should issue message ICE171I, set a return code of 4
and continue processing when the SORTOUT LRECL is larger than the
SORTIN/SORTINnn LRECL.
RC16
a return code of 16 when the SORTOUT LRECL is larger than the
RESALL
RESALL=
n
nK
nM
Temporarily overrides the RESALL installation option, which specifies the

number of bytes to be reserved in a REGION for system use when
SIZE/MAINSIZE=MAX is in effect. For more information, see the discussion of
the RESALL option in OPTION control statement on page 173.
n
specifies that n bytes of storage are to be reserved. If you specify less than
4096, 4096 is used.
Limit: 8 digits.
nK specifies that n times 1024 bytes of storage are to be reserved. If you

specify less than 4K, 4K is used.
Limit: 5 digits.
nM specifies that n times 1048576 bytes of storage are to be reserved. If you
specify 0M, 4K is used.
Limit: 2 digits.
Note: RESERVE=value can be used instead of RESALL=value.
RESET or NORESET
RESET
NORESET
Temporarily overrides the RESET installation option, which specifies whether

DFSORT should process a VSAM output data set defined with REUSE as a
NEW or MOD data set.
50

RESET
specifies that DFSORT processes a VSAM output data set defined with
REUSE as a NEW data set. The high-used RBA is reset to zero and the
output data set is effectively treated as an initially empty cluster.
NORESET
REUSE as a MOD data set. The high-used RBA is not reset and the output
data set is effectively treated as an initially non-empty cluster.
Note: A VSAM output data set defined without REUSE is processed as a MOD
data set.
SDB
SDB=
LARGE
YES
INPUT
NO
Temporarily overrides the SDB installation option, which specifies whether

DFSORT should use the system-determined optimum block size for output
data sets when the block size is specified as zero or defaulted to zero. For
more information, see the discussion of the SDB option in OPTION control
LARGE
specifies that DFSORT is to use the system-determined optimum block size
for an output data set when its block size is zero. SDB=LARGE allows
DFSORT to select a block size greater than 32760 bytes for a tape output
data set, when appropriate.
YES
for an output data set when its block size is zero, but is to limit the
selected block size to a maximum of 32760 bytes.
INPUT
selected block size to a maximum of 32760 bytes if the input block size is
less than or equal to 32760 bytes.
NO specifies that DFSORT is not to use the system-determined optimum block
size.
Note: SDB, SDB=ON, and SDB=SMALL can be used instead of SDB=YES.
NOSDB and SDB=OFF can be used instead of SDB=NO.
51

SIZE
SIZE=
n
nK
nM
MAX
MAX-m
MAX-mK
MAX-mM
Temporarily overrides the SIZE installation option, which specifies the amount
of main storage available to DFSORT. For more information, see the discussion
of the MAINSIZE option in OPTION control statement on page 173.
n
specifies that n bytes of storage are to be allocated. If you specify more

than 2097152000, 2097152000 is used.
Limit: 10 digits.
nK specifies that n times 1024 bytes of storage are to be allocated. If you

specify more than 2048000K, 2048000K is used.
Limit: 7 digits.
nM specifies that n times 1048576 bytes of storage are to be allocated. If you
Limit: 4 digits.
MAX
instructs DFSORT to calculate the amount of virtual storage available and
allocate an amount of storage up to the TMAXLIM or DSA value when
Blockset is selected, or up to the MAXLIM value when Blockset is not
selected.
MAX-m
specifies the RESALL value (m) in bytes. MAX-m instructs DFSORT to
calculate the amount of storage available and allocate this amount up to
the MAX value minus the amount of storage reserved for system and
application use (RESALL).
If you specify less than 4096 for m, 4096 is used.
Limit for m: 8 digits.
MAX-mK
specifies the RESALL value (m times 1024) in KBs. MAX-mK instructs
DFSORT to calculate the amount of storage available and allocate this
amount up to the MAX value minus the amount of storage reserved for
system and application use (RESALL).
If you specify less than 4K for m, 4K is used.
MAX-mM
specifies the RESALL value (m times 1048576) in s. MAX-mM instructs the
program to calculate the amount of storage available and allocate this
amount up to the MAX value minus the amount of storage reserved for
system and application use (RESALL).
If you specify 0M for m, 4K is used.
52

Note: The forms SIZE(value), CORE=value, and CORE(value) can be used
instead of SIZE=value.
SKIPREC
SKIPREC=z
Specifies the number of records (z) you want to skip (delete) before starting to
sort or copy the input data set. SKIPREC is typically used to bypass records
not processed from the previous DFSORT job. For more information, see the
discussion of the SKIPREC option in OPTION control statement on page 173.
z
specifies the number of records to be skipped.

Limit: 28 digits (15 significant digits).

SOLRF or NOSOLRF
SOLRF
NOSOLRF
Temporarily overrides the SOLRF installation option, which specifies whether

DFSORT should set the SORTOUT LRECL to the reformatted record length
when the SORTOUT LRECL is unknown. For more information, see the
discussion of the SOLRF and NOSOLRF options in OPTION control
SOLRF
specifies that DFSORT should use the reformatted record length for the
SORTOUT LRECL when the SORTOUT LRECL is not specified or
available.
NOSOLRF
specifies that DFSORT should not use the reformatted record length for the
SORTOUT LRECL.
SPANINC
SPANINC=
RC0
RC4
RC16
Temporarily overrides the SPANINC installation option, which specifies the

action to be taken by DFSORT when one or more incomplete spanned records
are detected in a variable spanned input data set. For more information, see
the discussion of the SPANINC option in OPTION control statement on page
173.
53

RC0
code of 0 and eliminate all incomplete spanned records it detects.
RC4
code of 4 and eliminate all incomplete spanned records it detects.
RC16
a return code of 16 when an incomplete spanned record is detected.
override of DFSORT options, on page 863.
STOPAFT
STOPAFT=n
Specifies the maximum number of records you want accepted for sorting or
copying (that is, read from SORTIN or inserted by E15 and not deleted by
SKIPREC, E15, or an INCLUDE/OMIT statement). For more information, see
the discussion of the STOPAFT option in OPTION control statement on page
173.
n
specifies the maximum number of records to be accepted.

Limit: 28 digits (15 significant digits).
Note: If you specify (1) FILSZ=x in the EXEC PARM, or (2) SIZE=x or FILSZ=x
on the OPTION or SORT statement, and the number of records accepted for
processing does not equal x, DFSORT issues an error message and terminates
unless installation option FSZEST=YES was specified.
SZERO or NOSZERO
SZERO
NOSZERO
Temporarily overrides the SZERO installation option, which specifies-+

whether DFSORT should treat numeric 0 and 0 values as signed (that is,
different) or unsigned (that is, the same) for collation, comparisons, editing,
conversions, minimums and maximums. For more information, see the
discussion of the SZERO and NOSZERO options in OPTION control
SZERO
specifies that DFSORT should treat numeric zero values as signed.
NOSZERO
specifies that DFSORT should treat numeric zero values as unsigned.
54

TRUNC
TRUNC=
RC0
RC4
RC16
Temporarily overrides the TRUNC installation option, which specifies the

action to be taken by DFSORT when the SORTOUT LRECL is smaller than the
truncation. For more information, see the discussion of the TRUNC option in
RC0
and continue processing when the SORTOUT LRECL is smaller than the
RC4
RC16
a return code of 16 when the SORTOUT LRECL is smaller than the
VERIFY or NOVERIFY
VERIFY
NOVERIFY
Temporarily overrides the VERIFY installation option, which specifies whether

sequence checking of the final output records must be performed.
VERIFY
specifies that sequence checking is performed.
NOVERIFY
specifies that sequence checking is not performed.
Note:
1. Using VERIFY can degrade performance.
2. SEQ=YES can be used instead of VERIFY. SEQ=NO can be used instead of
NOVERIFY.
55

VLLONG or NOVLLONG
VLLONG
NOVLLONG
Temporarily overrides the VLLONG installation option, which specifies

whether DFSORT is to truncate "long" variable-length output records. For more
information, see the discussion of the VLLONG and NOVLLONG options in
VLLONG
specifies that DFSORT truncates long variable-length output records to the
LRECL of the SORTOUT or OUTFIL data set.
NOVLLONG
specifies that DFSORT terminates if a long variable-length output record is
found.
VLSCMP or NOVLSCMP
VLSCMP
NOVLSCMP
Temporarily overrides the VLSCMP installation option, which specifies

whether DFSORT is to pad "short" variable-length INCLUDE/OMIT compare
fields with binary zeroes. For more information, see the discussion of the
VLSCMP and NOVLSCMP options in OPTION control statement on page
173.
VLSCMP
specifies that short variable-length compare fields are padded with binary
zeros.
NOVLSCMP
specifies that short variable-length compare fields are not padded.
VLSHRT or NOVLSHRT
VLSHRT
NOVLSHRT
Temporarily overrides the VLSHRT installation option, which specifies whether

DFSORT is to continue processing if a "short" variable-length SORT/MERGE
control field, INCLUDE/OMIT compare field, or SUM summary field is found.
For more information, see the discussion of the VLSHRT and NOVLSHRT
options in OPTION control statement on page 173.
VLSHRT
specifies that DFSORT continues processing if a short control field or
compare field is found.
56

NOVLSHRT
specifies that DFSORT terminates if a short control field or compare field is
found.
VSAMEMT or NVSAMEMT
VSAMEMT
NVSAMEMT
Temporarily overrides the VSAMEMT installation option, which specifies

whether DFSORT should accept an empty VSAM input data set.
VSAMEMT
specifies that DFSORT accepts an empty VSAM input data set and
processes it as having zero records.
NVSAMEMT
specifies that DFSORT terminates if an empty VSAM input data set is
found.
Note: VSAMEMT=YES can be used instead of VSAMEMT. VSAMEMT=NO
can be used instead of NVSAMEMT.
VSAMIO or NOVSAMIO
VSAMIO
NOVSAMIO
Temporarily overrides the VSAMIO installation option, which specifies

whether DFSORT should allow a VSAM data set defined with REUSE to be
sorted in-place.
VSAMIO
specifies that DFSORT can use the same VSAM data set for input and
output: when all of the following conditions are met
v The application is a sort.
v RESET is in effect.
v The VSAM data set was defined with REUSE.
These conditions ensure that the VSAM data set is processed as NEW for
output and will contain the sorted input records; that is it will be sorted
in-place.
DFSORT terminates if the same VSAM data set is specified for input and
output, and any of the previous conditions are not met.
57

NOVSAMIO
specifies that DFSORT terminates if the same VSAM data set is specified
for input and output.
WRKREL or NOWRKREL
WRKREL
NOWRKREL
Temporarily overrides the WRKREL installation option, which specifies

whether unused temporary SORTWKdd data set space will be released.
WRKREL
specifies that unused space is released.
NOWRKREL
specifies that unused space is not released.
Note:
1. If you have dedicated certain volumes for SORTWKdd data sets, and you
do not want unused temporary space to be released, you should specify
NOWRKREL.
2. If WRKREL is in effect, DFSORT releases space for the SORTWKdd data
sets just prior to termination. Space is released only for those SORTWKdd
data sets that were used for the sort application.
3. RELEASE=OFF and RLS=0 can be used instead of NOWRKREL.
RELEASE=ON and RLS=n (n greater than 0) can be used instead of
WRKREL.
WRKSEC or NOWRKSEC
WRKSEC
NOWRKSEC
Temporarily overrides the WRKSEC installation option, which specifies

whether DFSORT uses automatic secondary allocation for temporary JCL
SORTWKdd data sets.
WRKSEC
specifies that automatic secondary allocation for temporary JCL
SORTWKdd data sets is used and that 25 percent of the primary allocation
will be used as the secondary allocation.
NOWRKSEC
SORTWKdd data sets is not used.
Note: SECOND=OFF and SEC=0 can be used instead of NOWRKSEC.
SECOND=ON and SEC=n (n greater than 0) can be used instead of WRKSEC.
58

Y2PAST
Y2PAST=
s
f
Temporarily overrides the Y2PAST installation option, which specifies the

sliding (s) or fixed (f) century window. The century window is used with
DFSORT's Y2 formats to correctly interpret two-digit year data values as
four-digit year data values.
s
specifies the number of years DFSORT is to subtract from the current year
to set the beginning of the sliding century window. Since the Y2PAST
value is subtracted from the current year, the century window slides as the
current year changes. For example, Y2PAST=81 would set a century
window of 1925-2024 in 2006 and 1926-2025 in 2007. s must be a value
between 0 and 100.
specifies the beginning of the fixed century window. For example,

Y2PAST=1962 would set a century window of 1962-2061. f must be a value
between 1000 and 3000.
Note: CENTWIN=value can be used instead of Y2PAST=value.

ZDPRINT or NZDPRINT
ZDPRINT
NZDPRINT
Temporarily overrides the ZDPRINT installation option, which specifies

whether positive zoned-decimal (ZD) fields resulting from summing must be
converted to printable numbers. For more information, see the discussion of
the ZDPRINT and NZDPRINT options in OPTION control statement on
page 173.
ZDPRINT
means convert positive ZD summation results to printable numbers.
NZDPRINT
means do not convert positive ZD summation results to printable numbers.
Note: ZDPRINT=YES can be used instead of ZDPRINT. ZDPRINT=NO can be
used instead of NZDPRINT.
59
Aliases for PARM Options
Aliases for PARM options

For compatibility reasons, the following EXEC/DFSPARM PARM options can be
specified by using the aliases shown in the following table. See the indicated
PARM options for complete details.
Table 6. Aliases for PARM Options
PARM Option
60
CENTWIN=value
Y2PAST=value
CORE=value
SIZE=value
FLAG(I)
MSGPRT=ALL
FLAG(U)
MSGPRT=CRITICAL
L5=value
AVGRLEN=value
MSG=value
MSGPRT=value
MSGDD=value
MSGDDN=value
NOFLAG
MSGPRT=NONE
NORC16
NOABEND
NORLSOUT
NOOUTREL
NOSDB
SDB=NO
RC16=ABE
ABEND
RELEASE=ON
WRKREL
RELEASE=OFF
NOWRKREL
RESERVE=value
RESALL=value
RESERVEX=value
ARESALL=value
RLS=n
WRKREL
RLS=0
NOWRKREL
RLSOUT
OUTREL
SDB
SDB=YES
SDB=ON
SDB=YES
SDB=OFF
SDB=NO
SDB=SMALL
SDB=YES
SEC=n
WRKSEC
SEC=0
NOWRKSEC
SECOND=ON
WRKSEC
SECOND=OFF
NOWRKSEC
SEQ=YES
VERIFY
SEQ=NO
NOVERIFY
VSAMEMT=YES
VSAMEMT
VSAMEMT=NO
NVSAMEMT
ZDPRINT=YES
ZDPRINT
ZDPRINT=NO
NZDPRINT
Using DD Statements
Using DD statements
A DFSORT job always requires DD statements after the EXEC statement. DD:
statements fall into two categories
v System DD statements (discussed in detail in System DD statements on page
63)
v Program DD statements (discussed in detail in Program DD statements on
page 65).
System DD statements, and some program DD statements, are usually supplied
automatically when you use a cataloged procedure. Others you must always
supply yourself.
The DD statement parameters, the conditions under which they are required, and
the default values, are summarized in Table 7. The subparameters of the DCB
parameter (a DD statement parameter) are described similarly in Table 8 on page
62.
Note:
1. Performance is enhanced if the LRECL subparameter of the DCB is accurately
specified for variable-length records. The maximum input record length you
can specify for your particular configuration is given in Data set notes and
limitations on page 13.
2. When using DFSORT applications, FREE=CLOSE cannot be used on any DD
statements except DFSPARM.
Table 7. DD Statement Parameters Used by DFSORT
Parameter
When Required
Parameter Values
Default Value
{AMP | BUFSP}
When password-protected
VSAM data sets are used
and the password is
supplied through E18, E38,
or E39.
Minimum buffer pool value None.

given when creating the
data set.
DCB
Required when 7-track tape

is used; for input on tape
without standard labels;
and when the default
values are not applicable.
Specifies information used (See separate

to fill the data control block subparameters in Table 8 on
(DCB) associated with the
page 62.)
data set.
DISP
When the default value is

not applicable.
Indicates the status and

disposition of the data set.
DSNAME or DSN
Specifies the fully qualified

When the DD statement
defines a labeled input data or temporary name of the
set (for example, SORTIN), data set.
or when the data set being
created is to be kept or
cataloged (for example,
SORTOUT), or passed to
another step.
The system assigns a

unique name.
LABEL
When the default value is

not applicable.
Specifies information about

labeling and retention for
the data set.
The system assumes

standard labeling.
SPACE
When the DD statement

defines a new data set on
disk.
Specifies the amount of

See z/OS MVS JCL
space needed to contain the Reference.
data set.
The system assumes (NEW,

DELETE).
61
Using DD Statements
Table 7. DD Statement Parameters Used by DFSORT (continued)
Parameter
When Required
Parameter Values
Default Value
UNIT
When the input data set is

neither cataloged nor
passed or when the data set
is being created.
Specifies (symbolically or
actually) the type and
quantity of I/O units
required by the data set.
See z/OS MVS JCL

Reference.
VOLUME or VOL
When the input data set is

neither cataloged nor
passed, for multireel input
or when the output data set
is on disk and is to be kept
or cataloged.
Specifies information used

to identify the volume or
volumes occupied by the
data set.
See z/OS MVS JCL

Reference.
Default Value
Table 8. DCB Subparameters Used by DFSORT

Subparameter
Condition When Required
Subparameter Values
BUFOFF
When processing data in

ASCII format.
Specifies the length of the

buffer offset or specifies
that the buffer offset is the
block length indicator.
DEN
When the data set is

located on a 7-track tape
unit.
Specifies the density at

which the tape was
recorded.
OPTCD
When processing data in

ASCII format.
Specifies that the tape

processed is in ASCII
format.
TRTCH
When the data set is

located on a tape device
with IDRC and system
IDRC is not used.
Specifies whether data set

is compacted.
BLKSIZE 1, 2
When the DCB parameter

is required and the default
value is not suitable except
on SORTWKdd statements.
Specifies the maximum

v For old data sets, the
length (in bytes) of the
value in the data set
physical records in the data
label.
set.
v For new output data sets,
Specifies the maximum
appropriate values based
length (in bytes) of the
on the input data set or
logical records in the data
RECORD statement
set.
values.
LRECL 2, 3
RECFM
Specifies the format of the

records in the data set.
800 bpi
System default option.
Unless SDB=NO is in
effect, Blockset uses the
system-determined
optimum block size
when the output data set
block size is zero.
Applications which
require a specific output
data set block size
should be changed to
specify that block size
EXPLICITY.
v No default if input on
unlabeled tape or BLP or
NSL specified.
62
Using DD Statements
Duplicate ddnames
If you specify a particular ddname (such as SORTIN) more than once within the
same step, DFSORT uses the first ddname and ignores subsequent duplicates.
Processing continues normally.
In addition, SORTIN0, SORTIN1...SORTIN9 can be specified instead of SORTIN00,
SORTIN01...SORTIN09, respectively. If you specify both SORTINn and SORTIN0n
in the same job step, DFSORT treats them as duplicates, and ignores each usage
after the first. For example, SORTIN2 and SORTIN02 are treated as duplicates and
only SORTIN2 is used.
Note: For a Conventional merge, SORTINn will not be recognized because of the
existing restriction which allows only SORTIN01, SORTIN02...SORTIN16.
Duplicates of these accepted ddnames will be ignored.
Duplicate OUTFIL ddnames are ignored at the OUTFIL statement level as
explained in OUTFIL statements notes on page 374.
Shared tape units

The following pairs of DFSORT data sets can be assigned to a single tape unit:
v The SORTIN data set and the SORTWK01 data set (tape work data set sorts
only)
v The SORTIN data set and the SORTOUT data set or one OUTFIL data set (sort
applications only).
If you want to associate the SORTIN data set with SORTWK01, you can include
the parameter UNIT=AFF=SORTIN in the DD statement for SORTWK01. The AFF
subparameter causes the system to place the data set on the same unit as the
dataset with the ddname following the subparameter (SORTIN, in this case).
In the same way, you can associate the SORTIN data set with the SORTOUT data
set or an OUTFIL data set by including UNIT=AFF=SORTIN in the SORTOUT or
OUTFIL DD statement.
SORTINnn tape data sets must all be on different tape units because they are read
concurrently. SORTOUT and OUTFIL tape data sets must all be on different tape
units because they are written concurrently.
System DD statements
If you choose not to use the SORT or SORTD cataloged procedures to invoke
DFSORT, you might need to supply system DD statements in your input job
stream (See also the following section for DD statements dedicated to DFSORT,
such as SORTIN). The DD statements contained in the cataloged procedure (or
provided by you) are:
//JOBLIB DD
1. See SORTIN DD statement on page 67 and SORTINnn DD statement on page 69.

2. This is the only subparameter allowed for DD * data sets.
3. For padding and truncating fixed-length records, see Data set notes and limitations on page 13.
63
Using DD Statements
Defines your program link library if it is not already known to the system.
//STEPLIB DD
Same as //JOBLIB DD.
//SYSIN DD
Contains DFSORT control statements, comment statements, blank
statements and remarks when DFSORT is invoked with JCL rather than by
another program. It can also contain user exit routines, in object deck
format, to be bound or link-edited by DFSORT.
v If you use DFSPARM, then SYSIN is not necessary unless your job
requires binding or link-editing.
v The SYSIN data set usually resides in the input stream; however, it can
be defined as a sequential data set or as a member of a partitioned data
set.
v The data set must be defined with a RECFM of F or FB. The LRECL can
be 80, or more (when valid). If the LRECL is greater than 80, DFSORT
will use the first 80 bytes of each record.
If user exit routines are in SYSIN, the LRECL must be 80.
v DFSORT supports concatenated SYSIN data sets to the extent that the
system supports like concatenated data sets for BSAM. Refer to z/OS
DFSMS Using Data Sets for further information about like
concatenated data sets.
Note: The OPTION statement keywords EFS, LIST, NOLIST, LISTX,
NOLISTX, LOCALE, MSGPRT, MSGDDN, SMF, SORTDD, SORTIN, and
SORTOUT are used only when they are passed by an extended parameter
list or when in the DFSPARM data set. If they are specified on an OPTION
statement read from the SYSIN or SORTCNTL data set, the keyword is
recognized, but the parameters are ignored.
If you use the DFSPARM DD statement instead, you can specify both
EXEC PARM options and DFSORT control statements in a single source
data set that overrides all other sources. See DFSPARM DD statement on
page 76.
If user exit routines are in SYSIN, make sure that:
v The LRECL of SYSIN is 80.
v The END statement is the last control statement.
v The user exit routines are arranged in numeric order (for example, E11
before E15).
v The user exit routines are supplied immediately after the END control
statement.
v Nothing follows the last object deck in SYSIN.
v A SORTMODS DD statement is included.
If DFSORT is program invoked, and you supply the DFSORT control
statements through the 24-bit or extended parameter list, SORTCNTL, or
DFSPARM, SYSIN remains the source of user exit routines placed in the
system input stream.
//SYSOUT DD
Identifies the DFSORT message data set. The default ddname is SYSOUT,
but you can specify an alternate ddname for the message data set using
the MSGDDN installation or run-time option. Always supply a DD
statement for the message data set if a catalogued procedure is not used.
64
Using DD Statements
(If you are invoking DFSORT from a COBOL program and are using the
ddname SYSOUT for the message data set, the use of DISPLAY in your
COBOL program can produce uncertain printing results.)
DFSORT uses RECFM=FBA, LRECL=121, and the specified BLKSIZE for
the message data set. If the BLKSIZE you specify is not a multiple of 121,
DFSORT uses BLKSIZE=121. If you do not specify the BLKSIZE, DFSORT
selects the block size as directed by the SDBMSG installation option (see
z/OS DFSORT Installation and Customization).
If you use a temporary or permanent message data set, it is best to specify
a disposition of MOD to ensure you see all messages and control
statements in the message data set.
//SYSUDUMP DD
Defines the data set for output from a system ABEND dump routine.
//SYSMDUMP DD
Same as //SYSUDUMP DD.
//SYSABEND DD
Same as //SYSUDUMP DD.
If you are using the supplied SORT cataloged procedure, the DD statements that
follow are automatically supplied. If you are not using the SORT cataloged
procedure and you are using the linkage editor, you must supply the following:
DD statements
//SYSPRINT DD
Contains messages from the binder or linkage editor.
//SYSUT1 DD
Defines the intermediate storage data set for the linkage editor. The binder
does not use this data set.
//SYSLIN DD
Defines a data set for control information for the binder or linkage editor.
//SYSLMOD DD
Defines a data set for output from the binder or linkage editor.
Note: If you do not include user routines, or if you include user routines that do
not require binding or link-editing, you can use the supplied SORTD cataloged
procedure. If you include user routines that require binding or link-editing, you
can use the SORT cataloged procedure.
Program DD statements
Even if you use the SORT or SORTD cataloged procedure to invoke DFSORT, you
might need to supply additional dedicated DD statements. The following list
summarizes each of these statements, and a more detailed explanation of each one
follows.
//SORTLIB DD
Defines the data set that contains special load modules for DFSORT. Can
usually be omitted.
//SYMNAMES DD
65
Using DD Statements
symbol processing. Required only if symbol processing is to be performed.
//SYMNOUT DD
table are to be listed. Optional if SYMNAMES DD is specified. Otherwise
ignored.
//SORTIN DD
Defines the input data set for a sorting or copying application. Will not be
used for a merging application.
//SORTINnn DD
Defines the input data sets for a merging application. Will not be used for
a sorting or copying application.
//SORTWKdd DD
Defines intermediate storage data sets. Usually needed for a sorting
application unless dynamic allocation is requested. Will not be used for a
copying or merging application.
//SORTOUT DD
Defines the SORTOUT output data set for a sorting, merging, or copying
application.
//outfil DD
Defines an OUTFIL output data set for a sorting, merging, or copying
application.
//SORTCKPT DD
Defines the data set used to store the information that the system needs to
restart the sort from the last checkpoint. This is only needed if you are
using the checkpoint facility.
//SORTCNTL DD
Defines the data set from which additional or changed DFSORT control
statements can be read when DFSORT is program-invoked.
//DFSPARM DD
Defines the data set from which both additional or changed DFSORT
program control statements and EXEC statement PARM options can be
read when DFSORT is directly invoked or program invoked.
//SORTDKdd DD
Defines the data set used for a VIO SORTWKdd allocation by DFSORT if it
is dynamically reallocated; SORTDKdd must never be specified in the job
stream.
//SORTDIAG DD
Specifies that all messages and control statements are printed. Used
primarily for diagnostics and debugging.
//SORTSNAP DD
Defines the snap dump data set dynamically allocated by DFSORT.
SORTSNAP must never be specified in the job stream.
66
Using DD Statements
//SORTMODS DD
Defines a temporary partitioned data set. This temporary data set must be
large enough to contain all your user exit routines that appear in SYSIN for
a given application. If none of your routines appear in SYSIN, this
statement is not required. If your routines are in libraries, you must
include DD statements defining the libraries.
DFSORT temporarily transfers the user exit routines in SYSIN to the data
set defined by this DD statement. Then DFSORT calls the binder or linkage
editor for processing.
SORTLIB DD statement
The SORTLIB DD statement can usually be omitted. This statement describes the
data set that contains special DFSORT load modules.
When Required: If installation option SORTLIB=PRIVATE is in effect or dynamic
link edit of user exits is specified
v For sort applications using tape work data sets
v For merge applications for which Blockset cannot be used (see message ICE800I).
The SORTLIB installation option determines whether DFSORT searches a system
library or private library for the load modules required by tape work data set sorts
and Conventional merges.
Example 1 SORTLIB DD Statement:
//SORTLIB DD DSNAME=USORTLIB,DISP=SHR
This example shows DD statement parameters that define a previously cataloged

input data set:
DSNAME
causes the system to search the catalog for a data set with the name
USORTLIB. When the data set is found, it is associated with the ddname
SORTLIB. The control program obtains the unit assignment and volume serial
number from the catalog and, if the volume is not already mounted, writes a
mounting message to the operator.
DISP
indicates that the data set existed before this job step, that it should be kept
after this job step, and that it can be used concurrently by several jobs (SHR).
None of the jobs should change the data set in any way.
For information on the parameters used in the SORTLIB DD statement, the
conditions under which they are required, and the default values assumed if a
parameter is not included, see Table 7 on page 61. The subparameters of the DCB
parameter are described in the same detail in Table 8 on page 62. For more detailed
information, see z/OS MVS JCL Reference and z/OS MVS JCL User's Guide
SYMNAMES DD and SYMNOUT DD statements

See Chapter 8, Using symbols for fields and constants, on page 731 for details.
SORTIN DD statement
The SORTIN DD statement describes the characteristics of the data set in which the
records to be sorted or copied reside and also indicates its location.
When Required: If installation option SORTLIB=PRIVATE is in effect or dynamic
link edit of user exits is specified
67
Using DD Statements
v For sort applications using tape work data sets
v For merge applications for which Blockset cannot be used (see message ICE800I).
The SORTLIB installation option determines whether DFSORT searches a system
library or private library for the load modules required by tape work data set sorts
and Conventional merges.
Data set characteristics: DFSORT accepts empty and null non-VSAM data sets,
and DUMMY data sets, for sorting and copying (be sure to supply RECFM, LRECL
and BLKSIZE). DFSORT also accepts empty VSAM data sets for sorting and
copying provided VSAMEMT is in effect. For non-VSAM data sets, DFSORT
examines the DS1LSTAR field in the format-1 DSCB to determine whether the data
set is empty or null. If DS1LSTAR is zero, DFSORT treats the data set as empty or
null. If the data set is a null multivolume data set and the DS1IND80 flag is off in
the format-1 DSCB of the first volume of the multivolume data set, DFSORT opens
the data set for output to force an end of file (EOF) mark before using the data set
for input.
Note that a null data set is one that has been newly created, but never successfully
closed. Null data sets cannot be processed successfully for a tape work data set
sort. The System Code field in the data set label in the disk Volume Table of
Contents (DSCB in the VTOC) indicates a data set created by the VSE operating
system if it contains the letters DOS or VSE within it. Such data sets are never
treated as null; however, they may be empty. DFSORT cannot process VSE disk
data sets that do not have DOS or VSE within the System Code field.
DFSORT may set what it considers to be appropriate values for missing attributes
(RECFM, LRECL, BLKSIZE) of input data sets based on other attributes, or may
terminate due to a missing attribute. If a missing attribute results in termination, or
you don't want to use a missing attribute set by DFSORT, specify that attribute
explicitly (for example, specify RECFM=VB).
See Data set considerations on page 12 for additional considerations.
The following rules apply to concatenated data sets:
v RECFM must be either all fixed-length or all variable-length for the data sets in
the concatenation.
v BLKSIZE can vary. However, if a tape data set has the largest block size and is
not first in the concatenation, you must specify BLKSIZE explicitly on its DD
statement in the following two situations:
Blockset is selected and the tape data set has a block size greater than 32760
bytes, but the block size is not available from DFSMSrmm or ICETPEX.
Blockset is not selected.
v With fixed-length records, LRECL must be the same for all data sets. With
variable-length records, LRECL can vary. However:
If Blockset is selected: If a tape data set has the largest LRECL and is not first
in the concatenation, you must specify LRECL explicitly on its DD statement
if the LRECL is not available from DFSMSrmm or ICETPEX.
If Blockset is not selected, the first data set in the concatenation must have the
largest LRECL (LRECL can be specified explicitly on its DD statement).
v If the data sets are on unlike devices, you cannot use the EXLST parameter at
user exit E18.
v If Blockset is not selected and BSAM is used, all null data sets must precede all
non-null data sets; otherwise, the results are unpredictable.
68
Using DD Statements
v DFSORT forces an EOF mark on all null data sets whose format-1 DSCB
DS1IND80 flag is off before using BSAM to process the null data sets.
v If you define a data set using the DUMMY parameter, do not concatenate other
data sets to it; the system ignores data sets concatenated to a DUMMY data set.
v VSAM data sets must not be concatenated (system restriction).
v Input cannot consist of both VSAM and non-VSAM data sets.
General coding notes:
v For a copy application, the SORTIN data set should not be the same as the
SORTOUT data set or any OUTFIL data set because this can cause lost or
incorrect data or unpredictable results.
v For a sort application, the SORTIN data set should not be the same as any
SORTWKdd data set because this can cause lost or incorrect data or
unpredictable results. The SORTIN data set can be the same as the SORTOUT
data set or an OUTFIL data set, but this situation can lead to the loss of the data
set if the sort application does not end successfully.
v FREE=CLOSE cannot be specified. User labels are not copied.
Example 2 SORTIN DD statement:
//SORTIN DD DSNAME=INPUT,DISP=SHR
This example shows DD statement parameters that define a previously cataloged

input data set:
DSNAME
causes the system to search the catalog for a data set with the name INPUT.
When the data set is found, it is associated with the ddname SORTIN. The
control program obtains the unit assignment and volume serial number from
the catalog and, if the volume is not already mounted, writes a mounting
message to the operator.
DISP
indicates that the data set existed before this job step, that it should be kept
after this job step, and that it can be used concurrently by several jobs (SHR).
None of the jobs should change the data set in any way.
Example 3 volume parameter on SORTIN DD:
//SORTIN
//
DD
DSN=SORTIN,DISP=(OLD,KEEP),UNIT=3490,
VOL=SER=(75836,79661,72945)
If the input data set is contained on more than one reel of magnetic tape, the
VOLUME parameter must be included on the SORTIN DD statement to indicate
the serial numbers of the tape reels. In this example, the input data set is on three
reels that have serial numbers 75836, 79661, and 72945.
If a data set is not on a disk or on a standard-labeled tape, you must specify DCB
parameters in its DD statement.
SORTINnn DD statement
The SORTINnn DD statements describe the characteristics of the data sets in which
records to be merged reside and indicate the locations of these data sets.
When required: SORTINnn DD statements are always needed for a merge, unless
the merge is invoked from another program and all input is supplied through a
routine at user exit E32.
69
Using DD Statements
Data set characteristics: Input data sets can be either non-VSAM or VSAM, but
not both. DFSORT accepts empty and null non-VSAM data sets, and DUMMY data
sets, for merging (be sure to supply RECFM, LRECL and BLKSIZE). DFSORT also
accepts empty VSAM data sets for merging provided VSAMEMT is in effect. For
non-VSAM data sets, DFSORT examines the DS1LSTAR field in the format-1 DSCB
to determine whether the data set is null or empty. If DS1LSTAR is zero, DFSORT
treats the data set as null or empty. A null data set is one that has been newly
created but never successfully closed. Null data sets cannot be processed
successfully by the Conventional merge technique.
RECFM must be the same for all input data sets.
BLKSIZE can vary, but for a Conventional merge, SORTIN01 must specify the
largest block size.
With fixed-length records, LRECL must be the same for all data sets. With
variable-length records, LRECL can vary.
Data sets can be multivolume but not concatenated. If a SORTINnn data set is
multivolume and null, DFSORT forces an EOF mark on the data set before use.
DFSORT may set what it considers to be appropriate values for missing attributes
(RECFM, LRECL, BLKSIZE) of input data sets based on other attributes, or may
terminate due to a missing attribute. If a missing attribute results in termination, or
you don't want to use a missing attribute set by DFSORT, specify that attribute
explicitly (for example, specify RECFM=VB).
See Data set notes and limitations on page 13 for additional considerations.
v A SORTINnn data set should not be the same as the SORTOUT data set or any
OUTFIL data set because this can cause lost or incorrect data or unpredictable
results.
v You can merge up to 100 data sets with Blockset merge or up to 16 data sets
with Conventional merge. If Conventional merge is selected, check message
ICE800I for the reason Blockset could not be used and correct the indicated
condition, if possible.
With Blockset merge, nn can be any integer from 00 (the initial zero is
optional) to 99, in any order. Blockset merge treats ddnames of the form
SORTINn and SORTIN0n as duplicates, and ignores any occurrences after: the
first. For example, if you have
//SORTIN4 DD . . .
//SORTIN04 DD . . .
the SORTIN04 DD will be ignored.

With Conventional merge, nn can range from 01 to 16. The first number you
use must be 01 and the remainder must follow in numeric order. Numbers
cannot be skipped. Conventional merge cannot use ddnames of the form
SORTIN0-SORTIN9, SORTIN00 or SORTIN17-SORTIN99.
v FREE=CLOSE cannot be specified. User labels are not copied.
Example 4 SORTIN01-03 DD statements (merge):
//SORTIN01 DD DSNAME=MERGE1,VOLUME=SER=000111,DISP=OLD,
//
LABEL=(,NL),UNIT=3590,
//
DCB=(RECFM=FB,LRECL=80,BLKSIZE=32000)
70
Using DD Statements
//
//
//
//
Example 5 SORTIN01-02 DD statements (merge):

//SORTIN01 DD DSNAME=INPUT1,VOLUME=SER=000101,
//
UNIT=3390,DISP=OLD
//SORTIN02 DD DSNAME=INPUT2,VOLUME=SER=000201,
//
UNIT=3390,DISP=OLD
*
*DCB PARAMETERS
*SUPPLIED FROM
*LABELS
SORTWKdd DD statement
The SORTWKdd DD statements describe the characteristics of the data sets used as
intermediate storage areas for records to be sorted; they also indicate the location
of these data sets.
Up to 255 SORTWKdd DD statements can be specified. However, if you specify
more than 32 and the Blockset technique is not selected, only the first 32 are used.
When required: One or more SORTWKdd statements are required for each sort
application (but not a merge or copy), unless:
v Input can be contained in main storage
v Dynamic work space allocation has been requested (DYNALLOC)
v dataspace sorting, Hipersorting, or memory object sorting is used.
For information on using work data sets, see Appendix A, Using work space, on
page 853.
Diagnostic message ICE803I gives information on intermediate storage allocation
and use.
Devices: SORTWKdd data sets can be on disk or on tape, but not both. Disk
types can be mixed.
Tape must be nine-track unless input is on seven-track tape, in which case work
tapes can (but need not) be seven-track.
v Unless the input file is very large, two or three SORTWKdd data sets are usually
sufficient. Two or three large SORTWKdd data sets are preferable to several
small data sets. Placing each SORTWKdd data set on a separate device can
improve performance.
For optimum allocation of resources such as virtual storage, avoid specifying a
large number of work data sets unnecessarily.
v A SORTWKdd data set should not be the same as the SORTIN data set, the
SORTOUT data set, any OUTFIL data set, or any other SORTWKdd data set
because this can cause lost or incorrect data or unpredictable results.
v Cylinder allocation is preferable for performance reasons. Temporary
SORTWKdd data sets allocated in tracks or blocks (without ROUND) are
readjusted to cylinders by DFSORT.
v For disk work data sets, any valid ddname of the form SORTWKdd or
SORTWKd can be used (for example, SORTWK01, SORTWKC3, SORTWK2,
SORTWK#5, SORTWKA, SORTWKXY and so on). The ddnames can be in any
order. SORTWKd and SORTWK0d are not treated as duplicate ddnames (for
71
Using DD Statements
v
v
v
example, SORTWK5 and SORTWK05 will both be used if specified, as will

SORTWKQ and SORTWK0Q). If you specify more than 255 ddnames and the
Blockset technique is selected, only the first 255 ddnames are used. If you
specify more than 32 ddnames and the Blockset technique is not selected, only
the first 32 ddnames are used.
For tape work data sets, at least three SORTWKdd data sets must be specified.
The first three ddnames must be SORTWK01, SORTWK02 and SORTWK03.
Subsequent ddnames, if specified, must be in order from SORTWK04 through
SORTWK32, with no numbers skipped,
FREE=CLOSE cannot be specified.
Spool, dummy, subsystem, VSAM and z/OS UNIX file system data sets, and
z/OS UNIX files, must not be specified as work data sets
Parameters relating to ASCII data should not be included for tape work data
sets.
Disk work data set coding notes:

v Data sets must be physical sequential; they cannot be partitioned or extended
format.
v The SPLIT cylinder parameter must not be specified.
v If no secondary allocation is requested for temporary SORTWKdd data sets,
automatic secondary allocation will be used unless NOWRKSEC is in effect.
(Secondary allocation is limited to 12 work data sets in the Peerage and Vale
sorting techniques only.)
v If the data set is allocated to VIO, there is no automatic secondary allocation.
v Secondary allocation can be requested for work data sets. For the Peerage and
Vale sorting techniques only, secondary allocation is limited to the first 12 work
data sets; if more work data sets are defined, they are used with only the
primary allocation.
v DFSORT uses only the space on the first volume specified for a multivolume
data set. Space on the second and subsequent volumes is not used. Multivolume
SORTWKdd data sets are, therefore, treated as single-volume SORTWKdd data
sets.
v DFSORT work data sets can exceed 65536 tracks if allocated as large format data
sets on disk devices. Dynamically allocated disk work data sets are
automatically allocated by DFSORT as large format. DSNTYPE=LARGE must be
specified on SORTWKdd DD statements to allocate JCL work data sets as large
format.
v If primary space is fragmented, all but the first fragment are handled as
secondary space.
Virtual I/O: If a SORTWKdd data set is specified on a virtual device:
v With VIO=NO: DFSORT performs dynamic reallocation using the ddname
SORTDKdd on a real device with the same device type as the virtual device. If a
real device corresponding to the virtual device is not available in the system,
DFSORT terminates with an ICE083A message; see z/OS DFSORT Messages, Codes
and Diagnosis Guide for more information about this error. Non-VIO SORTWKdd
data sets are also reallocated when VIO SORTWKdd data sets are present.
v With VIO=YES: the virtual device is used; performance may be degraded.
v DFSORT work data sets cannot be allocated as large format data sets if VIO is
used.
72
Using DD Statements
The following is an example of a SORTWKdd DD statement using a disk work:
data set
Example 6 SORTWK01 DD statement, disk work data set:
//SORTWK01
DD
SPACE=(CYL,(15,5)),UNIT=3390
If you use the checkpoint/restart facility and need to make a deferred restart, you
must make the following additions to the previous statement so that the sort work
data set is not lost:
DSNAME=name1,DISP=(NEW,DELETE,CATLG)
Thus the same SORTWKdd DD statement for a deferred restart would be:
//SORTWK01
//
DD
DSNAME=name1,UNIT=3390,SPACE=(CYL,(15,5)),
DISP=(NEW,DELETE,CATLG)
The following is an example of SORTWKdd DD statements using three tape

devices.
Example 7 SORTWK01-03 DD statement, tape intermediate storage:
//SORTWK01
//SORTWK02
//SORTWK03
DD
DD
DD
UNIT=3480,LABEL=(,NL)
If DFSORT terminates unsuccessfully and the previous DD statements have been

specified, the intermediate storage data sets remain in the system until the step has
been successfully rerun or until the data sets have been deleted by some other
means.
These parameters specify unlabeled data sets on three 3480 tape units. Because the
DSNAME parameters are omitted, the system assigns unique names.
SORTOUT and OUTFIL DD statements

The SORTOUT and OUTFIL DD statements describe the characteristics of the data
sets in which the processed records are to be placed and indicate their location.
The SORTOUT DD statement specifies the single non-OUTFIL output data set for a
sort, copy, or merge application. OUTFIL processing does not apply to SORTOUT.
The FNAMES and FILES parameters of one or more OUTFIL statements specify
the ddnames of the OUTFIL data sets for a sort, copy, or merge application. The
parameters specified for each OUTFIL statement define the OUTFIL processing to
be performed for the OUTFIL data sets associated with that statement. For specific
information about OUTFIL processing, see OUTFIL control statements on page
223.
Although the ddname SORTOUT can actually be used for an OUTFIL data set, the
term SORTOUT will be used to denote the single non-OUTFIL output data set.
When required: Each ddname specified in an OUTFIL statement requires a
corresponding DD statement for that OUTFIL data set.
If you do not specify OUTFIL statements, a SORTOUT DD statement is required
unless you provide an E35 user exit that disposes of all output. A SORTOUT DD
statement is ignored if your program invokes DFSORT and passes the address of
an E35 user exit in the parameter list.
73
SORTOUT and OUTFIL DD Statements

If you specify OUTFIL statements, you do not have to specify a SORTOUT DD
statement or an E35 user exit, although you can use either or both.
Data set characteristics: See Data set considerations on page 12 for additional
considerations.
Block size: Unless SDB=NO is in effect, Blockset uses the system-determined
optimum block size in most cases when the output data set block size is zero. See
the discussion of the SDB option in OPTION control statement on page 173 for
complete details about DFSORT's use of system-determined block size.
For some jobs, the selection of a larger output data set block size can require an
increase in the amount of storage needed for successful DFSORT processing.
Applications which require a specific output data set block size should be changed
to specify that block size explicitly.
If SDB=NO is in effect, DFSORT selects an appropriate (though not necessarily
optimum) block size for the output data set based on the available attributes from
the output data set, the input data set, and the RECORD statement. The output
data set block size will not necessarily be the same as the input block size.
Reblockable indicator: DFSORT sets the reblockable indicator in the output data
set label when
Blockset is selected and
v DFSORT sets the system-determined optimum block size for the output data set
(see Block size) or
v Allocation sets the system-determined optimum block size for the output data
set before DFSORT gets control.
v For a copy application, neither the SORTOUT data set nor any OUTFIL data set
should be the same as the SORTIN data set because this can cause lost or
v For a merge application, neither the SORTOUT data set nor any OUTFIL data set
should be the same as any SORTINnn data set because this can cause lost or
v For a sort application, the SORTOUT data set or an OUTFIL data set can be the
same as the SORTIN data set, but this situation can lead to the loss of the data
set if the sort application does not end successfully.
v An OUTFIL data set should not be the same as the SORTOUT data set or any
other OUTFIL data set because this can cause lost or incorrect data or
unpredictable results.
v Do not specify OPTCD=W for a full function IBM 3480 tape unit; it is
overridden. For a 3480 operating in 3420 compatibility mode (specified as
3400-9), the OPTCD=W request is not overridden, but performance might be
degraded.
v If no secondary allocation is requested for a temporary or new output data set,
automatic secondary allocation will be used unless NOOUTSEC is in effect.
v The RECFM, LRECL, and BLKSIZE in a tape label are used only for a tape
output data set with DISP=MOD, a DD volser present, and an AL, SL, or NSL
label, when appropriate.
v FREE=CLOSE cannot be specified.
74
SORTOUT and OUTFIL DD Statements

v See the discussion of the SOLRF and NOSOLRF options in OPTION control
statement on page 173 for information related to the SORTOUT LRECL.
Example 8 SORTOUT DD statement:
//SORTOUT DD DSN=C905460.OUTPT,UNIT=3390,SPACE=(CYL,5),
//
DISP=(NEW,CATLG)
DISP
specifies the data set unknown to the operating system (NEW) and catalogs
(CATLG) it under the name C905460.OUTPT.
DSNAME
specifies that the data set is called C905460.OUTPT.
SPACE
requests five cylinders of storage for the data set.
UNIT
Indicates that the data set is on a 3390.
SORTCKPT DD statement
The SORTCKPT data set can be allocated on any device that operates with the
Basic Sequential Access Method (BSAM). Processing must be restarted only from
the last checkpoint taken.
Example 9 SORTCKPT DD statement:
//SORTCKPT DD DSNAME=CHECK,VOLUME=SER=000123,
//
DSP=(NEW,KEEP),UNIT=3480
When you allocate the SORTCKPT data set, you must include at least one work
data set.
If the CKPT operand is specified on the OPTION or SORT control statement, more
intermediate storage could be required.
If you want to use the Checkpoint/Restart Facility, refer to Checkpoint/restart
on page 907.
SORTCNTL DD statement
The SORTCNTL data set can be used to supply DFSORT control statements,
comment statements, blank statements, and remarks when DFSORT is invoked
from another program (written, for example, in COBOL or PL/I).
v The SORTCNTL data set usually resides in the input stream, but can be defined
as a sequential data set or as a member of a partitioned data set.
v The data set must be defined with RECFM of F or FB. The LRECL can be 80, or
more (when valid). If the LRECL is greater than 80, DFSORT will use the first 80
bytes of each record.
v DFSORT supports concatenated SORTCNTL data sets to the extent that the
system supports like concatenated data sets for BSAM. Refer to z/OS DFSMS
Using Data Sets for further information about like concatenated data sets.
v When DFSORT is invoked from a PL/I program, the SORTCNTL or DFSPARM
data set must not be used to supply a new RECORD control statement.
Example 10 SORTCNTL DD statement:
//SORTCNTL DD *
OPTION MAINSIZE=8M
75
SORTCNTL DD Statement
Note:
1. The OPTION statement keywords EFS, LIST, NOLIST, LISTX, NOLISTX,
LOCALE, MSGPRT, MSGDDN, SMF, SORTDD, SORTIN, and SORTOUT are
used only when they are passed by an extended parameter list or when in the
DFSPARM data set. If they are specified on an OPTION statement read from
the SYSIN or SORTCNTL data set, the keyword is recognized, but the
parameters are ignored.
If your program invokes DFSORT more than once, you can direct DFSORT to
read different versions of the SORTCNTL data set at each call. See the
explanation of the SORTDD parameter in OPTION control statement on page
173.
2. If you use the DFSPARM DD statement instead of the SORTCNTL DD
statement, you can specify both EXEC PARM options and DFSORT control
statements in a single source data set that overrides all other sources. See
DFSPARM DD statement. For override rules, see Appendix B,
Specification/override of DFSORT options, on page 863.
DFSPARM DD statement
The DFSPARM DD statement can be used to supply DFSORT program control
statements and EXEC statement PARM options from a single DD source. Because
statements in the DFSPARM data set are read whether DFSORT is program
invoked or directly invoked, you can specify EXEC PARM options when invoking
DFSORT from another program (unlike SORTCNTL). DFSPARM accepts all
DFSORT program control statements and all EXEC statement PARM options
(including those ignored by SYSIN and SORTCNTL) and any equivalent options
specified on a DFSORT OPTION statement.
DFSPARM also accepts comment statements, blank statements, and remarks.
For examples of using DFSPARM when you call DFSORT from a program, see
Overriding DFSORT control statements from programs on page 542.
Full override and applicability details are listed as follows and in Appendix B,
v If you use DFSPARM, SYSIN is not necessary unless your job requires
link-editing.
v The DFSPARM data set usually resides in the input stream, but it can be defined
as a sequential data set or as a member of a partitioned data set.
v The data set must be defined with RECFM of F or FB. The LRECL can be 80, or
more (when valid). If the LRECL is greater than 80, DFSORT will use the first 80
bytes of each record.
v DFSORT supports concatenated DFSPARM data sets to the extent that the
system supports like concatenated data sets for BSAM. Refer to z/OS DFSMS
Using Data Sets for further information about like concatenated data sets.
v When DFSORT is invoked from a PL/I program, the SORTCNTL or DFSPARM
data set must not be used to supply a new RECORD control statement.
Note: The ddname DFSPARM is used throughout this document to refer to this
data set source for EXEC PARM options and DFSORT program control statements.
When your system programmers installed DFSORT, they might have changed this
name to one more appropriate for your site with the PARMDDN installation
option. However, DFSORT will always use a DFSPARM data set of present, unless
a DD statement with the PARMDDN name is also present.
76
DFSPARM DD Statement
General coding notes: Coding of parameters in the DFSPARM DD statement
follows the same rules used for the JCL EXEC statement PARM options and the
program control statements specified in SYSIN or SORTCNTL. The following
exceptions apply:
v Labels are not allowed.
v PARM options and program control statements cannot be mixed on the same
line, but can be specified in any order on different lines.
v PARM options must be specified without the PARM= keyword and without
quote marks.
v Commas (or semicolons) are accepted, but not required, to continue PARM
options to another line.
v Leading blanks are not required for PARM options, but at least one leading
blank is required for program control statements.
FREE=CLOSE can be used for applicable DFSPARM data sets (for example, with
temporary and permanent sequential data sets, but not with DD * data sets).
When DFSORT is called from another program, FREE=CLOSE causes the
DFSPARM data set to be released when DFSORT returns to the caller. This allows
another DFSPARM data set to be used for a subsequent call.
For example, if a COBOL program contains three SORT verbs, the following would
cause the control statements in DP1 to be used for the first SORT verb, the control
statements in DP2 to be used for the second SORT verb, and the: control
statements in DP3 to be used for the third SORT verb
//DFSPARM DD DSN=DP1,DISP=SHR,FREE=CLOSE
Without FREE=CLOSE, DP1 would be used for all three SORT verbs.
Example 11 DFSPARM DD statement:
//DFSPARM DD *
SORT FIELDS=(1,2,CH,A),STOPAFT=300
ABEND
OPTION SORTIN=DATAIN
STOPAFT=500
In this example, the DFSPARM DD data set passes a DFSORT SORT statement, the
ABEND and STOPAFT parameters equivalent to specifying
PARM=ABEND,STOPAFT=500 in a JCL EXEC statement, and a DFSORT OPTION
statement.
Note:
1. SORT and OPTION are control statements. ABEND and STOPAFT=500 are
PARM options.
2. The PARM option STOPAFT=500 overrides the SORT control statement option
STOPAFT=300.
3. When PARMDDN=DFSPARM is specified or defaulted
v if a //DFSPARM DD data set is available at run-time, DFSORT will use it
v if a //DFSPARM DD data set is not available at run-time, DFSORT will use
a //$ORTPARM DD data set if available.
77
DFSPARM DD Statement
Thus with PARMDDN=DFSPARM, you can choose to specify either a
//DFSPARM DD data set or a //$ORTPARM DD data set for a particular
DFSORT application.
4. When PARMDDN=ddname is specified
v if a //ddname DD data set is available at run-time, DFSORT will use it
v if a //ddname DD data set is not available at run-time, DFSORT will use a
//DFSPARM DD data set if available.
Thus with PARMDDN=ddname, you can choose to specify either a //ddname
DD data set or a //DFSPARM DD data set for a particular DFSORT
application.
Example 12 DFSPARM DD statement:
//DFSPARM DD *
SORT FIELDS=(5,2,CH,D),SKIPREC=10
STOPAFT=100,BSAM,SKIPREC=5
OPTION SORTIN=DATAIN,SKIPREC=20
In this example, the DFSPARM DD data set contains a SORT program control
statement, three PARM options on one line, and an OPTION program control
statement.
Note: Because PARM options override program control statements, DFSORT uses
SKIPREC=5 and ignores the other SKIPREC specifications.
For information on the parameters used in the DFSPARM DD statement, the
conditions under which they are required, and any default values assumed if a
parameter is omitted, see Specifying EXEC/DFSPARM PARM options on page 32
and Chapter 3, Using DFSORT program control statements, on page 81.
SORTDKdd DD statement
SORTWKdd data sets can be assigned to VIO. If the VIO installation option is
specified or defaults to NO, SORTWKdd data sets are deallocated and reallocated
by DFSORT using SORTDKdd ddnames. SORTDKdd ddnames are reserved for use
by DFSORT.
SORTDIAG DD statement
The SORTDIAG DD statement specifies that all messages, including diagnostic
messages (ICE800I through ICE999I), and control statements are to be written to
the message data set. The statement can be used for all DFSORT techniques and
provides information on EXCP counts, intermediate storage allocation and use, and
so on. The SORTDIAG DD statement has no effect on console messages. The
statement is intended as a diagnostic tool.
When SORTDIAG is used, a SYSOUT DD statement or a ddname DD statement
(where ddname is the alternate message data set ddname specified during
installation or run-time) should be provided. If installation option
NOMSGDD=QUIT is in effect and neither an alternate message data set ddname
statement nor a SYSOUT ddname statement is provided, DFSORT terminates with
a return code of 20.
Example 13 SORTDIAG DD statement:
//SORTDIAG DD
DUMMY
SORTSNAP DD statement
The SORTSNAP DD statement defines the data set where the snap dumps
requested by the ESTAE recovery routine, or the snap dumps requested before or
78
SORTSNAP DD Statement
after a call to an EFS program are printed. SORTSNAP is dynamically allocated by
DFSORT whenever it is required. The ddname, SORTSNAP, is reserved for
DFSORT.
79
80
Chapter 3. Using DFSORT program control statements

Using program control statements
Program control statements direct DFSORT in processing your records. Some
program control statements are required while others are optional. You use the
control statements to:
v
v
v
v
v
v
Indicate whether a sort, merge, or copy is performed.

Describe the control fields to be used.
Indicate program exits for transferring control to your own routines.
Describe DFSORT functions you want to have invoked.
Describe input and output files.
Indicate various options you want to use during processing.
You can supply program control statements to DFSORT from:

v A SYSIN data set
v A SORTCNTL data set
v A DFSPARM data set
v A 24-Bit parameter list
v An extended parameter list
See Appendix B, Specification/override of DFSORT options, on page 863 for an
explanation of when to use each source.
This chapter begins with a summary of DFSORT program control statements and
coding rules. A detailed description of each statement follows.
Control statement summary

Describing the primary task
The only required program control statement in a DFSORT application is a SORT,
MERGE, or OPTION statement that specifies whether you want to sort, merge, or
copy records. (Copying can be specified on any of the three statements.)
SORT
Describes control fields if you are coding a sort application, or specifies a
copy application. Indicates whether you want ascending or descending
order for the sort.
MERGE
Describes control fields if you are coding a merge application, or specifies
a copy application. Indicates whether you want ascending or descending
order for the merge.
OPTION
Overrides installation defaults (such as EQUALS, CHALT, and CHECK)
and supplies optional information (such as DYNALLOC and SKIPREC).
Can specify a copy application.
81
Control Statement Summary
Including or omitting records

You can specify whether certain records are included in the output data sets or
omitted from them.
INCLUDE
Specifies that only records whose fields meet certain criteria are included.
OMIT
Specifies that any records whose fields meet certain criteria are deleted.
OUTFIL
Specifies the records to be included or omitted in multiple output data
sets.
Reformatting and editing records

You can modify individual records by deleting and reordering fields and inserting
blanks, zeros, or constants.
INREC
Specifies how records are reformatted before they are sorted, copied, or
merged.
OUTREC
Specifies how records are reformatted after they are sorted, copied, or
merged.
OUTFIL
Specifies how records are reformatted in multiple output data sets.
Producing multiple output and reports and converting records

You can produce multiple output data sets and reports, convert variable-length
records to fixed-length records, and convert fixed-length records to variable-length
records.
OUTFIL
Specifies the output data sets and which records are to appear in each.
Specifies how records are to be converted from variable-length to
fixed-length or from fixed-length to variable-length.
Joining two files

You can perform various types of "join" applications on two files (F1 and F2) by
one or more keys.
JOINKEYS
One JOINKEYS statement is required for each input file to indicate the
ddname of the file, describe the keys, indicate whether the file is already
sorted by those keys, and so on.
JOIN
An inner join is performed by default, but a JOIN statement can be used to

specify a different type of join.
REFORMAT
Describes the fields from the two files to be included in the joined records,
and optionally an indicator of where the key was found ('B' for both files,
'1' for file1 only or '2' for file2 only.
82
Control Statement Summary

The three JOINKEYS application control statements (JOINKEYS, JOIN and
REFORMAT) are completely described in Chapter 4, Using a JOINKEYS
application for joining two files, on page 457.
Invoking additional functions and options

You can use the remaining control statements to perform a variety of tasks.
ALTSEQ
Specifies changes to the ALTSEQ translation table to be used for SORT,
MERGE, INCLUDE or OMIT fields with format AQ, for INREC, OUTREC,
and OUTFIL fields with TRAN=ALTSEQ, and for INREC, OUTREC, and
OUTFIL fields with format AQ in logical expressions.
DEBUG
Specifies various diagnostic options.
END
Causes DFSORT to discontinue reading SYSIN, SORTCNTL, or DFSPARM.
MODS
Specifies use of one or more user exit routines in a DFSORT application.
See Chapter 5, Using your own user exit routines, on page 487 for
information about user exit routines.
RECORD
Can be used to supply length and type information.
SUM
Specifies that numeric summary fields in records with equal control fields
are summed in one record and that the other records are deleted.
Using symbols
You can define and use a symbol for any field or constant in the following
DFSORT control statements: INCLUDE, INREC, JOINKEYS, MERGE, OMIT,
OUTFIL, OUTREC, REFORMAT, SORT and SUM. You can use a symbol for a
number associated with various keywords in the following DFSORT control
statements: INREC, JOINKEYS, MERGE, OPTION, OUTFIL, OUTREC and SORT.
You can also use a symbol for an output column in the following DFSORT control
statements: INREC, OUTFIL and OUTREC. This makes it easy to create and reuse
collections of symbols (that is, mappings) representing information associated with
various record layouts. You can use system symbols (for example, &JOBNAME.) in
your symbol constants. You can use SET and PROC symbols in your symbol
constants. See Chapter 8, Using symbols for fields and constants, on page 731 for
complete details.
General coding rules

See Inserting comment statements on page 87 for an explanation of how to use
comment statements, blank statements, and remarks. DFSORT program control
statements and EXEC PARM options can also be specified together in a
user-defined DD data set. See DFSPARM DD statement on page 76 for special
coding conventions that apply to this DD source.
All other DFSORT control statements have the same general format, shown in
Figure 4 on page 84. The illustrated format does not apply to control statements
83
General Coding Rules

you supply in a parameter list. See Chapter 6, Invoking DFSORT from a
program, on page 541 for information on the special rules that apply.
Column 1 must be blank
unless a label is present
72 73
(Label)
Operation
Operand
(Remarks)
80
(Sequence or
Identification)
(Continuation column)
Figure 4. Control Statement Format
The control statements are free-form; that is, the operation definer, operands, and
comment field can appear anywhere in a statement, provided they appear in the
proper order and are separated by one or more blank characters. Column 1 of each
control statement must be blank, unless the first field is a label or a comment
statement (see Inserting comment statements on page 87).
v Label Field
A label can be specified on any control statement in SYSIN or SORTCNTL. A
label is never required. If present, a label must begin in column 1 with any
character other than a blank or asterisk (*). A label can be 1 to 70 characters and
ends when a blank character is found. Any character can be used in a label. A
label followed only by blanks is printed but otherwise not processed.
Labels cannot be specified in the parameter list, in DFSPARM or in continuation
lines.
To skip the label, specify one or more blanks starting in column 1.
The following illustrates the use of control statements with and without labels:
OPTION EQUALS
MYSORT SORT FIELDS=(5,4,CH,A)
OUTREC FIELDS=(1,20,51,30)
OUT_1 OUTFIL FNAMES=OUT1,INCLUDE=(5,1,CH,EQ,CA)
OUT_2 OUTFIL FNAMES=OUT2,INCLUDE=(5,1,CH,EQ,CB)
v Operation Field
This field can appear anywhere between column 2 and column 71 of the first
line. It contains a word (for example, SORT or MERGE) that identifies the
statement type to the program. In the example shown later in this section, the
operation definer, SORT, is in the operation field of the sample control statement.
v Operand Field
The operand field is composed of one or more operands separated by commas
or semicolons. This field must follow the operation field, and be separated from
it by at least one blank. No blanks are allowed within the parameters, but a
blank is required at the end of all parameters. If the statement occupies more
than one line, the operand must begin on the first line. Each operand has an
operand definer, or parameter (a group of characters that identifies the operand
type to DFSORT). A value or a list of values can be associated with a parameter.
The five possible operand forms shown in this chapter are:
parameter. Examples: CHALT, NOCHALT, REMOVECC
parameter(c). Examples: DATE1(/), TIME2(:), Y2W(-)
parameter=value. An operand shown in the form parameter=value can also
be specified in the equivalent form parameter(value) or parameter=(value).
84

Examples: AVGRLEN=100, AVGRLEN(100), and AVGRLEN=(100) are
equivalent, DYNALLOC=SYSDA, DYNALLOC(SYSDA), and
DYNALLOC=(SYSDA) are equivalent, SAMPLE=20, SAMPLE(20), and
SAMPLE=(20) are equivalent.
parameter=(list). An operand shown in the form parameter=(list) can also be
specified in the equivalent form parameter(list). Examples:
FIELDS=(21,5,CH,A) and FIELDS(21,5,CH,A) are equivalent,
BUILD=(1,20,HEX) and BUILD(1,20,HEX) are equivalent.
parameter=(value). An operand shown in the form parameter=(value) can
also be specified in the equivalent form parameter(value). Examples:
DATE=(4MD/) and DATE(4MD/) are equivalent, TIMENS=(12) and
TIMENS(12) are equivalent, NOMATCH=(C'***') and NOMATCH(C'***') are
equivalent.
The following example illustrates the parameter, parameter=value and
parameter=(list) forms.
SORT EQUALS,FORMAT=CH,FIELDS=(10,30,A),STOPAFT=1000
The following example illustrates the parameter, parameter(value),

parameter(list) and parameter=(value) forms.
SORT EQUALS,FORMAT(CH),FIELDS(10,30,A),STOPAFT=(1000)
The two previous SORT statements are equivalent.

v Remark Field
This field can contain any information. It is not required, but if it is present, it
must be separated from the last operand field by at least one blank.
v Continuation Column (72)
Any character other than a blank in this column indicates that the present
statement is continued on the next line. However, as long as the last character of
the operand field on a line is a comma or semicolon or colon followed by a
blank, the program assumes that the next line is a continuation line. The
nonblank character in column 72 is required only when a remark field is to be
continued or when an operand is broken at column 71.
v Columns 73 through 80
This field can be used for any purpose.
Continuation lines
The format of the DFSORT continuation line is shown in Figure 5 on page 86.
85

Column 1 must
be blank
16
72 73
Continued operand or remarks
80
Optional use
(Continuation column)
Figure 5. Continuation Line Format
The continuation column and columns 73 through 80 of a continuation line have

the same purpose as they do on the first line of a control statement. Column 1
must be blank.
A continuation line is treated as a logical extension of the preceding line. Either an
operand or a remark field can begin on one line (referred to as "line 1" in the
bullets that follow) and continue on the next line (referred to as "line 2" in the
bullets that follow). The following are the rules for continuation illustrated with
examples (these different types of continuation can be intermixed):
v Implicit continuation in 2-71: If line 1 breaks at a comma-blank or
semicolon-blank or colon-blank, DFSORT continues on line 2 with the first
nonblank character it finds in columns 2-71. For example:
*
1
2
3
4
5
6
7
*23456789012345678901234567890123456789012345678901234567890123456789012
INCLUDE COND=(5,4,CH,EQ,
CABCD)
SORT FIELDS=(9,
3,
ZD,
A)
OUTREC FIELDS=(1,27,2X,
FIRST FIELD AND TWO BLANKS
51,2,BI,M11,
SECOND FIELD
60:9,3,ZD,PD) THIRD FIELD
The previous statements will be treated as if they were specified as:

INCLUDE COND=(5,4,CH,EQ,CABCD)
SORT FIELDS=(9,3,ZD,A)
OUTREC FIELDS=(1,27,2X,51,2,BI,M11,60:9,3,ZD,PD)
v Explicit continuation in 16: If line 1 breaks at column 71 with a nonblank in

column 72, and columns 2-15 of line 2 are blank, DFSORT continues on line 2
with whatever character it finds in column 16 (blank or nonblank). For example:
*
1
2
3
4
5
6
7
*23456789012345678901234567890123456789012345678901234567890123456789012
INCLUDE COND=(5,4,CH,E*
Q,CABCD)
SORT FIELDS=(9,3,*
ZD,A)
OUTREC FIELDS=(1,80,CBLANK WITHIN A*
LITERAL)

OUTREC FIELDS=(1,80,CBLANK WITHIN A LITERAL)
86

Attention: You should only start with a blank in column 16 of line 2 if you
need a blank as the first character of the continued operand, as shown in the
previous OUTREC statement. A blank in column 16 of line 2 will be included in
the operand and will result in invalid syntax if incorrectly placed. For example:
*
1
2
3
4
5
6
7
*23456789012345678901234567890123456789012345678901234567890123456789012
SORT FIELDS=(5,4,Z*
D,A)
SUM FIELDS=(5,4,Z*
D)

SUM FIELDS=(5,4,Z D)
With the 'D' in column 16 of line 2, we get 'ZD' in the SORT statement. But with
the 'D' in column 17 of line 2, we get 'Z D' in the SUM statement instead of 'ZD',
resulting in a syntax error.
v Explicit continuation in 2-15: If line 1 breaks at column 71 with a nonblank in
column 72, and columns 2-15 of line 2 are nonblank, DFSORT continues on line
2 with the first nonblank character it finds in columns 2-15. For example:
*
1
2
3
4
5
6
7
*23456789012345678901234567890123456789012345678901234567890123456789012
INCLUDE COND=(5,4,CH,EQ,CAB*
CD)
SORT FIELDS=(9,3,*
ZD,A)
OUTREC FIELDS=(5,4,2X*
,9,3,ZD,M26,80:X)

OUTREC FIELDS=(5,4,2X,9,3,ZD,M26,80:X)
v Remark continuation in 2-71: If a statement ends on line 1 with a blank before

column 72 and a nonblank in column 72, DFSORT treats the first nonblank
character it finds in columns 2-71 of line 2 as the start of a remark. For example:
*
1
2
3
4
5
6
7
*23456789012345678901234567890123456789012345678901234567890123456789012
THIS IS A
*
CONTINUED REMARK
Tip: A simpler way to do the same thing (without continuation) is to use a

comment statement for line 2. For example:
*
1
2
3
4
5
6
7
*23456789012345678901234567890123456789012345678901234567890123456789012
THIS IS A
*
CONTINUED REMARK
Inserting comment statements

v Specify comment statements by coding an asterisk (*) in column 1. A comment
statement is printed along with other DFSORT program control statements but is
not otherwise processed.
v A statement with blanks in columns 1 through 71 is treated as a comment
statement.
v Comment statements are allowed only in the DFSPARM, SYSIN, and
SORTCNTL data sets.
87
Coding restrictions
The following rules apply to control statement preparation:
v Operation definers and operands must be in uppercase EBCDIC.
v Column 1 of each control statement can be used only for a label or for a
comment statement that begins with an asterisk in column 1.
v If present, a label must begin in column 1. Labels are allowed only in the SYSIN
and SORTCNTL data sets.
v The entire operation definer must be contained on the first line of a control
statement.
v The first operand must begin on the first line of a control statement. The last
operand in a statement must be followed by at least one blank.
v Blanks are not allowed in operands. Anything following a blank is considered
part of the remark field.
v Remarks are allowed only in the DFSPARM, SYSIN, and SORTCNTL data sets.
v Commas, semicolons, and blanks can be used only as delimiters. They can be
used in values only if the values are constants.
v Each type of program control statement can appear only once within a single
source (for example, the SYSIN data set).
EFS restrictions when an EFS program is in effect

In addition to the previous items, the following restrictions apply to control
statement preparation for an EFS program.
v Non-DFSORT operation definers can be up to 8 bytes long.
v An operation definer with no operands is allowed only if:
It is supplied through SYSIN, SORTCNTL, or DFSPARM.
It is the only operation definer on a line; column 72 must contain a blank.
Using control statements from other IBM programs

The INPFIL control statement, which is used by other IBM sort programs, is
accepted but not processed. However, control statement errors can result from
continuation of an INPFIL statement. The information contained in the INPFIL
statement for other IBM sort programs is supplied to DFSORT with DD statements.
Because DFSORT uses the OPTION control statement, OPTION control statements
in any job streams from other IBM sort programs cause DFSORT to terminate
unless the parameters from the other program conform to the DFSORT OPTION
control statement parameters.
ALTSEQ control statement

,
ALTSEQ CODE=( fftt
The ALTSEQ control statement can be used to change the alternate translation table
(ALTSEQ table). Any modifications you specify are applied to the standard
EBCDIC translation table. The modified ALTSEQ table overrides the installation
default ALTSEQ table (the shipped default is the EBCDIC translation table).
The ALTSEQ table can be used in two ways as follows:
88
ALTSEQ Control Statement

v To apply an alternate collating sequence for SORT, MERGE, INCLUDE, or OMIT
fields, or for INREC, OUTREC, or OUTFIL fields in logical expressions with
format AQ (or format CH with CHALT in effect). In this case, the ALTSEQ table
is used to change only the order in which data is collated, not the data itself. If
you specify AQ (or CH with CHALT) without specifying an ALTSEQ control
statement, DFSORT uses the installation default ALTSEQ table.
For example, if you want to specify that the character $ (X'5B') is to collate at
position X'EA', after uppercase Z (X'E9'), you should specify:
ALTSEQ CODE=(5BEA)
v To convert characters for INREC, OUTREC, or OUTFIL OUTREC fields with

TRAN=ALTSEQ. In this case, the ALTSEQ table is used to change the actual
data. If you specify TRAN=ALTSEQ without specifying an ALTSEQ control
statement, DFSORT uses the installation default ALTSEQ table.
For example, if you want to change the character $ (X'5B') to the character *
(X'5C'), you should specify:
ALTSEQ CODE=(5B5C)
CODE
,
CODE=( fftt
Specifies the original and modified EBCDIC collating positions.

ff
specifies, in hexadecimal, the character whose position is to be changed

in the ALTSEQ table.
tt
specifies, in hexadecimal, the new position the character is to occupy in

the ALTSEQ table.
The order in which the parameters are specified is not important.

Note:
1. If CHALT is in effect, control fields with format CH are collated using the
ALTSEQ table, in addition to those with format AQ.
2. If you use locale processing for SORT, MERGE, INCLUDE, or OMIT fields,
you must not use CHALT. If you need alternate sequence processing for a
particular field, use format AQ.
3. Using ALTSEQ can degrade performance.
Default: Usually the installation option. See Appendix B, Specification/
Altering EBCDIC collating sequenceexamples

Example 1
SORT FIELDS=(18,20,AQ,A)
ALTSEQ CODE=(5BEA)
The character $ (X'5B') is to collate at position X' EA', that is, after uppercase Z
(X'E9').
89
ALTSEQ Control Statement
Example 2
MERGE FIELDS=(25,7,A,1,10,D),FORMAT=CH
OPTION CHALT
ALTSEQ CODE=(F0B0,F1B1,F2B2,F3B3,F4B4,F5B5,F6B6,
F7B7,F8B8,F9B9)
The numerals 0 through 9 are to collate before uppercase letters (but after
lowercase letters).
Example 3
ALTSEQ CODE=(C1F1,C2F2)
The uppercase A (X'C1') is to collate at the same position as the numeral 1 (X'F1')
and the uppercase B (X'C2') is to collate at the same position as the numeral 2
(X'F2').
Note that this ALTSEQ statement does NOT cause collating of A before or after 1,
or of B before or after 2.
Example 4
ALTSEQ
CODE=(81C1,82C2,83C3,84C4,85C5,86C6,87C7,
88C8,89C9,91D1,92D2,93D3,94D4,95D5,96D6,
97D7,98D8,99D9,A2E2,A3E3,A4E4,A5E5,A6E6,
A7E7,A8E8,A9E9)
Each lowercase letter is to collate at the same position as the corresponding

uppercase letter. For example, the lowercase a (X'81') is to collate at the same
position as the uppercase A (X'C1'). This results in case-insensitive collating.
Example 5
OPTION COPY
ALTSEQ CODE=(0040)
OUTREC FIELDS=(1,80,TRAN=ALTSEQ)
Each binary zero (X'00') is changed to a space (X'40').
90
DEBUG Control Statement
DEBUG control statement

,
DEBUG
ABEND
NOABEND
ABSTP
BSAM
CFW
NOCFW
CTRx=n
,
EFSDPAFT=( n
,
EFSDPBFR=( n
EQUCOUNT
ESTAE
NOESTAE
NOASSIST
The DEBUG control statement is not intended for regular use; only ABEND,
NOABEND, and BSAM are of general interest. For a tape work sort or a
Conventional merge, only the ABEND or NOABEND parameters of the DEBUG
statement are used. For more information about problem diagnosis, see z/OS
DFSORT Messages, Codes and Diagnosis Guide.
ABEND or NOABEND
ABEND
NOABEND
Temporarily overrides the ERET installation option, which specifies whether

DFSORT abends or terminates with a return code of 16, if your sort, copy, or
merge is unsuccessful.
ABEND
Specifies that if your sort, copy, or merge is unsuccessful, DFSORT abends
with a user completion code equal to the appropriate message number or
with a user-defined number between 1 and 99, as set during installation
with installation option ABCODE=n.
When DEBUG ABEND is in effect, a user abend code of zero might be
issued when a tape work data set sort or Conventional merge is
unsuccessful.
NOABEND
Specifies that an unsuccessful sort, copy, or merge terminates with a return
code of 16.
ABSTP
91

Prevents loss of needed information in a dump when Blockset terminates. This
option overrides ERET, ABEND, and NOABEND. If the DFSORT application is
unsuccessful, an abend is forced with a completion code equal to the
appropriate message number, or with the user ABEND code set with
installation option ABCODE=MSG or ABCODE=n. The message is not written
if NOESTAE is in effect.
BSAM
BSAM
Temporarily bypasses the EXCP access method for input and output data sets.
BSAM is ignored for VSAM input and output data sets.
Attention: If Blockset is not selected and BSAM processing is used with
concatenated SORTIN input, and both null and non-null data sets are specified,
all null data sets must precede all non-null data sets; otherwise, the results are
unpredictable.
CFW or NOCFW
CFW
NOCFW
Temporarily overrides the CFW installation option, which specifies whether

DFSORT can use cache fast write when processing SORTWKdd data sets that
reside on devices connected to cached 3990 control units.
CFW
Specifies that DFSORT can use cache fast write when processing
SORTWKdd data sets.
NOCFW
Specifies that DFSORT cannot use cache fast write.
Attention: The NOCFW option can degrade performance.
CTRx
CTRx=n
Keeps a count of the input and output records, and abends with code 0C1
when the count reaches n. The numbers that can be assigned to x are:
2
92
Counts the input records being moved from the input buffer (not used
for a copy).

3
Counts the output records being moved to the output buffer (not used
for a copy or merge).
Counts the input records inserted by E15 (not used for Blockset).
Counts the output records deleted by E35 (not used for Blockset).

EFSDPAFT
,
EFSDPAFT=( n
Initiates a SNAP dump after a Major Call to an EFS program. Any combination
of the numbers can be specified.
The numbers have the following meanings:
2
Takes the SNAP dump after Major Call 2 to the EFS program.

EFSDPBFR
,
EFSDPBFR=( n
Initiates a SNAP dump before a Major Call to an EFS program. Any

combination of the numbers can be specified.
The numbers have the following meanings:
2
Takes the SNAP dump before Major Call 2 to the EFS program.

EQUCOUNT
EQUCOUNT
Determines the number of records having equal keys (that is, duplicate keys)
93

which have been sorted by the Blockset technique (printed in message
ICE184I). For variable-length records, EQUCOUNT can only be used with
either Hiperspace (when Hipersorting is used) or work data sets.
Note:
1. Using EQUCOUNT can degrade performance.
2. ICETOOL's UNIQUE and OCCUR operators provide unique and
non-unique key reporting capabilities that may be more useful for your
application than EQUCOUNT.
3. If VLSHRT is in effect, EQUCOUNT will not be used.
ESTAE or NOESTAE
ESTAE
NOESTAE
Temporarily overrides the ESTAE installation option, which determines

whether DFSORT should delete its ESTAE recovery routine early or use it for
the entire run.
DFSORT normally establishes an ESTAE recovery routine at the beginning of a
run. If an abend occurs and the ESTAE option is in effect, the system passes
control to the recovery routine. The routine terminates the run after attempting
to:
v Print additional abend information
v Continue a sort, merge, or copy application after successful SORTOUT
output
v Call the EFS program at Major Calls 4 and 5 for cleanup and housekeeping
v Write an SMF record
v Call the ICETEXIT termination exit.
If an abend occurs and the ESTAE option is not in effect, these functions might
not be performed.
ESTAE
specifies that DFSORT can use its ESTAE recovery routine for the entire
run.
NOESTAE
specifies that DFSORT is to delete its ESTAE recovery routine at a point
early in its processing. If DFSORT terminates or abends before this point is
reached, it will not delete its ESTAE recovery routine; that is, NOESTAE
will not be in effect.
Note: See Appendix E, DFSORT abend processing, on page 907 for more
information on the DFSORT ESTAE recovery routine.
NOASSIST
94

NOASSIST
DFSORT uses Sorting Instructions when possible. If you do not want to use
these instructions, you can temporarily bypass them by specifying this
parameter.
Specifying diagnostic optionsexamples

Example 1
SORT FIELDS=(1,4,CH,A)
DEBUG EQUCOUNT
If the input records contain the following keys:

v KEYA, KEYA, KEYB, KEYB, KEYC, KEYD, KEYD, KEYE
the following message will be issued:
v ICE184I THE NUMBER OF RECORDS SORTED WITH EQUAL KEYS IS 3
The three equal keys are KEYA, KEYB, and KEYD.
Note: ICETOOL's UNIQUE and OCCUR operators provide full equal key
reporting capabilities and should be used instead of EQUCOUNT.
Example 2
SORT FIELDS=(12,2,BI,D)
DEBUG BSAM,ABEND
Directs DFSORT to use the BSAM access method for the SORTIN and SORTOUT
data sets and to abend if the sort application is unsuccessful.
END control statement

END
The END control statement allows DFSORT to discontinue reading SYSIN,

DFSPARM, or SORTCNTL before end of file (EOF).
When you link-edit user exit routines dynamically, the END statement marks the
end of the DFSORT control statements and the beginning of exit routine object
decks in SYSIN.
Discontinue reading control statementsexamples

Example 1
//SYSIN DD *
SORT FIELDS=(1,6,A,28,5,D),FORMAT=CH
RECORD TYPE=V,LENGTH=(200,,,,80)
END
OPTION DYNALLOC
95
END Control Statement

Because the OPTION statement appears after the END statement, it is not read.
Example 2
//SYSIN DD *
MODS E15=(E15,1024,SYSIN,T)
END
object deck for E15 user exit here
The END statement precedes the E15 user exit routine object deck in SYSIN.
INCLUDE control statement

INCLUDE COND=
(logical expression)

FORMAT=f
ALL
NONE
Use an INCLUDE statement if you want only certain records to appear in the
output data set. The INCLUDE statement selects the records you want to include.
You can specify either an INCLUDE statement or an OMIT statement in the same
DFSORT run, but not both.
The way in which DFSORT processes short INCLUDE/OMIT compare fields
depends on the settings for VLSCMP/NOVLSCMP and VLSHRT/NOVLSHRT. A
short field is one where the variable-length record is too short to contain the entire
field, that is, the field extends beyond the record. For details about including or
omitting short records, see the discussion of the VLSCMP and NOVLSCMP options
A logical expression is one or more relational conditions logically combined, based
on fields in the input record, and can be represented at a high level as follows:
relational condition1

.
,
AND
OR
,relational condition2
If the logical expression is true for a given record, the record is included in the
output data set.
Five types of relational conditions can be used as follows:
1. Comparisons:
Compare two compare fields or a compare field and a decimal, hexadecimal,
character, or current, future, or past date constant.
For example, you can compare the first 6 bytes of each record with its last 6
bytes, and include only those records in which those fields are identical. Or you
can compare a date field with today's date, yesterday's date, or tomorrow's
date, and include records accordingly.
See Comparisons on page 99 for information about comparisons.
2. Substring Comparison Tests:
96
INCLUDE Control Statement

Search for a constant within a field value or a field value within a constant.
For example, you can search the value in a 6-byte field for the character
constant C'OK', and include only those records for which C'OK' is found
somewhere in the field. Or you can search the character constant C'J69,L92,J82'
for the value in a 3-byte field, and include only those records for which C'J69',
C'L92', or C'J82' appears in the field.
See Substring comparison tests on page 110 for information about substring
comparison tests.
3. Bit Logic Tests:
Test the state (on or off) of selected bits in a binary field using a bit or
hexadecimal mask or a bit constant.
For example, you can include only those records which have bits 0 and 2 on in
a 1-byte field. Or you can include only those records which have bits 3 and 12
on and bits 6 and 8 off in a 2-byte field.
See Bit logic tests on page 112 for information about bit logic tests.
4. Date Comparisons:
Compare a two-digit year date field to a two-digit year date constant, a current,
future or past two-digit year date, or to another two-digit year date field, using
the century window in effect.
For example, you can include only those records for which a Z'yymm' date
field is between January 1996 and March 2005. Or you can include only those
records for which a P'dddyy' field is less than another P'dddyy' field. Or you
can include only those records for which a C'yyddd' field is between today's
date and 5 days earlier than today's date.
See Date comparisons on page 117 for information about date comparisons.
5. Numeric Tests:
Test a field for numerics or non-numerics in character, zoned decimal or packed
decimal format.
For example, you can include only those records in which a 5-byte field
contains only '0'-'9' characters (that is, numerics). Or you can include only those
records in which a 9-byte field contains invalid ZD numeric data (that is,
non-numerics). Or you can include only those records in which a 12-byte field
contains valid PD numeric data (that is, numerics).
See Numeric tests on page 120 for information about numeric tests.
6. Alphanumeric Tests:
Test a field for alphanumerics or non-alphanumerics in character format.
Various combinations of uppercase characters (A-Z), lowercase characters (a-z)
and numeric characters (0-9) can be used.
For example, you can include only those records in which a 10-byte field
contains only 'A'-'Z' characters (that is, uppercase characters) or '0'-'9' characters
(that is, numeric characters). Or you can include only those records in which a
20-byte field contains characters other than 'a'-'z' (that is, lowercase characters).
See Alphanumeric tests on page 122 for information about aphanumeric tests.
By nesting relational conditions within parentheses, you can create logical
expressions of higher complexity.
Although comparisons, substring comparison tests, bit logic tests, date
comparisons, numeric tests, and alphanumeric tests are explained separately later
in this section for clarity, they can be combined to form logical expressions.
97

The INCLUDE control statement differs from the INCLUDE parameter of the
OUTFIL statement in the following ways:
v The INCLUDE statement applies to all input records; the INCLUDE parameter
applies only to the OUTFIL input records for its OUTFIL group.
v FORMAT=f can be specified with the INCLUDE statement but not with the
INCLUDE parameter. Thus, you can use FORMAT=f and p,m or p,m,f fields
with the INCLUDE statement, but you must only use p,m,f fields with the
INCLUDE parameter. For example:
INCLUDE FORMAT=BI,
COND=(5,4,LT,11,4,OR,21,4,EQ,31,4,OR,
61,20,SS,EQ,CFLY)
OUTFIL INCLUDE=(5,4,BI,LT,11,4,BI,OR,21,4,BI,EQ,31,4,BI,OR,
61,20,SS,EQ,CFLY)
v D2 format can be specified with the INCLUDE statement but not with the
INCLUDE parameter.
See OUTFIL control statements on page 223 for more details on the OUTFIL
INCLUDE parameter.
COND
COND=
ALL
NONE
logical expression
specifies one or more relational conditions logically combined, based
on fields in the input record. If the logical expression is true for a
given record, the record is included in the output data sets.
ALL
specifies that all of the input records are to be included in the output
data sets.
NONE
specifies that none of the input records are to be included in the output
data sets.
Default: ALL. See Appendix B, Specification/override of DFSORT options, on
page 863 for full override details.
FORMAT
FORMAT=f
FORMAT=f can be used to specify a particular format for one or more compare
fields. f from FORMAT=f is used for p,m fields. f from FORMAT=f is ignored
for p,m,f fields. For example, the following are all equivalent:
INCLUDE COND=(5,5,ZD,EQ,12,3,PD,OR,21,3,PD,NE,35,5,ZD)
INCLUDE FORMAT=ZD,COND=(5,5,EQ,12,3,PD,OR,21,3,PD,NE,35,5)
INCLUDE COND=(5,5,ZD,EQ,12,3,OR,21,3,NE,35,5,ZD),FORMAT=PD
The permissible field formats for comparisons are shown in Table 9 on page
100. SS (substring) is the only permissible field format for substring
comparison tests. BI (unsigned binary) is the only permissible field format for
bit logic tests. The Y2x formats are the only permissible field formats for date
98

comparisons. The FS or CSF (floating sign), ZD (zoned decimal), and PD
(packed decimal) formats are the only permissible field formats for numeric
tests. BI (unsigned binary) is the only permissible format for alphanumeric
tests.
Default: None. FORMAT=f must be specified if any field is specified as p,m
rather than p,m,f. See Appendix B, Specification/override of DFSORT
Note: DFSORT issues an informational message and ignores FORMAT=f if all of
the fields are specified as p,m,f.
Relational condition
A relational condition specifies a comparison, substring comparison test, bit logic
test, date comparison, numeric test or alphanumeric test to be performed.
Relational conditions can be logically combined, with AND or OR, to form a
logical expression. If they are combined, the following rules apply:
v AND statements are evaluated before OR statements unless parentheses are used
to change the order of evaluation; expressions inside parentheses are always
evaluated first. (Nesting of parentheses is limited only by the amount of storage
available.)
v The symbols & (AND) and | (OR) can be used instead of the words.
Comparisons
Relational condition format
Two formats for the relational condition can be used:
(p1,m1,BI,
EQ
NE
GT
GE
LT
LE
p2,m2,f2
constant
Or, if the FORMAT=f operand is used:

(p1,m1,
f1,
EQ
NE
GT
GE
LT
LE
p2,m2
,f2
constant
Comparison operators are as follows:

EQ
Equal to
NE
Not equal to
GT
Greater than
GE
Greater than or equal to
LT
Less than
99

LE
Less than or equal to.
Fields:
p1,m1,f1: These variables specify a field in the input record to be compared either
to another field in the input record or to a constant.
v p1 specifies the first byte of the compare field relative to the beginning of the
input record.4 The first data byte of a fixed-length record (FLR) has relative
position 1. The first data byte of a variable-length (VLR) record has relative
position 5 (because the first 4 bytes contain the record descriptor word). All
compare fields must start on a byte boundary, and no compare field can extend
beyond byte 32752.
v m1 specifies the length of the compare field. Acceptable lengths for different
formats are in Table 9.
v f1 specifies the format of the data in the compare field. Permissible formats are
given in Table 9.
You can use p1,m1 rather than p1,m1,f1 if you use FORMAT=f to supply the
format for the field.
Table 9. Compare Field Formats and Lengths
Format Code
Length
Description
CH
1 to 256 bytes
Character
AQ
1 to 256 bytes
Character with alternate

collating sequence
ZD
1 to 256 bytes
Signed zoned decimal
PD
1 to 255 bytes
Signed packed decimal
PD0
2 to 8 bytes
Packed decimal with sign

and first digit ignored
FI
1 to 256 bytes
Signed fixed-point
BI
1 to 256 bytes
Unsigned binary
AC
1 to 256 bytes
ASCII character
CSF or FS
1 to 32 bytes
Signed numeric with optional

leading floating sign
UFF
1 to 44 bytes
Unsigned free form numeric
SFF
1 to 44 bytes
Signed free form numeric
CSL or LS
2 to 256 bytes
Signed numeric with leading

separate sign
CST or TS
2 to 256 bytes
Signed numeric with trailing

separate sign
CLO or OL
1 to 256 bytes

overpunch sign
CTO or OT
1 to 256 bytes

overpunch sign
ASL
2 to 256 bytes
Signed ASCII numeric with

leading separate sign
AST
2 to 256 bytes

trailing separate sign
4. If your E15 user exit routine formats the record, p1 must refer to the record as reformatted by the exit.
100

Table 9. Compare Field Formats and Lengths (continued)
Format Code
Length
Description
D2
1 to 256 bytes
User-defined data type

(requires an EFS program)
Note: See Appendix C, Data format descriptions, on page 891 for detailed format
descriptions.
p2,m2,f2: These variables specify another field in the input record with which the
p1,m1,f1 field will be compared. Permissible comparisons between compare fields
with different formats are shown in Table 10 and Table 11.
AC, ASL, and AST formats sequence EBCDIC data using the ASCII collating
sequence.
You can use p2,m2 rather than p2,m2,f2 if you use FORMAT=f to supply the
Table 10. Permissable Field-to-Field Comparisons for INCLUDE/OMIT (Group 1)
Field Format
BI
CH
ZD
PD
BI
CH
ZD
PD
CSF or
FS
UFF
SFF
CSL or
LS
CST or
TS
CSF or FS
UFF
SFF
CSL or LS
CST or TS
FI
FI
Table 11. Permissable Field-to-Field Comparisons for INCLUDE/OMIT (Group 2)

Field Format
PD0
PD0
ASL
AST
CLO or
OL
CTO or
OT
ASL
AST
CLO or OL
CTO or OT
AC
AC
AQ
D2
AQ
D2
X
X
Note: D2 field formats are user-defined.
5. If CHALT is in effect, CH is treated as AQ.

101

Constants: A constant can be a decimal number (n, +n, -n), character string
(C'xx...x'), or hexadecimal string (X'yy...yy').
The current date can be used as a decimal number (DATE1P, &DATE1P, DATE2P,
&DATE2P, DATE3P, &DATE3P) or character string (DATE1, &DATE1, DATE1(c),
&DATE1(c), DATE2, &DATE2, DATE2(c), &DATE2(c), DATE3, &DATE3, DATE3(c),
&DATE3(c), DATE4, &DATE4, DATE5, &DATE5).
A future date can be used as a decimal number (DATE1P+d, &DATE1P+d,
DATE2P+m, &DATE2P+m, DATE3P+d, &DATE3P+d) or character string
(DATE1+d, &DATE1+d, DATE1(c)+d, &DATE1(c)+d, DATE2+m, &DATE2+m,
DATE2(c)+m, &DATE2(c)+m, DATE3+d, &DATE3+d, DATE3(c)+d, &DATE3(c)+d).
A past date can be used as a decimal number (DATE1P-d, &DATE1P-d,
DATE2P-m, &DATE2P-m, DATE3P-d, &DATE3P-d) or character string (DATE1-d,
&DATE1-d, DATE1(c)-d, &DATE1(c)-d, DATE2-m, &DATE2-m, DATE2(c)-m,
&DATE2(c)-m, DATE3-d, &DATE3-d, DATE3(c)-d, &DATE3(c)-d).
The different constants are explained in detail in the following table. Permissible
comparisons between compare fields and constants are shown in Table 12.
Table 12. Permissible Field-to-Constant Comparisons for INCLUDE/OMIT
Self-Defining Term
Decimal Number
BI
CH
ZD
PD
Character String
Hexadecimal String
PD0
FI
AC
ASL
AST
CSF or FS
UFF
SFF
CSL or LS
CST or TS
CLO or OL
CTO or OT
AQ
D2
Note: D2 field formats are user-defined.
Decimal number format: The format for coding a decimal constant is:
102
[]n
When an FI field is compared with a decimal constant, n or +n cannot be larger

than +9223372036854775807 and -n cannot be smaller than -9223372036854775808.
When a BI field is compared with a decimal constant, n or +n cannot be larger
than +18446744073709551615 nor smaller than +0. A BI field cannot be compared to
a negative number (-n). A BI field cannot be compared to -0 even if NOSZERO is
in effect.
Examples of valid and invalid decimal constants are:
Table 13. Valid and Invalid Decimal Constants
Valid
Invalid
Explanation
15
++15
Too many sign characters
+15
15+
Sign in wrong place
-15
1.5
Contains invalid character
18000000
1,500
Contains invalid character
Current date as decimal number: DATE1P, &DATE1P, DATE2P, &DATE2P,

DATE3P, or &DATE3P can be used to generate a decimal number for the current
date of the run.
Table 14 shows the form of the decimal number constant generated for each
current date operand along with an example of the actual decimal number
generated when the date of the run is June 21, 2005. yyyy represents the year, mm
represents the month (01-12), dd represents the day (01-31) and ddd represents the
day of the year (001-366).
Table 14. Decimal Numbers for Current Date
Format of Constant
Example of Constant
DATE1P
+yyyymmdd
+20050621
DATE2P
+yyyymm
+200506
DATE3P
+yyyyddd
+2005172
Note: You can precede each of the operands in the table with an & with identical
results.
Future date as decimal number: DATE1P+d, &DATE1P+d, DATE2P+m,
&DATE2P+m, DATE3P+d, or &DATE3P+d can be used to generate a decimal
number for a future date relative to the current date of the run. d is days in the
future and m is months in the future. d and m can be 0 to 9999.
Table 15 on page 104 shows the form of the decimal number constant generated for
each future date operand along with an example of the actual decimal number
represents the month (01-12) , dd represents the day (01-31) and ddd represents the
day of the year (001-366).
103

Table 15. Decimal Numbers for Future Dates
Format of Constant
Example of Operand Example of Constant
DATE1P+d
+yyyymmdd
DATE1P+11
+20050702
DATE2P+m
+yyyymm
DATE2P+2
+200508
DATE3P+d
+yyyyddd
DATE3P+200
+2006007
results.
Past date as decimal number: DATE1P-d, &DATE1P-d, DATE2P-m, &DATE2P-m,
DATE3P-d, or &DATE3P-d can be used to generate a decimal number for a past
date relative to the current date of the run. d is days in the past and m is months
in the past. d and m can be 0 to 9999.
Table 16 shows the form of the decimal number constant generated for each past
date operand along with an example of the actual decimal number generated when
the date of the run is June 21, 2005. yyyy represents the year, mm represents the
month (01-12), dd represents the day (01-31) and ddd represents the day of the
year (001-366).
Table 16. Decimal Numbers for Past Dates
Format of Constant
DATE1P-d
+yyyymmdd
DATE1P-30
+20050522
DATE2P-m
+yyyymm
DATE2P-12
+200406
DATE3P-d
+yyyyddd
DATE3P-172
+2004366
results.
Character string format: The format for coding a character string constant is:
C'xx...x'
The value x may be any EBCDIC character (the EBCDIC character string is
translated appropriately for comparison to an AC or AQ field). You can specify up
to 256 characters.
If you want to include a single apostrophe in the character string, you must specify
it as two single apostrophes. Thus:
Required:
ONEILL
Specify: CONEILL
Examples of valid and invalid character string constants are shown in Table 17:
Table 17. Valid and Invalid Character String Constants
104
Valid
Invalid
Explanation
C'JDCO'
C'''''
Apostrophes not paired
C'$@#'
'ABCDEF'
C identifier missing
C'+0.193'
C'ABCDEF
Apostrophe missing
C'Frank''s'
C'Frank's'
Two single apostrophes needed for one
Double-byte data may be used in a character string for INCLUDE/OMIT

comparisons. The start of double-byte data is delimited by the shift-out (SO)
control character (X'0E'), and the end by the shift-in (SI) control character (X'0F').
SO and SI control characters are part of the character string and must be paired
with zero or an even number of intervening bytes. Nested shift codes are not
allowed. All characters between SO and SI must be valid double-byte characters.
No single-byte meaning is drawn from the double-byte data.
Examples of valid and invalid character string constants containing double-byte
characters are shown in Table 18 using:
v < to represent SO
v > to represent SI
v Dn to represent a double-byte character
Table 18. Valid and Invalid Strings with Double-Byte Data
Invalid
Explanation
C'Q<D1D2>T'
C'Q<R>S'
Single-byte data within SO/SI
C'<D1D2D3>'
C'D1D2D3'
Missing SO/SI; treated as single-byte

data
C'Q<D1>R<D2>'
C'Q<D1<D2>>'
Nested SO/SI
Tip: X'0E', X'0F', and X'7D' are treated as the special characters shift-out, shift-in,
and single apostrophe in a character string. If you don't want to treat one or more
of these characters as special in a particular value, use a hexadecimal string instead
of a character string. For example, if you want to treat the binary value 000E0E7D
as its decimal equivalent of 921213, use X'000E0E7D'; 0E will not be treated as
shift-out and 7D will not be treated as a single apostrophe.
Current date as character string: DATE1, &DATE1, DATE1(c), &DATE1(c),
DATE2, &DATE2, DATE2(c), &DATE2(c), DATE3, &DATE3, DATE3(c), &DATE3(c),
DATE4, &DATE4, DATE5 or &DATE5 can be used to generate a character string
for the current date of the run.
Table 19 shows the form of the character string constant generated for each current
date operand along with an example of the actual character string generated when
the date of the run is June 21, 2005 at 04:42:45 PM, using (/) for (c) where relevant.
yyyy represents the year, mm (for date) represents the month (01-12), dd represents
the day (01-31), ddd represents the day of the year (001-366), hh represents the
hour (00-23), mm (for time) represents the minutes (00-59), ss represents the
seconds (00-59), nnnnnn represents the microseconds (000000-999999) and c can be
any character except a blank.
Table 19. Character Strings for Current Date
Format of Constant
Example of Constant
DATE1
C'yyyymmdd'
C'20050621'
DATE1(c)
C'yyyycmmcdd'
C'2005/06/21'
DATE2
C'yyyymm'
C'200506'
DATE2(c)
C'yyyycmm'
C'2005/06'
DATE3
C'yyyyddd'
C'2005172'
105

Table 19. Character Strings for Current Date (continued)
Format of Constant
Example of Constant
DATE3(c)
C'yyyycddd'
C'2005/172'
DATE4
C'yyyy-mm-dd-hh.mm.ss'
C'2005-06-21-16.52.45'
DATE4
C'2005-06-21-16.52.45'
DATE5
C'yyyy-mm-dd-hh.mm.ss.nnnnnn'
C'2005-06-21-16.52.45.582013'
results.
Tip: When a field is shorter than the character string it's compared to, DFSORT
truncates the string on the right. You can take advantage of this to compare a field
to only part of the DATE4 timestamp when appropriate. For example:
INCLUDE COND=(1,13,CH,GT,DATE4)
would compare the field in positions 1-13 to the truncated DATE4 constant
C'yyyy-mm-dd-hh'.
Future date as character string: DATE1+d, &DATE1+d, DATE1(c)+d,
&DATE1(c)+d, DATE2+m, &DATE2+m, DATE2(c)+m, &DATE2(c)+m, DATE3+d,
&DATE3+d, DATE3(c)+d or &DATE3(c)+d can be used to generate a character
string for a future date relative to the current date of the run. d is days in the
Table 20 shows the form of the character string constant generated for each future
date operand along with an example of the actual character string generated when
the date of the run is June 21, 2005. yyyy represents the year, mm represents the
month (01-12), dd represents the day (01-31), ddd represents the day of the year
(001-366), and c can be any character except a blank.
Table 20. Character Strings for Future Dates
Format of Constant
DATE1+d
C'yyyymmdd'
DATE1+11
C'20050702'
DATE1(c)+d
C'yyyycmmcdd'
DATE1(/)+90
C'2005/09/19'
DATE2+m
C'yyyymm'
DATE2+2
C'200508'
DATE2(c)+m
C'yyyycmm'
DATE2(.)+25
C'2007.07'
DATE3+d
C'yyyyddd'
DATE3+200
C'2006007'
DATE3(c)+d
C'yyyycddd'
DATE3(-)+1
C'2005-171'
results.
Past date as character string: DATE1-d, &DATE1-d, DATE1(c)-d, &DATE1(c)-d,
DATE2-m, &DATE2-m, DATE2(c)-m, &DATE2(c)-m, DATE3-d, &DATE3-d,
DATE3(c)-d or &DATE3(c)-d can be used to generate a character string for a past
date relative to the current date of the run. d is days in the past and m is months
in the past. d and m can be 0 to 9999.
Table 21 on page 107 shows the form of the character string constant generated for
each past date operand along with an example of the actual character string
106

represents the month (01-12), dd represents the day (01-31), ddd represents the day
of the year (001-366), and c can be any character except a blank.
Table 21. Character Strings for Past Dates
Format of Constant
DATE1-d
C'yyyymmdd'
DATE1-1
C'20050620'
DATE1(c)-d
C'yyyycmmcdd'
DATE1(-)-60
C'2005-04-22'
DATE2-m
C'yyyymm'
DATE2-6
C'200412'
DATE2(c)-m
C'yyyycmm'
DATE2(/)-1
C'2005/05'
DATE3-d
C'yyyyddd'
DATE3-300
C'2004238'
DATE3(c)-d
C'yyyycddd'
DATE3(.)-21
C'2005.151'
results.
Hexadecimal string format: The format for coding a hexadecimal string constant
is:
X'yy...yy'
The value yy represents any pair of hexadecimal digits. You can specify up to 256
pairs of hexadecimal digits.
Because the first digit and sign are ignored in a PD0 field, you should not include
the first digit or sign in a hexadecimal constant to be compared to a PD0 field. For
example, 3-byte PD0 values like X'01234C' and X'01234D' would be equal to a
hexadecimal constant of X'1234'.
Examples of valid and invalid hexadecimal constants are shown in the following
table.
Table 22. Valid and Invalid Hexadecimal Constants
Valid
Invalid
Explanation
X'ABCD'
X'ABGD'
Invalid hexadecimal digit
X'BF3C'
X'BF3'
Incomplete pair of digits
X'AF050505'
'AF050505'
Missing X identifier
X'BF3C'
'BF3C'X
X identifier in wrong place

In a field-to-field comparison, the shorter compare field is padded appropriately. In
a field-to-constant comparison, the constant is padded or truncated to the length of
the compare field.
Character and hexadecimal strings are truncated and padded on the right.
The padding characters are:
v X'40' For a character string
v X'00' For a hexadecimal string.
107

Decimal constants are padded and truncated on the left. Padding is done with
zeros in the proper format.
Cultural environment considerations

environment. The cultural environment is established by selecting the active locale.
The active locale's collating rules affect INCLUDE and OMIT processing as follows:
v DFSORT includes or omits records for output according to the collating rules
defined in the active locale. This provides inclusion or omission for single- or
multi-byte character data, based on defined collating rules which retain the
cultural and local characteristics of a language.
If locale processing is to be used, the active locale will only be used to process
character (CH) compare fields and character and hexadecimal constants compared
to character (CH) compare fields.
For more information on locale processing, see Cultural environment
considerations on page 7 or LOCALE in OPTION control statement on page
173.
Including records in the output data setcomparison

examples
Example 1
INCLUDE
COND=(5,8,GT,13,8,|,105,4,LE,1000),FORMAT=CSF
This example illustrates how to only include records in which:

v The floating sign number in bytes 5 through 12 is greater than the floating sign
number in bytes 13 through 20
OR
v The floating sign number in bytes 105 through 108 is less than or equal to 1000.
Note that all three compare fields have the same format.
Example 2
INCLUDE
COND=(1,10,CH,EQ,CSTOCKHOLM,
AND,21,8,ZD,GT,+50000,
OR,31,4,CH,NE,CHERR)

v The first 10 bytes contain STOCKHOLM (this nine-character string was padded
on the right with a blank) AND the zoned-decimal number in bytes 21 through
28 is greater than 50 000
OR
v Bytes 31 through 34 do not contain HERR.
Note that the AND is evaluated before the OR. ( Omitting records from the
output data setexample on page 171 illustrates how parentheses can be used to
change the order of evaluation.) Also note that ending a line with a comma or
semicolon followed by a blank indicates that the parameters continue on the next
line, starting in any position from columns 2 through 71.
108
Example 3
INCLUDE FORMAT=CH,
COND=((5,1,EQ,8,1),&,
((20,1,EQ,CA,&,30,1,FI,GT,10),|,
(20,1,EQ,CB,&,30,1,FI,LT,100),|,
(20,1,NE,CA,&,20,1,NE,CB)))

v Byte 5 equals byte 8
AND
v At least one of the following is true:
Byte 20 equals 'A' and byte 30 is greater than 10
Byte 20 equals 'B' and byte 30 is less than 100
Byte 20 is not equal to 'A' or 'B'.
Note that p,m,FI is used for the FI fields, and p,m with FORMAT=CH is used for
all of the CH fields. With FORMAT=f, you can mix p,m and p,m,f fields when
that's convenient such as when all or most of the fields have the same format
(although you can always code p,m,f for all fields and not use FORMAT=f, if you
prefer).
Example 4
INCLUDE COND=(7,2,CH,EQ,CT1,OR,
(1,2,BI,GE,X001A,AND,20,2,CH,EQ,25,2,CH))
This example shows the effects of VLSCMP/NOVLSCMP and

VLSHRT/NOVLSHRT on INCLUDE processing when short records are present.
Consider the records shown in Figure 6 on page 110:
v If VLSCMP is in effect, the first record is included because bytes 7-8 are equal to
C'T1', even though the comparison of bytes 20-21 to 25-26 involves short fields.
The second record is included or omitted based on the comparison of bytes
20-21 to bytes 25-26.
v If NOVLSCMP and VLSHRT are in effect, the first record is omitted because the
comparison of bytes 20-21 to 25-26 involves short fields. The second record is
included or omitted based on the comparison of bytes 20-21 to bytes 25-26.
v If NOVLSCMP and NOVLSHRT are in effect, the first record causes message
ICE015A or ICE218A to be issued because the comparison of bytes 20-21 to bytes
25-26 involves short fields.
109
RDW
compare
field A
compare
field B
T1
10
compare
field C
RDW
compare
field D
T2
7
20
25
Figure 6. Sample Records
Example 5
INCLUDE COND=(21,8,ZD,GT,DATE1P)
This example illustrates how to include records in which a zoned-decimal date of

the form Z'yyyymmdd' in bytes 21-28 is greater than today's date. DATE1P
generates a decimal number for the current date in the form +yyyymmdd.
Example 6
INCLUDE COND=(15,7,CH,GE,DATE3-7,AND,15,7,CH,LE,DATE3+7)
This example illustrates how to include records in which a character date of the
form C'yyyyddd' in bytes 15-21 is between 7 days in the past and 7 days in the
future, relative to the current date. DATE3-7 generates a character constant in the
form C'yyyyddd' where yyyyddd is the current date minus 7 days. DATE3+7
generates a character constant in the form C'yyyyddd' where yyyyddd is the
current date plus 7 days.
Example 7
INCLUDE COND=(21,10,CH,GE,DATE1(-)-365)
This example illustrates how to include records in which a character date of the
form C'yyyy-mm-dd' in bytes 21-30 is within 365 days of the current date.
DATE1(-)-365 generates a character constant in the form C'yyyy-mm-dd' where
yyyymmdd is the current date minus 365 days.
Substring comparison tests

Two types of substring comparison tests are offered, as follows:
1. Find a constant within a field value. For example, you can search the value in a
6-byte field for the character constant C'OK'. If the field value is, for example,
C'**OK**' or C'****OK', the relational condition is true; if the field value is
C'**ERR*', the relational condition is false.
2. Find a field value within a constant. For example, you can search the character
constant C'J69,L92,J82' for the value in a 3-byte field. If the field value is C'J69',
C'L92', or C'J82', the relational condition is true; if the field value is C'X24', the
relational condition is false. Note that the comma is used within the constant to
separate the valid 3-character values; any character that will not appear in the
field value can be used as a separator in the constant.
110

(p1,m1,SS,
EQ
NE
, constant )
Or, if the FORMAT=SS operand is used:

(p1,m1,
SS,
EQ
NE
, constant )
Restriction: FORMAT=SS can precede COND but cannot follow it.

Substring comparison operators are as follows:
EQ
Equal to
NE
Not equal to
Fields:
p1,m1: These variables specify the character field in the input record for the
substring test.
v p1 specifies the first byte of the character input field for the substring test,
relative to the beginning of the input record.6 The first data byte of a
fixed-length record (FLR) has relative position 1. The first data byte of a
variable-length (VLR) record has relative position 5 (because the first 4 bytes
contain the record descriptor word). All fields to be tested must start on a byte
boundary and must not extend beyond byte 32752.
v m1 specifies the length of the field to be tested. The length can be 1 to 32752
bytes.
Constant: The constant can be a character string or a hexadecimal string. See
Character string format on page 104 and Hexadecimal string format on page
107 for details.
If m1 is greater than the length of the constant, the field value will be searched for
the constant and the condition will be true if a match is found when the EQ
comparison operator is specified or if a match is not found when the NE
comparison operator is specified.
If m1 is smaller than the length of the constant, the constant will be searched for
the field value and the condition will be true if a match is found when the EQ
comparison operator is specified or if a match is not found when the NE
comparison operator is specified.
111
Including records in the output data setsubstring

comparison example
Example
INCLUDE FORMAT=SS,COND=(11,6000,EQ,COK,OR,5,3,EQ,CJ69,L92,J82)
This example illustrates how to include only records in which:

v OK is found somewhere within bytes 11 through 6010
OR
v Bytes 5 through 7 contain J69, L92 or J82.
Bit logic tests

Two methods for bit logic testing are offered as follows:
v Bit operator with hexadecimal or bit mask
v Bit comparison tests
While any bit logic test can be specified using either of the two methods, each of
them offers unique advantages not found with the other.
The ability to specify selected bits in a field, by either of the two methods, can
greatly reduce the number of INCLUDE conditions that must be specified to
achieve a given result, because the need to account for unspecified bits is
eliminated.
Method 1: Bit operator tests

This method of bit logic testing allows you to test whether selected bits in a binary
field are all on, all off, in a mixed on-off state, or in selected combinations of these
states. While this method allows you to test many different possible bit
combinations with a single operation, similar to the Test Under Mask (TM)
machine instruction, it is less suited to determine if a field contains exactly one
particular combination of on and off bits than Method 2 described later in this
section.

(p1,m1,BI,
ALL
SOME
NONE
NOTALL
NOTSOME
NOTNONE
BO
BM
BZ
BNO
BNM
BNZ
, mask )
Or, if the FORMAT=BI operand is used:
112

(p1,m1,
BI,
ALL
SOME
NONE
NOTALL
NOTSOME
NOTNONE
BO
BM
BZ
BNO
BNM
BNZ
, mask )
Bit operators describe the input field to mask relationship to be tested as follows:
ALL or BO
All mask bits are on in the input field
SOME or BM
Some, but not all mask bits are on in the input field
NONE or BZ
No mask bits are on in the input field
NOTALL or BNO
Some or no mask bits are on in the input field
NOTSOME or BNM
All or no mask bits are on in the input field
NOTNONE or BNZ
All or some mask bits are on in the input field
The first set of operators (ALL, SOME, and so on) are intended for those who like
meaningful mnemonics. The second set of operators (BO, BM, and so on) are
intended for those familiar with the conditions associated with the Test Under
Mask (TM) instruction.
Fields
p1,m1: These variables specify the binary field in the input record to be tested
against the mask.
v p1 specifies the first byte of the binary input field to be tested against the mask,
relative to the beginning of the input record.7 The first data byte of a
v m1 specifies the length of the field to be tested. The length can be 1 to 256 bytes.
Mask
A hexadecimal string or bit string that indicates the bits in the field selected for
testing. If a mask bit is on (1), the corresponding bit in the field is tested. If a mask
bit is off (0), the corresponding bit in the field is ignored.
113

Hexadecimal string format: The format for coding a hexadecimal string mask is:
X'yy...yy'
The value yy represents any pair of hexadecimal digits that constitute a byte (8
bits). Each bit must be 1 (test bit) or 0 (ignore bit). You can specify up to 256 pairs
of hexadecimal digits.
Bit string format: The format for coding a bit string mask is:
B'bbbbbbbb...bbbbbbbb'
The value bbbbbbbb represents 8 bits that constitute a byte. Each bit must be 1
(test bit) or 0 (ignore bit). You can specify up to 256 groups of 8 bits. The total
number of bits in the mask must be a multiple of 8. A bit mask string can only be
used with a bit operator.

The hexadecimal or bit mask is truncated or padded on the right to the byte length
of the binary field. The padding character is X'00' (all bits off and thus not tested).
Including records in the output data setbit operator test

examples
Example 1
INCLUDE
COND=(27,1,CH,EQ,CD,AND,18,1,BI,ALL,B10000000)

v Byte 27 contains D
AND
v Byte 18 has bit 0 on.
Example 2
INCLUDE
COND=(11,1,BI,BM,X85)
This example illustrates how to only include records in which byte 11 has some,
but not all of bits 0, 5 and 7 on. Results for selected field values are shown as
follows:
Table 23. Bit Comparison Example 2: Results for Selected Field Values
11,1,BI Result
Action
X'85'
False
Omit Record
X'C1'
True
Include Record
X'84'
True
Include Record
X'00'
False
Omit Record
Example 3
INCLUDE
114
COND=(11,2,ALL,B0001001000110100,
OR,21,1,NONE,B01001100),FORMAT=BI

v Bytes 11 through 12 have all of bits 3, 6, 10, 11 and 13 on
OR
v Byte 21 has none of bits 1, 4, or 5 on.
Results for selected field values are shown as follows:
11,2,BI Result
21,1,BI Value
21,1,BI Result
Action
X'1234'
True
X'4C'
False
Include Record
X'02C4'
False
X'81'
True
Include Record
X'0204'
False
X'40'
False
Omit Record
X'F334'
True
X'00'
True
Include Record
X'1238'
False
X'4F'
False
Omit Record
Method 2: Bit comparison tests

This method of bit logic testing allows you to test whether selected bits in a binary
field are either in an exact pattern of on and off bits, or not in that exact pattern.
Unlike Method 1 described previously, only equal and unequal comparisons
are allowed; however, this method has the advantage of being able to test for a
precise combination of on and off bits.

(p1,m1,BI,
EQ
NE
, constant )

(p1,m1,
BI,
EQ
NE
, constant )
Bit comparison operators are as follows:

EQ
Equal to
NE
Not equal to
Fields
p1,m1: These variables specify the binary field in the input record to be compared
to the bit constant.
v p1 specifies the first byte of the binary input field to be compared to the bit
constant, relative to the beginning of the input record.8 The first data byte of a
115

v m1 specifies the length of the field to be tested. The length can be 1 to 256 bytes.
Bit constant
A bit string constant that specifies the pattern to which the binary field is
compared. If a bit in the constant is 1 or 0, the corresponding bit in the field is
compared to 1 or 0, respectively. If a bit in the constant is . (period), the
corresponding bit in the field is ignored.
Bit string format: The format for coding a bit string constant is:
B'bbbbbbbb...bbbbbbbb'
The value bbbbbbbb represents 8 bits that constitute a byte. Each bit must be 1
(test bit for 1), 0 (test bit for 0) or . (ignore bit). You can specify up to 256 groups of
8 bits. The total number of bits in the mask must be a multiple of 8. A bit constant
can only be used for bit comparison tests (BI format and EQ or NE operator).

The bit constant is truncated or padded on the right to the byte length of the
binary field. The padding character is B'00000000' (all bits equal to 0). Note that the
padded bytes are compared to the excess bytes in the binary field.
Recommendation: To ensure that comparison of the padded bytes to the excess
bytes in the binary field does not cause unwanted results, shorten the field length
to eliminate the padding characters, or increase the length of the bit constant to
specify the exact test pattern you want.
Including records in the output data setbit comparison test

examples
Example 1
INCLUDE
COND=(27,1,CH,EQ,CD,AND,18,1,BI,EQ,B1.......)

AND
v Byte 18 is equal to the specified pattern of bit 0 on.
Example 2
INCLUDE
COND=(11,1,BI,NE,B10...1.1)
This example illustrates how to only include records in which byte 11 is not equal
to the specified pattern of bit 0 on, bit 1 off, bit 5 on and bit 7 on. Results for
selected field values are shown :
116
11,1,BI Result
Action
X'85'
False
Omit Record
X'C1'
True
Include Record
X'84'
True
Include Record
X'97'
False
Omit Record
Example 3
INCLUDE
COND=(11,2,EQ,B..01....0......1,
OR,21,1,EQ,B01......),FORMAT=BI

v Bytes 11 through 12 are equal to the specified pattern of bit 2 off, bit 3 on, bit 8
off and bit 15 on
OR
v Byte 21 is equal to the specified pattern of bit 0 off and bit 1 on.
Results for selected field values are shown in Table 26
11,2,BI Result
21,1,BI Value
21,1,BI Result
Action
X'1221'
True
X'C0'
False
Include Record
X'02C4'
False
X'41'
True
Include Record
X'1234'
False
X'00'
False
Omit Record
X'5F7F'
True
X'7F'
True
Include Record
X'FFFF'
False
X'2F'
False
Omit Record
Date comparisons
You can use DFSORT's Y2 formats in conjunction with the century window in
effect, as follows:
v Use the full date formats (Y2T, Y2U, Y2V, Y2W, Y2X and Y2Y) to compare a
two-digit year date field to a two-digit year date constant, a current, future or
past two-digit year date (Y constant), or to another two-digit year date field.
v Use the year formats (Y2C, Y2Z, Y2S, Y2P, Y2D and Y2B) to compare a two-digit
year field to a two-digit year constant (Y constant) or to another two-digit year
field.
For example, you can include only those records for which a Z'yymm' date field is
between January 1996 and March 2005. Or you can include only those records for
which a P'dddyy' field is less than another P'dddyy' field. Or you can include only
those records for which a C'yyddd' field is between today's date and 5 days earlier
than today's date.
The ordering of dates and special indicators used for comparisons with Y2 fields
and Y constants is the same as the ascending orders for sorting and merging Y2
fields (see SORT control statement on page 442 for details).

(p1,m1,Y2x,
EQ
NE
GT
GE
LT
LE
p2,m2,Y2x
constant
Or, if the FORMAT=Y2x operand is used:

117

(p1,m1,
Y2x,
EQ
NE
GT
GE
LT
LE
p2,m2
,Y2x
constant
Comparison operators are as follows:

EQ
Equal to
NE
Not equal to
GT
Greater than
GE
Greater than or equal to
LT
Less than
LE
Less than or equal to.
Fields:
p1,m1,Y2x: These variables specify a two-digit year date field in the input record
to be compared either to another two-digit year date field in the input record or to
a two-digit year date constant.
v p1 specifies the first byte of the date field relative to the beginning of the input
record.9 The first data byte of a fixed-length record (FLR) has relative position 1.
The first data byte of a variable-length (VLR) record has relative position 5
(because the first 4 bytes contain the record descriptor word). All date fields
must start on a byte boundary, and no date field can extend beyond byte 32752.
v m1 specifies the length of the date field. Appendix C, Data format
descriptions, on page 891 describes the length and format for each type of date
field.
v Y2x specifies the Y2 format. Appendix C, Data format descriptions, on page
891 describes the length (m) and format (Y2x) for each type of date field.
You can use p1,m1 rather than p1,m1,Y2x if you use FORMAT=Y2x to supply the
format for the date field.
p2,m2,Y2x: These variables specify another two-digit year date field in the input
record with which the p1,m1,Y2x field will be compared.
You can use p2,m2 rather than p2,m2,Y2x if you use FORMAT=Y2x to supply the
format for the date field.
Constant: A two-digit year date constant in the form Y'string', Y'DATE1',
Y'DATE1'+d, Y'DATE1'-d, Y'DATE2', Y'DATE2'+m, Y'DATE2'-m, Y'DATE3',
Y'DATE3'+d, or Y'DATE3'-d, with which the p1,m1,Y2x field will be compared.
Comparisons: A date field can be compared to a date constant or another date
field with the same number of non-year (x) digits. Table 27 on page 119 shows the
type of field-to-field and field-to-constant comparisons you can use. The fields
shown for any type of date (for example, yyx and xyy) can be compared to any
other fields shown for that type of date or to the Y constant shown for that type of
date.
118

Table 27. Permissible Comparisons for Dates
Type of Date
Fields (m,f)
Y Constant
yyx and xyy
3,Y2T
3,Y2W
2,Y2U
2,Y2X
Y'yyx'
yyxx and xxyy
4,Y2T
4,Y2W
3,Y2V
3,Y2Y
Y'yyxx'
Y'DATE2'
Y'DATE2'+m
Y'DATE2'-m
yyxxx and xxxyy
5,Y2T
5,Y2W
3,Y2U
3,Y2X
Y'yyxxx'
Y'DATE3'
Y'DATE3'+d
Y'DATE3'-d
yyxxxx and xxxxyy
6,Y2T
6,Y2W
4,Y2V
4,Y2Y
Y'yyxxxx'
Y'DATE1'
Y'DATE1'+d
Y'DATE1'-d
yy
2,Y2C
2,Y2S
1,Y2D
2,Y2Z
2,Y2P
1,Y2B
Y'yy'
Y constants for current, future, and past two-digit year dates are as follows. d can
be 0 to 9999 days and m can be 0 to 9999 months.
v Y'DATE1' generates a Y constant for the current date in the form Y'yymmdd'
v Y'DATE1'+d generates a Y constant for the current date plus d days in the form
Y'yymmdd'
v Y'DATE1'-d generates a Y constant for the current date minus d days in the form
Y'yymmdd'
v Y'DATE2' generates a Y constant for the current date in the form Y'yymm'
v Y'DATE2'+m generates a Y constant for the current date plus m months in the
form Y'yymm'
v Y'DATE2'-m generates a Y constant for the current date minus m months in the
form Y'yymm'
v Y'DATE3' generates a Y constant for the current date in the form Y'yyddd'
v Y'DATE3'+d generates a Y constant for the current date plus d days in the form
Y'yyddd'
v Y'DATE3'-d generates a Y constant for the current date minus d days in the form
Y'yyddd'.
You must use the same number of digits in a Y constant as the type of date;
leading zeros must be specified (for example, for Y'yymm', use Y'0001' for January
2000 and Y'0501' for January 2005).
You can also use Y constants for special indicators as follows:
v Y'0...0' (CH/ZD/PD zeros) and Y'9...9' (CH/ZD/PD nines) can be used with
Y2T, Y2U, Y2V, Y2W, Y2X and Y2Y dates. You must use the same number of
digits as the type of date (for example, Y'000' for yyq or qyy, Y'0000' for yymm
or mmyy, and so forth).
v
Y'LOW' (BI zeros), Y'BLANKS' (blanks) and Y'HIGH' (BI ones) can be used with
Y2T, Y2W and Y2S dates.
119
Including records in the output data setdate comparisons

Example 1
INCLUDE FORMAT=Y2T,
COND=(3,4,GE,Y9901,AND,
3,4,LE,Y0312,OR,
3,4,LE,Y0000)

v A C'yymm' date field in bytes 3 through 6 is between January 1999 and
December 2003
OR
v Bytes 3 through 6 contain CH zeros (C'0000'), ZD zeros (Z'0000') or BI zeros
(X'00000000').
Note that the century window in effect will be used to interpret the Y'9901' and
Y'0312' date constants, as well as real dates in the C'yymm' date field. However,
the century window will not be used to interpret the Y'0000' special indicator
constant or special indicators in the C'yymm' date field.
Example 2
INCLUDE COND=(2,3,Y2X,LT,36,5,Y2T)
This example illustrates how to only include records in which a P'dddyy' date field
in bytes 2 through 4 is less than a Z'yyddd' date field in bytes 36 through 40.
Note that the century window in effect will be used to interpret real dates in the
P'dddyy' and Z'yyddd' date fields. However, the century window will not be used
to interpret special indicators in the P'dddyy' and Z'yyddd' date fields.
Numeric tests
You can test a field for numerics or non-numerics in character, zoned decimal or
packed decimal format.
For example, you can include only those records in which a 5-byte field contains
only '0'-'9' characters (that is, character numerics). Or you can include only those
records in which a 9-byte field contains invalid zoned decimal data (that is, zoned
decimal non-numerics). Or you can include only those records in which a 12-byte
field contains valid packed decimal data (that is, packed decimal numerics).
A field to be tested for numerics in character format looks like this in hexadecimal:
FdFd...Fd
The field is considered to be character numeric if every d is 0-9. (This is equivalent

to '0'-'9' for each character.) Otherwise, the field is considered to be character
non-numeric. For example, '1234', '0001' and '9999' are all considered to be
character numeric, whereas 'A234', '12.3', ' 1' and '123D' are all considered to be
character non-numeric.
A field to be tested for numerics in zoned decimal format looks like this in
hexadecimal:
zdzd...sd
The field is considered to be zoned decimal numeric if every z is F, every d is 0-9,

and s is C, D or F. Otherwise, the field is considered to be zoned decimal
120

non-numeric. For example, '1234' (X'F1F2F3F4'), '123D' (X'F1F2F3C4') and '123M'
(X'F1F2F3D4') are all considered to be zoned decimal numeric, whereas 'A234'
(X'C1F2F3F4'), '12.3' (X'F1F24BF3') and '123X' (X'F1F2F3E7') are all considered to be
zoned decimal non-numeric.
A field to be tested for numerics in packed decimal format looks like this in
hexadecimal:
dddd...ds
The field is considered to be packed decimal numeric if every d is 0-9, and s is C,

D or F. Otherwise, the field is considered to be packed decimal non-numeric. For
example, X'12345C', X'12345D' and X'12345F' are all considered to be packed
decimal numeric, whereas X'A2345C', X'1B345D' and X'12F45F' are all considered to
be packed decimal non-numeric.

(p1,m1,f1,
EQ
NE
,NUM)
Or, if the FORMAT=f operand is used:

(p1,m1,
f1,
EQ
NE
,NUM)
Numeric test operators are as follows:

EQ
Equal to numerics
NE
Not equal to numerics (non-numerics)
Fields:
p1,m1,f1: These variables specify the field in the input record for the numeric test.
v p1 specifies the first byte of the field relative to the beginning of the input
record 10. The first data byte of a fixed-length record (FLR) has relative position
1. The first data byte of a variable-length (VLR) record has relative position 5
(because the first 4 bytes contain the record descriptor word). All fields must
start on a byte boundary, and no field can extend beyond byte 32752.
v m1 specifies the length of the field. The length can be 1 to 256 bytes.
v f1 specifies the type of numerics the field is to be tested for as follows:
FS
test for numerics in character format ('0'-'9' (X'F0'-X'F9') in all bytes).

Note: CSF can be used instead of FS.
ZD
test for numerics in zoned decimal format ('0'-'9' (X'F0'-X'F9') in all

non-sign bytes; X'F0'-X'F9', X'D0'-X'D9' or X'C0'-X'C9' in the sign byte)
PD
test for numerics in packed decimal format (0-9 for all digits; F, D or C
for the sign).
121

You can use p1,m1 rather than p1,m1,f1, if you use FORMAT=f to supply the
NUM: Specifies a test for numerics or non-numerics. The condition will be true if
the field is numeric when the EQ operator is specified or if the field is
non-numeric when the NE operator is specified.
Including records in the output data set--numeric tests

Example 1
INCLUDE COND=(1,20,FS,EQ,NUM)
This example illustrates how to only include records in which the field in bytes 1
through 20 contains valid character numeric data (that is, '0'-'9' in all bytes).
Example 2
INCLUDE COND=(21,8,ZD,NE,NUM,OR,31,5,PD,NE,NUM)
through 28 contains invalid zoned decimal data, or the field in bytes 31 through 35
contains invalid packed decimal data (that is, one of the fields is non-numeric).
Alphanumeric tests
You can test a field for alphanumerics or non-alphanumerics in character format.
Various combinations of uppercase characters (A-Z), lowercase characters (a-z) and
numeric characters (0-9) can be used.
For example, you can include only those records in which a 10-byte field contains
only 'A'-'Z' characters (that is, uppercase characters) or '0'-'9' characters (that is,
numeric characters). Or you can include only those records in which a 20-byte field
contains characters other than 'a'-'z' (that is, lowercase characters).
The following combinations of alphanumeric characters can be used:
v Uppercase characters (A-Z)
v Lowercase characters (a-z)
v Mixed case characters (A-Z, a-z)
v Uppercase and numeric characters (A-Z, 0-9)
v Lowercase and numeric characters (a-z, 0-9)
v Mixed case and numeric characters (A-Z, a-z, 0-9)

(p1,m1,BI,
EQ
NE
UC
LC
MC
UN
LN
MN
122

(p1,m1,
BI,
EQ
NE
UC
LC
MC
UN
LN
MN
Alphanumeric test operators are as follows:

EQ
Equal to specified set of alphanumerics
NE
Not equal to specified set of alphanumerics (non-alphanumerics)
Fields:
p1,m1
These variables specify the field in the input record for the alphanumeric
test.
v p1 specifies the first byte of the field relative to the beginning of the
input record11. The first data byte of a fixed-length record (FLR) has
relative position 1. The first data byte of a variable-length (VLR) record
has relative position 5 (because the first 4 bytes contain the record
descriptor word). All fields must start on a byte boundary, and no field
can extend beyond byte 32752.
v m1 specifies the length of the field. The length can be 1 to 256 bytes.
UC
Specifies a test for the set of uppercase characters (A-Z). The condition will
be true if the field is all uppercase characters when the EQ operator is
specified or if the field is not all uppercase characters when the NE
operator is specified.
LC
Specifies a test for the set of lowercase characters (a-z). The condition will
be true if the field is all lowercase characters when the EQ operator is
specified or if the field is not all lowercase characters when the NE
operator is specified.
MC
Specifies a test for the set of mixed case characters (A-Z and a-z). The
condition will be true if the field is all mixed case characters when the EQ
operator is specified or if the field is not all mixed case characters when
the NE operator is specified.
UN
Specifies a test for the set of uppercase and numeric characters (A-Z and
0-9). The condition will be true if the field is all uppercase or numeric
characters when the EQ operator is specified or if the field is not all
uppercase or numeric characters when the NE operator is specified.
LN
Specifies a test for the set of lowercase and numeric characters (a-z and
0-9). The condition will be true if the field is all lowercase or numeric
characters when the EQ operator is specified or if the field is not all
lowercase or numeric characters when the NE operator is specified.
MN
Specifies a test for the set of mixed case and numeric characters (A-Z, a-z
or 0-9). The condition will be true if the field is all mixed case or numeric
characters when the EQ operator is specified or if the field is not all mixed
case or numeric characters when the NE operator is specified.
123
Including records in the output data set--alphanumeric tests

Example 1
INCLUDE COND=(11,10,BI,EQ,MC)
through 20 contains only mixed case characters (that is, 'A'-'Z' or 'a'-'z' in all bytes).
A record with 'AaBbZRStuv' in 11-20 would be included, whereas a record with
'Aa7BtuvZQR' would not be included.
Example 2
INCLUDE COND=(21,4,BI,NE,LN)
through 24 does not contain all lowercase or numeric characters (that is, one of the
bytes is not 'a'-'z' or '0'-'9'). A record with 'a,23' would be included, whereas a
record with 'a2b9' would not be included.
INCLUDE/OMIT statement notes

v Floating point compare fields cannot be referenced in INCLUDE or OMIT
statements.
v
v Any selection can be performed with either an INCLUDE or an OMIT statement.
INCLUDE and OMIT are mutually exclusive.
v If several relational conditions are joined with a combination of AND and OR
logical operators, the AND statement is evaluated first. The order of evaluation
can be changed by using parentheses inside the COND expression.
v If any changes are made to record formats by user exits E15 or E32, the
INCLUDE or OMIT statement must apply to the newest formats.
v DFSORT issues a message and terminates if an INCLUDE or OMIT statement is
specified for a tape work data set sort or conventional merge application.
v If SZERO is in effect, -0 compares as less than +0 when numeric fields and
constants are used. If NOSZERO is in effect, -0 compares as equal to +0 when
numeric fields and constants are used.
Table 28 shows how DFSORT reacts to the result of a relational condition
comparison, depending on whether the statement is INCLUDE or OMIT and
whether the relational condition is followed by an AND or an OR logical operator.
When writing complex statements, the table in Table 28 helps you get the result
that you want.
Table 28. Logic Table for INCLUDE/OMIT.
Relational Condition Program action if next logical operator is:
124
Statement
Compare
AND
OR
OMIT
True
Check next compare,

or if last compare
OMIT record.
OMIT record
OMIT
False
INCLUDE record
Check next compare,

or if last compare,
INCLUDE record.

Table 28. Logic Table for INCLUDE/OMIT. (continued)
Relational Condition Program action if next logical operator is:
Statement
Compare
AND
OR
INCLUDE
True
Check next compare,

or if last compare,
INCLUDE record.
INCLUDE record
INCLUDE
False
OMIT record
Check compare, or if
last compare, OMIT
record.
INREC control statement

,
INREC
PARSE=( definition
,
FIELDS=
BUILD=
( item
,
OVERLAY=( item
,
FINDREP=( item
,
IFTHEN=(clause)
IFOUTLEN=n
The INREC control statement allows you to reformat the input records before they
are sorted, merged, or copied.
The INREC control statement supports a wide variety of parsing, editing, and
reformatting tasks, including:
records.
v
125
INREC Control Statement
v
v
Sophisticated editing capabilities, such as control of the way numeric fields are
so on.
forms.

lookup table, based on a character, hexadecimal, or bit string as input (that is,
lookup and change).
You can create the reformatted INREC records in one of the following ways using
unedited, edited, or converted input fields (p,m for fixed fields, or %nn for parsed
fields - see PARSE), and a variety of constants:
v BUILD or FIELDS: Reformat each record by specifying all of its items one by
one. Build gives you complete control over the items you want in your
reformatted INREC records and the order in which they appear. You can delete,
rearrange and insert fields and constants. Example:
INREC BUILD=(1,20,CABC,26:5C*,
15,3,PD,EDIT=(TTT.TT),21,30,80:X)
v OVERLAY: Reformat each record by specifying just the items that overlay
specific columns. Overlay lets you change specific existing columns without
affecting the entire record. Example:
INREC OVERLAY=(45:45,8,TRAN=LTOU)
v FINDREP: Reformat each record by doing various types of find and replace
operations. Example:
INREC FINDREP=(IN=CMr.,OUT=CMister)
v IFTHEN clauses: Reformat different records in different ways by specifying how

build, overlay, find/replace, or group operation items are applied to records that
meet given criteria. IFTHEN clauses let you use sophisticated conditional logic
to choose how different record types are reformatted. Example:
INREC IFTHEN=(WHEN=(1,5,CH,EQ,CTYPE1),
BUILD=(1,40,C**,+1,TO=PD)),
IFTHEN=(WHEN=(1,5,CH,EQ,CTYPE2),
BUILD=(1,40,+2,TO=PD,XFFFF)),
IFTHEN=(WHEN=NONE,OVERLAY=(45:CNONE))
You can choose to include any or all of the following items in your reformatted
INREC records:
v Fixed position/length fields or variable position/length fields. For fixed fields,
you specify the starting position and length of the field directly. For variable
fields, such as delimited fields, comma separated values (CSV), tab separated
values, blank separated values, keyword separated fields, null-terminated strings
126

(and many other types), you define rules that allow DFSORT to extract the
relevant data into fixed parsed fields, and then use the parsed fields as you
would use fixed fields.
v Blanks, binary zeros, character strings, and hexadecimal strings
v Current date, future date, past date, and current time in various forms
v Unedited input fields aligned on byte, halfword, fullword, and doubleword
boundaries
v Replaced or removed strings.
v Hexadecimal or bit representations of binary input fields
v Characters translated from uppercase to lowercase, lowercase to uppercase,
ASCII to EBCDIC or EBCDIC to ASCII.
v Left-justified, right-justified, left-squeezed, or right-squeezed input fields.
v Numeric input fields of various formats converted to different numeric formats,
or to character format edited to contain signs, thousands separators, decimal
points, leading zeros or no leading zeros, and so on.
v Decimal constants converted to different numeric formats, or to character format
edited to contain signs, thousands separators, decimal points, leading zeros or
no leading zeros, and so on.
v The results of arithmetic expressions combining fields, decimal constants,
operators (MIN, MAX, MUL, DIV, MOD, ADD and SUB) and parentheses
converted to different numeric formats, or to character format edited to contain
signs, thousands separators, decimal points, leading zeros or no leading zeros,
and so on.
v SMF, TOD and ETOD date and time fields converted to different numeric
formats, or to character format edited to contain separators, leading zeros or no
leading zeros, and so on.
v Input date fields of one type (CH, ZD, PD, 2-digit year, 4-digit year, Julian,
Gregorian) converted to corresponding output date fields of another type or to a
corresponding day of the week.
v The results of various types of arithmetic operations for input date fields.
v Sequence numbers in various formats.
v A character constant, hexadecimal constant or input field selected from a lookup
table, based on a character, hexadecimal or bit constant as input.
v A zoned decimal group identifier, a zoned decimal group sequence number, or a
field propagated from the first record of a group to all of the records of a group.
For information concerning the interaction of INREC and OUTREC, see INREC
statement notes on page 148.
PARSE
127

,
,
PARSE=(
%n=
%nn=
%nnn=
%=
FIXLEN=m
ABSPOS=p
ADDPOS=x
SUBPOS=y
,
STARTAFT=string
STARTAFT=an
STARTAFT=BLANKS
STARTAT=string
STARTAT=an
STARTAT=BLANKS
STARTAT=NONBLANK
,
ENDBEFR=string
ENDBEFR=an
ENDBEFR=BLANKS
ENDAT=string
ENDAT=an
ENDAT=BLANKS
PAIR=APOST
PAIR=QUOTE
REPEAT=v
This operand allows you to extract variable position/length fields into fixed
parsed fields. Parsed fields (%n, %nn or %nnn) can be used where fixed
position/length fields (p,m) can be used in the BUILD (or FIELDS) or
OVERLAY operands as described later in this section.
Note: Although you can use %n (%0-%9), %nn (%00-%99) or %nnn
(%000-%999) for a parsed field, for convenience in this book %nn will be used
in general when referring to a parsed field. %n, %0n or %00n can be used
interchangeably for parsed field n (for example, %1, %01 or %001 for parsed
field 1). %nn or %0nn can be used interchangeably for parsed field nn (for
example, %12 or %012 for parsed field 12).
PARSE can be used for many different types of variable fields including
delimited fields, comma separated values (CSV), tab separated values, blank
separated values, keyword separated fields, null-terminated strings, and so on.
You can assign up to 1000 parsed fields (%0-%999) to the variable fields you
want to extract.
Note that if all of the fields in your records have fixed positions and lengths,
you don't need to use PARSE. But if any of the fields in your records have
variable positions or lengths, you can use PARSE to treat them as fixed parsed
fields in BUILD or OVERLAY. You can mix p,m fields (fixed fields) and %nn
fields (parsed fields) in BUILD and OVERLAY.
See PARSE under "OUTFIL Control Statements" for complete details.
Sample Syntax
INREC PARSE=(%00=(ENDBEFR=C*,FIXLEN=3),
%01=(ENDBEFR=BLANKS,FIXLEN=6),
%02=(STARTAT=CMAX,FIXLEN=8),
%03=(STARTAFT=C(,ENDBEFR=C),FIXLEN=6),
128

%04=(STARTAFT=BLANKS,FIXLEN=5)),
BUILD=(%03,X,%03,HEX,21:%02,31:%01,SFF,M26,LENGTH=7,
18,6,%00,UFF,M11,LENGTH=3,%04,JFY=(SHIFT=RIGHT))
Default for PARSE: None; must be specified. See Appendix B,

FIELDS or BUILD
,
FIELDS=
BUILD=

c:
s
p,m
,a
%nn
p
p,m
%nn
p
,TRAN
LTOU
UTOL
ALTSEQ
ATOE
ETOA
HEX
UNHEX
BIT
UNBIT
p,m
,HEX
%nn
p
p,m,f
,edit
%nn,f
,to
(p,m,f)
(%nn,f)
deccon
,edit
(deccon)
,to
arexp
,edit
(arexp)
,to
p,m
,Y2x
%nn
,Y4x
,edit
,to
,todate
,dateop
,Y2x(s)
,Y4x(s)
,Y2xP
p,m
,lookup
%nn
p,m
,justify
%nn
p,m
,squeeze
%nn
seqnum
Specifies all of the items in the reformatted INREC record in the order in
which they are to be included. The reformatted INREC record consists of the
separation fields, edited and unedited input fields (p,m for fixed fields, or %nn
for parsed fields - see PARSE),, edited decimal constants, edited results of
arithmetic expressions, and sequence numbers you select, in the order in which
you select them, aligned on the boundaries or in the columns you indicate.
129

For variable-length records, the first item in the BUILD or FIELDS parameter
must specify or include the unedited 4-byte record descriptor word (RDW),
that is, you must start with 1,m with m equal to or greater than 4. If you want
to include the bytes from a specific position to the end of each input record at
the end of each reformatted output record, you can specify that starting
position (p) as the last item in the BUILD or FIELDS parameter. For example:
INREC FIELDS=(1,4,
1,2,BI,TO=ZD,LENGTH=5,
C|,
5)
unedited RDW
display RDW length in decimal
| separator
display input positions 5 to end
For fixed-length records, the first input and output data byte starts at position
1. For variable-length records, the first input and output data byte starts at
position 5, after the RDW in positions 1-4.
c: Specifies the position (column) for a separation field, input field, decimal
constant, arithmetic expression, or sequence number, relative to the start of
the reformatted input record. Unused space preceding the specified column
is padded with EBCDIC blanks. The following rules apply:
v c must be a number between 1 and 32752.
v c: must be followed by a separation field, input field, decimal constant,
or arithmetic expression.
v c must not overlap the previous input field or separation field in the
reformatted input record.
v for variable-length records, c: must not be specified before the first input
field (the record descriptor word) nor after the variable part of the input
record.
v The colon (:) is treated like the comma (,) or semicolon (;) for
continuation to another line.
Both valid and invalid examples are shown in Table 29.
Table 29. Examples of Valid and Invalid Column Alignment. Examples of Valid and Invalid
Column Alignment
Validity
Specified
Result
Valid
33:C'State '
Columns 1-32 blank

Columns 33-38 'State '
Valid
20:5,4,30:10,8
Columns
Columns
Columns
Columns
Invalid
0:5,4
Column value cannot be zero.
Invalid
:25Z
Column value must be specified.
Invalid
32753:21,8
Invalid column value must be less

than 32753.
Invalid
5:10:2,5
Column values cannot be adjacent.
Invalid
20,10,6:C'AB'
Column value overlaps previous field.
130
1-19 blank
20-23 input field (5,4)
24-29 blank
30-37 input field (10,8)
Specifies that a separation field (blanks, zeros, character string,

hexadecimal string, current date, future date, past date, or current time) is
to appear in the reformatted input record. It can be specified before or after
any input field. Consecutive separation fields can be specified. For
variable-length records, separation fields must not be specified before the

first input field (the record descriptor word), or after the variable part of
the input record. Permissible values are nX, nZ, nC'xx...x', nX'yy...yy', and
various date and time constants.
nX
Blank separation. n bytes of EBCDIC blanks (X'40') are to appear in

the reformatted input records. n can range from 1 to 4095. If n is
omitted, 1 is used.
Examples of valid and invalid blank separation are shown in
Table 30.
Table 30. Examples of Valid and Invalid Blank Separation

Specified
Result
Valid
X or 1X
1 blank
Valid
4095X
4095 blanks
Invalid
5000X
Too many repetitions. Use two adjacent separation

fields instead (2500X,2500X, for example)
Invalid
0X
0 is not allowed.
nZ
Binary zero separation. n bytes of binary zeros (X'00') are to appear
in the reformatted input records. n can range from 1 to 4095. If n is
omitted, 1 is used.
Examples of valid and invalid binary zero separation are shown in
Table 31.
Table 31. Examples of Valid and Invalid Binary Zero Separation
Specified
Result
Valid
Z or 1Z
1 binary zero
Valid
4095Z
4095 binary zeros
Invalid
4450Z
Too many repetitions. Use two adjacent separation

fields instead (4000Z,450Z for example).
Invalid
0Z
0 is not allowed.
nC'xx...x'
Character string separation. n repetitions of the character string
constant (C'xx...x') are to appear in the reformatted input records. n
can range from 1 to 4095. If n is omitted, 1 is used. x can be any
EBCDIC character. You can specify from 1 to 256 characters.
If you want to include a single apostrophe in the character string,
you must specify it as two single apostrophes:
Required:
ONEILL
Specify:
CONEILL
Examples of valid and invalid character string separation are

shown in Table 32.
Table 32. Examples of Valid and Invalid Character String Separation
Specified
Result
Length
Valid
C'John Doe'
John Doe
Valid
C'JOHN DOE'
JOHN DOE
Valid
C'$@#'
$@#
131

Table 32. Examples of Valid and Invalid Character String Separation (continued)
Specified
Result
Length
Valid
C'+0.193'
+0.193
Valid
4000C'
8000 blanks
8000
Valid
20C'**FILLER**'
**FILLER** repeated 20 times
200
Valid
C'Frank''s'
Frank's
Invalid
C'''''
Apostrophes not paired
n/a
Invalid
'ABCDEF'
C identifier missing
n/a
Invalid
C'ABCDE
Apostrophe missing
n/a
Invalid
4450C'1'
Too many repetitions. Use two

adjacent separation fields instead
(4000C'1',450C'1', for example).
n/a
Invalid
0C'ABC'
0 is not allowed
n/a
Invalid
C''
No characters specified
n/a
Invalid
C'Frank's'
Two single apostrophes needed for

one
n/a
'
nX'yy...yy'
Hexadecimal string separation. n repetitions of the hexadecimal
string constant (X'yy...yy') are to appear in the reformatted input
records. n can range from 1 to 4095. If n is omitted, 1 is used.
The value yy represents any pair of hexadecimal digits. You can
specify from 1 to 256 pairs of hexadecimal digits. Examples of
valid and invalid hexadecimal string separation are shown in
Table 33.
Table 33. Examples of Valid and Invalid Hexadecimal String Separation
Specified
Result
Length
Valid
X'FF'
FF
Valid
X'BF3C'
BF3C
Valid
3X'00000F'
00000F00000F00000F
Valid
4000X'FFFF'
FF repeated 8000 times
8000
Invalid
X'ABGD'
G is not a hexadecimal digit
n/a
Invalid
X'F1F'
Incomplete pair of digits
n/a
Invalid
'BF3C'
X identifier missing
n/a
Invalid
'F2F1'X
X in wrong place
n/a
Invalid
8000X'01'
Too many repetitions. Use two

adjacent separation fields instead
(4000X'01',4000X'01', for example).
n/a
Invalid
0X'23AB'
0 is not allowed
n/a
Invalid
X''
No hexadecimal digits specified
n/a
DATEn, DATEn(c), DATEnP

Constant for current date. The current date of the run is to appear in the
reformatted input records. See DATEn, DATEn(c), DATEnP under OUTFIL
OUTREC for details.
132

&DATEn, &DATEn(c), &DATEnP
Can be used instead of DATEn, DATEn(c) and DATEnP, respectively.
DATEn+r, DATEn(c)+r, DATEnP+r
Constant for future date. A future date relative to the current date of the
run is to appear in the reformatted input records. See DATEn+r,
DATEn(c)+r, DATEnP+r under OUTFIL OUTREC for details.
&DATEn+r, &DATEn(c)+r, &DATEnP+r
Can be used instead of DATEn+r, DATEn(c)+r and DATEnP+r, respectively.
DATEn-r, DATEn(c)-r, DATEnP-r
Constant for past date. A past date relative to the current date of the run is
to appear in the reformatted input records. See DATEn-r, DATEn(c)-r,
DATEnP-r under OUTFIL OUTREC for details.
&DATEn-r, &DATEn(c)-r, &DATEnP-r
Can be used instead of DATEn-r, DATEn(c)-r and DATEnP-r, respectively.
TIMEn, TIMEn(c), TIMEnP

Constant for current time. The time of the run is to appear in the
reformatted input records. See TIMEn, TIMEn(c), TIMEnP under OUTFIL
OUTREC for details.
&TIMEn, &TIMEn(c), &TIMEnP
Can be used instead of TIMEn, TIMEn(c) and TIMEnP, respectively.
DATE
Specifies that the current date is to appear in the reformatted input records
in the form 'mm/dd/yy'. See DATE under OUTFIL OUTREC for details.
&DATE
Can be used instead of DATE.
DATE=(abcd)
in the form 'adbdc'. See DATE=(abcd) under OUTFIL OUTREC for details.
&DATE=(abcd)
Can be used instead of DATE=(abcd).
DATENS=(abc)
in the form 'abc'. See DATENS=(ab) under OUTFIL OUTREC for details.
&DATENS=(abc)
Can be used instead of DATENS=(abc).
YDDD=(abc)
specifies that the current date is to appear in the reformatted input records
in the form 'acb'. See YDDD=(abc) under OUTFIL OUTREC for details.
&YDDD=(abc)
Can be used instead of YDDD=(abc).
YDDDNS=(ab)
specifies that the current date is to appear in the reformatted input records
in the form 'ab'. See YDDDNS=(ab) under OUTFIL OUTREC for details.
&YDDDNS=(ab)
can be used instead of YDDDNS=(ab).
133

TIME
specifies that the current time is to appear in the reformatted input records
in the form 'hh:mm:ss'. See TIME under OUTFIL OUTREC for details.
&TIME
Can be used instead of TIME.
TIME=(abc)
specifies that the current time is to appear in the reformatted input records
in the form 'hhcmmcss' (24-hour time) or 'hhcmmcss xx' (12-hour time). See
TIME=(abc) under OUTFIL OUTREC for details.
&TIME=(abc)
Can be used instead of TIME=(abc).
TIMENS=(ab)
specifies that the current time is to appear in the reformatted input record
in the form 'hhmmss' (24-hour time) or 'hhmmss xx' (12-hour time). See
TIMENS=(ab) under OUTFIL OUTREC for details.
&TIMENS=(ab)
Can be used instead of TIMENS=(ab).
p,m,a
Specifies that an unedited input field is to appear in the reformatted input
record.
p
Specifies the first byte of the input field relative to the beginning of
the input record.12 The first data byte of a fixed-length record has
relative position 1. The first data byte of a variable-length record
has relative position 5 (because the first 4 bytes contain the RDW).
All fields must start on a byte boundary, and no field can extend
beyond byte 32752. For special rules concerning variable-length
records, see INREC statement notes on page 148.
Specifies the length of the input field. It must include the sign if
the data is signed, and must be an integer number of bytes. See
INREC statement notes on page 148 for more information.
Specifies the alignment (displacement) of the input field in the

reformatted input record relative to the start of the reformatted
input record.
Permissible values of a are:
H
Halfword aligned. The displacement (p-1) of the field from

the beginning of the reformatted input record, in bytes, is a
multiple of two (that is, position 1, 3, 5, and so forth).
Fullword aligned. The displacement is a multiple of four

(that is, position 1, 5, 9, and so forth).
Doubleword aligned. The displacement is a multiple of

eight (that is, position 1, 9, 17, and so forth).
Alignment can be necessary if, for example, the data is to be used

in a COBOL application program where items are aligned through
the SYNCHRONIZED clause. Unused space preceding aligned
fields will always be padded with binary zeros.
12. If your E15 user exit reformats the record, p must refer to the record as reformatted by the exit.
134

%nn
specifies that an unedited parsed input field is to appear in the reformatted
input record. See PARSE for details of parsed fields. See p,m,a for further
details. Note that alignment (H, F, D) is not permitted for %nn fields (for
example, %nn,F results in an error message and termination).
p
specifies that the unedited variable part of the input record (that part
beyond the minimum record length), is to appear in the reformatted input
record, as the last field. p without m can only be used for variable-length
records; not for fixed-length records.
Attention: If 1,4,p is specified (only RDW and variable part of record),
"null" records containing only an RDW will result if the record length is
less than p.
A value must be specified for p that is less than or equal to the minimum
record length (RECORD statement L4 value) plus 1 byte.
p,m,TRAN=keyword
Specifies that an input field is to be translated as indicated by the
keyword. See p,m,TRAN=keyword under OUTFIL OUTREC for details.
%nn,TRAN=keyword
Specifies that a parsed input field is to be translated as indicated by the
keyword. See %nn,TRAN=keyword under OUTFIL OUTREC for details.
p,TRAN=keyword
Specifies that the variable part of the input record is to be translated as
indicated by the keyword. See p,TRAN=keyword under OUTFIL OUTREC
for details.
p,m,HEX
Can be used instead of p,m,TRAN=HEX. See p,m,HEX under OUTFIL
OUTREC for details.
%nn,HEX
Can be used instead of %nn,TRAN=HEX. See %nn,HEX under OUTFIL
OUTREC for details.
p,HEX
Can be used instead of p,TRAN=HEX. See p,HEX under OUTFIL OUTREC
for details.
p,m,f,edit or (p,m,f),edit
specifies that an edited numeric input field is to appear in the reformatted
input record. You can edit BI, FI, PD, PD0, ZD, FL, CSF, FS, UFF, SFF, DC1,
DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3, TC4, TE1, TE2,
TE3, TE4, TM1, TM2, TM3 or TM4 fields using either pre-defined edit
masks (M0-M26) or specific edit patterns you define. You can control the
way the edited fields look with respect to length, leading or suppressed
zeros, thousands separators, decimal points, leading and trailing positive
and negative signs, and so on.
See p,m,f,edit under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(5:21,8,ZD,M19,X,46,5,ZD,M13,
31:(35,6,FS),SIGNS=(,,+,-),LENGTH=10,
51:8,4,PD,EDIT=(**II,IIT.TTXS),SIGNS=(,,+,-))
135

%nn,f,edit or (%nn,f),edit
specifies that an edited numeric parsed input field is to appear in the
reformatted input record. See PARSE for details of parsed fields. See
p,m,f,edit or (p,m,f),edit for further details.
p,m,f,to or (p,m,f),to
specifies that a converted numeric input field is to appear in the
reformatted input record. You can convert BI, FI, PD, PD0, ZD, FL, CSF, FS,
UFF, SFF, DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3,
TC4, TE1, TE2, TE3, TE4, TM1, TM2, TM3, or TM4 fields to BI, FI, PD,
PDC, PDF, ZD, ZDF, ZDC, or CSF/FS fields.
See p,m,f,to under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(21,5,ZD,TO=PD,X,(8,4,ZD),FI,LENGTH=2,X,55,4,FL,TO=FS)
%nn,f,to or (%nn,f),to
specifies that a converted numeric parsed input field is to appear in the
reformatted input record. See PARSE for details of parsed fields. See
p,m,f,to or (p,m,f),to for further details.
deccon,edit or (deccon),edit
specifies that an edited decimal constant is to appear in the reformatted
input record. The decimal constant must be in the form +n or -n where n is
1 to 31 decimal digits. The sign (+ or -) must be specified. A decimal
constant produces a signed, 31-digit zoned decimal (ZD) result to be edited
as specified.
See deccon,edit under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(5:+5000,EDIT=(T,TTT),X,
(-25500),M18,LENGTH=8)
deccon,to or (deccon),to
specifies that a converted decimal constant is to appear in the reformatted
input record. The decimal constant must be in the form +n or -n where n is
1 to 31 decimal digits. The sign (+ or -) must be specified. A decimal
constant produces a signed, 31-digit zoned decimal (ZD) result to be
converted as specified.
See deccon,to under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(+0,TO=PD,LENGTH=6,3Z,(-512000),FI)
arexp,edit or (arexp),edit
specifies that the edited result of an arithmetic expression is to appear in
the reformatted input record. The arithmetic expression can consist of
input fields, decimal constants, operators and parentheses. An arithmetic
expression produces a signed, 31-digit zoned decimal (ZD) result to be
edited as specified.
See arexp,edit under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(C**,27,2,FI,MIN,
83,4,PD,EDIT=(STTTTTTT),SIGNS=(+,-),
15:(((15,5,ZD,ADD,+1),MUL,+100),DIV,62,2,PD),M25,LENGTH=10)
136

arexp,to or (arexp),to
specifies that the converted result of an arithmetic expression is to appear
in the reformatted input record. The arithmetic expression can consist of
input fields, decimal constants, operators and parentheses. An arithmetic
expression produces a signed, 31-digit zoned decimal (ZD) result to be
See arexp,to under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=((15,6,FS,SUB,+5),ADD,(-1,MUL,36,6,FS),ZD,X,
3,2,FI,MIN,-6,LENGTH=4,TO=PD)
p,m,Y2x or p,m,Y4x
Specifies that an input date field is to be edited. Real Y2x dates are edited
using the century window established by the Y2PAST option in effect. Y2x
and Y4x dates with special indicators are expanded appropriately (for
example, p,6,Y2T transforms C'000000' to C'00000000').
See "p,m,Y2x or p,m,Y4x" under OUTFIL OUTREC for details.
Sample Syntax:
INREC BUILD=(21,3,Y2U,X,3,8,Y4W)
%nn,Y2x or %nn,Y4x
Specifies that a parsed input date field is to be edited. See PARSE for
details of parsed fields. See "p,m,Y2x or p,m,Y4x" for further details.
p,m,Y2x,edit or p,m,Y4x,edit
Specifies that the output for a p,m,Yxx input date field is to be edited
according to the edit parameters you specify. For example, if you specify:
INREC BUILD=(28,5,Y4V,EDIT=(TTTT-TT-TT)
The P'ccyymmdd' (X'0ccyymmddC') date value will be transformed to a

C'ccyymmdd' date value and then edited to a C'ccyy-mm-dd' date value.
See "p,m,Y2x or p,m,Y4x" and "p,m,f,edit" for related details.
%nn,Y2x,edit or %nn,Y4x,edit
Specifies that the output for a parsed Yxx input date field is to be edited
according to the edit parameters you specify. See PARSE for details of
parsed fields. See "p,m,Y2x,edit or p,m,Y4x,edit" for further details.
p,m,Y2x,to or p,m,Y4x,to
Specifies that an input date field is to be converted according to the to
parameters you specify. For example, if you specify:
INREC BUILD=(5,6,Y2W,TO=PD,LENGTH=5)
The C'mmddyy' date value will be transformed to a Z'mmddccyy' date

value and then converted to a P'mmddccyy' (X'0mmddccyyC') date value.
See "p,m,Y2x or p,m,Y4x" and "p,m,f,to" for related details.
%nn,Y2x,to or %nn,Y4x,to
Specifies that a parsed input date field is to be converted according to the
to parameters you specify. See PARSE for details of parsed fields. See
"p,m,Y2x,to or p,m,Y4x,to" for further details.
p,m,Y2x,todate or p,m,Y4x,todate
Specifies that an input date field of one type is to be converted to a
corresponding output date field of another type. You can convert date
fields between combinations of 2-digit and 4-digit year dates, CH/ZD and
137

PD dates, and Julian and Gregorian dates. You can also convert a date field
to a corresponding day of the week in several forms.
See "p,m,Y2x,todate or p,m,Y4x,todate" under OUTFIL OUTREC for details.
Sample Syntax:
* Convert a Pdddyy input date to a Cccyy/mm/dd output date
INREC BUILD=(21,3,Y2X,TOGREG=Y4T(/),X,
* Convert a Cccyymmdd input date to a Pccyyddd output date
42,8,Y4T,TOJUL=Y4U,X,
* Convert a Cmmddyy input date to a Cyymmdd output date
11,6,Y2W,TOGREG=Y2T)
%nn,Y2x,todate or %nn,Y4x,todate
corresponding output date field of another type. See PARSE for details of
parsed fields. See "p,m,Y2x,todate or p,m,Y4x,todate" for further details.
p,m,Y2x,dateop or p,m,Y4x,dateop
Specifies an arithmetic operation for an input date field. See
"p,m,Y2x,dateop or p,m,Y4x,dateop" under OUTFIL OUTREC for details.
%nn,Y2x,dateop or %nn,Y4x,dateop
Specifies an arithmetic operation for a parsed input date field. See
"%nn,Y2x,dateop or %nn,Y4x,dateop" under OUTFIL OUTREC for details.
p,m,Y2x(s) or p,m,Y4x(s)
Specifies that an input date field is to be edited with separators. s can be
any character except a blank. Real Y2x dates are edited using the century
window established by the Y2PAST option in effect. Y2x and Y4x dates
with special indicators are expanded appropriately (for example, p,8,Y4T(/)
transforms C'00000000' to C'0000/00/00').
See "p,m,Y2x(s) or p,m,Y4x(s)" under OUTFIL OUTREC for details.
Sample Syntax:
* Convert a Zdddccyy date to a Cddd/ccyy date.
INREC BUILD=(19,7,Y4W(/),X,
* Convert a Pccyymmdd date to a Cccyy-mm-dd date.
43,5,Y4V(-))
%nn,Y2x(s) or %nn,Y4x(s)
Specifies that a parsed input date field is to be edited with separators. See
PARSE for details of parsed fields. See "p,m,Y2x(s) or p,m,Y4x(s)" for
further details.
p,m,Y2xP
Specifies that an input date field is to be converted to a packed decimal
output date field. Real Y2x dates are edited using the century window
established by the Y2PAST option in effect. Y2x and Y4x dates with special
indicators are expanded appropriately (for example, p,6,Y2TP transforms
C'000000' to P'00000000').
See "p,m,Y2xP" under OUTFIL OUTREC for details.
Sample Syntax:
INREC BUILD=(11,3,Y2XP,X,21,4,Y2WP)
%nn,Y2xP
Specifies that a parsed input date field is to be converted to a packed
decimal output date field. See PARSE for details of parsed fields. See
"p,m,Y2xP" for further details.
138

p,m,lookup or %nn,lookup
specifies that a character constant, hexadecimal constant, input field (p,m),
or parsed input field (%nn) from a lookup table is to appear in the
reformatted input record. You can use p,m,lookup or %nn,lookup to select
a specified character set constant (that is, a character or hexadecimal string)
or set field (that is, an input field or parsed input field) based on matching
an input value against find constants (that is, character, hexadecimal, or bit
constants).
See p,m,lookup or %nn,lookup under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(11,1,
CHANGE=(6,
CR,CREAD,
CU,CUPDATE,
XFF,CEMPTY,
CA,CALTER),
NOMATCH=(11,6),
4X,
21,1,
CHANGE=(10,
B.1......,CVSAM,
B.0......,CNON-VSAM))
p,m,justify
specifies that a left-justified or right-justified input field is to appear in the
reformatted input record. For a left-justified field, leading blanks are
removed and the characters from the first nonblank to the last nonblank
are shifted left, with blanks inserted on the right if needed. For a
right-justified field, trailing blanks are removed and the characters from the
last nonblank to the first nonblank are shifted right, with blanks inserted
on the left if needed. Optionally:
v specific leading and trailing characters can be changed to blanks before
justification begins
v a leading string can be inserted
v a trailing string can be inserted
v the output length can be changed (it's equal to the input length by
default)
See p,m,justify under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(1,10,
21,20,JFY=(SHIFT=RIGHT),5X,
52,12,JFY=(SHIFT=LEFT,PREBLANK=C(),LEAD=CVOL=SER=,
LENGTH=16))
%nn,justify
specifies that a left-justified or right-justified parsed input field is to appear
in the reformatted input record. See PARSE for details of parsed fields. See
p,m,justify for further details.
p,m,squeeze
specifies that a left-squeezed or right-squeezed input field is to appear in
the reformatted input record. For a left-squeezed field, all blanks are
removed and the characters from the first nonblank to the last nonblank
are shifted left, with blanks inserted on the right if needed. For a
139

right-squeezed field, all blanks are removed and the characters from the
last nonblank to the first nonblank are shifted right, with blanks inserted
on the left if needed. Optionally:
specific characters can be changed to blanks before squeezing begins
a leading string can be inserted
a trailing string can be inserted
a string (for example, a comma delimiter) can be inserted wherever a
group of blanks is removed between the first nonblank and the last
nonblank
v blanks can be kept as is between paired apostrophes ('AB CD EF') or
paired quotes ("AB CD EF")
v the output length can be changed (it's equal to the input length by
default)
v
v
v
v
See p,m,squeeze under OUTFIL OUTREC for details.

Sample Syntax:
INREC FIELDS=(21,20,SQZ=(SHIFT=LEFT),5X,
152,18,SQZ=(SHIFT=RIGHT,PREBLANK=X00,
LEAD=C<,MID=C,,TRAIL=C>,PAIR=APOST))
%nn,squeeze
specifies that a left-squeezed or right-squeezed parsed input field is to
appear in the reformatted input record. See PARSE for details of parsed
fields. See p,m,squeeze for further details.
seqnum
specifies that a sequence number is to appear in the reformatted input
record. The sequence numbers are assigned in the order in which the
records are received for INREC processing. You can create BI, PD, ZD, CSF,
or FS sequence numbers and control their lengths, starting values and
increment values. You can restart the sequence number at the start value
each time a specified input field (p,m) or parsed input field (%nn) changes.
See seqnum under OUTFIL OUTREC for details.
Sample Syntax:
INREC FIELDS=(1,80,SEQNUM,8,ZD)
Default for BUILD or FIELDS: None; must be specified. See Appendix B,

OVERLAY
140

,
OVERLAY= (
c:
s
p,m
,a
%nn
p,m
%nn
,TRAN
LTOU
UTOL
ALTSEQ
ATOE
ETOA
HEX
UNHEX
BIT
UNBIT
p,m
,HEX
%nn
p,m,f
,edit
%nn,f
,to
(p,m,f)
(%nn,f)
deccon
,edit
(deccon)
,to
arexp
,edit
(arexp)
,to
p,m
,Y2x
%nn
,Y4x
,edit
,to
,todate
,dateop
,Y2x(s)
,Y4x(s)
,Y2xP
p,m
,lookup
%nn
p,m
,justify
%nn
p,m
,squeeze
%nn
seqnum
Specifies each item that is to overlay specific columns in the reformatted

record. Columns that are not overlaid remain unchanged. If you want to insert,
rearrange, or delete fields, use BUILD or FIELDS rather than OVERLAY. Use
OVERLAY only to overlay existing columns or to add fields at the end of every
record. OVERLAY can be easier to use then BUILD or FIELDS when you just
want to change a few fields without rebuilding the entire record.
Use c: (column) to specify the output positions to be overlaid. If you do not
specify c: for the first item, it defaults to 1:. If you do not specify c: for any
other item, it starts after the previous item. For example, if you specify:
INREC OVERLAY=(25,2,11:CA,15,3,C**)
141

Input positions 25-26 are placed at output positions 1-2; C'A' is placed at
output position 11; input positions 15-17 are placed at output positions 12-14;
and C'**' is placed at output positions 1516. The rest of the record remains
unchanged.
You can specify items in any order, you can change the same item multiple
times and you can overlap output columns. Changes to earlier items affect
changes to later items. For example, say you specify:
INREC OVERLAY=(21:8,4,ZD,MUL,+10,TO=ZD,LENGTH=6,
5:5,1,TRAN=UTOL,
5:5,1,CHANGE=(1,Ca,CX,Cb,CY),NOMATCH=(5,1))
and input position 5 has 'A'. The second item (UTOL) would change 'A' to 'a'
and the third item (CHANGE) would change 'a' again to 'X'.
If you specify an OVERLAY item that extends the overlay record beyond the
end of the input record, the reformatted record length is automatically
increased to that length, and blanks are filled in on the left as needed. For
variable-length records, the RDW length is also increased to correspond to the
larger reformatted record length after all of the OVERLAY items are processed.
For example, if your input record has a length of 40 and you specify:
INREC OVERLAY=(16:CABC,51:5C*,35:15,2)
the output record is given a length of 55. Blanks are filled in from columns
41-50. For variable-length records, the length in the RDW is changed from 40
to 55 after all of the OVERLAY items are processed.
Missing bytes in specified input fields are replaced with blanks so the padded
fields can be processed.
See INREC FIELDS for details of the items listed in the OVERLAY syntax
diagram previously. You can specify all of the items for OVERLAY in the same
way that you can specify them for BUILD or FIELDS with the following
exceptions:
v You cannot specify p or p,HEX or p,TRAN=keyword for OVERLAY.
v For p,m,H or p,m,F or p,m,D fields specified for OVERLAY, fields are
aligned as necessary without changing the preceding bytes.
v For variable-length records, you must not overlay positions 1-4 (the RDW)
for OVERLAY, so be sure to specify the first column (c:) as 5 or greater. If
you do not specify the first column, it will default to 1: which is invalid for
variable-length records with OVERLAY. Whereas FIELDS=(1,m,...) is
required, OVERLAY=(1,m) is not allowed, since it would overlay the RDW.
Sample Syntax:
Fixed input records:
INREC OVERLAY=(21:21,4,ZD,TO=PD,LENGTH=4,
2:5,8,HEX,45:C*,32,4,C*,81:SEQNUM,5,ZD)
Variable input records:

INREC OVERLAY=(5:X0001,28:CDate is ,YDDDNS=(4D),
17:5C*)
Default for OVERLAY: None; must be specified. See Appendix B,

FINDREP
142

FINDREP
IN=incon,OUT=outcon
,
IN=( incon
,
,OUT=outcon
INOUT=( incon,outcon
)

,
STARTPOS=p
ENDPOS=q
DO=n
MAXLEN=n
OVERRUN=ERROR
OVERRUN=TRUNC
SHIFT=YES
SHIFT=NO
You can use FINDREP to find constants anywhere in a record and replace
them with other constants of the same or different lengths. You can find
character or hexadecimal input constants anywhere in your records and replace
them with character, hexadecimal or null output constants. As appropriate,
bytes can be shifted left or right, blank padding can be added for fixed-length
records, and the length can be changed for variable-length records.
Various options of FINDREP allow you to define one or more input constants
and a corresponding output constant, define one or more pairs of input and
output constants, start and end the find scan at specified positions, stop after a
specified number of constants are replaced, increase or decrease the length of
the output record, define the action to be taken if nonblank characters overrun
the end of the record, and specify whether output constants are to replace or
overlay input constants.
See FINDREP under "OUTFIL Control Statements" for complete details.
Sample syntax
INREC FINDREP=(IN=CGoodbye,OUT=CBye)
Default for FINDREP: None; must be specified. See Appendix B,

IFTHEN
143

,
IFTHEN=(
WHEN=INIT
,PARSE=(definitions)
,BUILD=(items)
,OVERLAY=(items)
,FINDREP=(items)
WHEN=GROUP ,BEGIN=(logexp) ,PUSH=(items)
,KEYBEGIN=(p,m)
,END=(logexp)
,RECORDS=n
WHEN=(logexp) ,PARSE=(definitions)
,BUILD=(items)
,HIT=NEXT
,OVERLAY=(items)
,FINDREP=(items)
WHEN=ANY ,PARSE=(definitions)
,BUILD=(items)
,HIT=NEXT
,OVERLAY=(items)
,FINDREP=(items)
WHEN=NONE ,PARSE=(definitions)
,BUILD=(items)
,OVERLAY=(items)
,FINDREP=(items)
IFTHEN clauses allow you to reformat different records in different ways by

specifying how build, overlay, find/replace or group operation items are to be
applied to records that meet given criteria. IFTHEN clauses let you use simple
or complex conditional logic to choose how different record types are
reformatted.
If you want to insert, rearrange, or delete fields in the same way for every
record, use BUILD or FIELDS rather than IFTHEN. If you want to overlay
existing columns in the same way for every record, use OVERLAY rather than
IFTHEN. If you want to do find and replace operations in the same way for
every record, use FINDREP rather than IFTHEN. Use IFTHEN clauses if you
want to insert, rearrange, delete or overlay fields, or perform find/replace
operations in different ways for different records., or if you want to perform
operations on groups of records.
You can use five types of IFTHEN clauses as follows:
v WHEN=INIT: Use one or more WHEN=INIT clauses to apply build, overlay
or find/replace items to all of your input records. WHEN=INIT clauses and
WHEN=GROUP clauses are processed before any of the other IFTHEN
clauses.
v WHEN=GROUP: Use one or more WHEN=GROUP clauses to propagate
fields, identifiers and sequence numbers within specified groups of records.
WHEN=INIT clauses and WHEN=GROUP clauses are processed before any
of the other IFTHEN clauses.
v WHEN=(logexp): Use one or more WHEN=(logexp) clauses to apply build,
overlay or find/replace items to your input records that meet specified
criteria. A WHEN=(logexp) clause is satisfied when the logical expression
evaluates as true.
v WHEN=ANY: Use a WHEN=ANY clause after multiple WHEN=(logexp)
clauses to apply additional build, overlay or find/replace items to your
input records if they satisfied the criteria for any of the preceding
WHEN=(logexp) clauses.
v WHEN=NONE: Use one or more WHEN=NONE clauses to apply build,
overlay or find/replace items to your input records that did not meet the
criteria for any of the WHEN=(logexp) clauses. WHEN=NONE clauses are
processed after any of the other IFTHEN clauses. If you do not specify a
144

WHEN=NONE clause, only the WHEN=INIT and WHEN=GROUP changes
(if any) are applied to input records that do not meet the criteria for any of
the WHEN=(logexp) clauses.
IFTHEN clauses are processed in the following order:
v WHEN=INIT clauses and WHEN=GROUP clauses
v WHEN=(logexp) clauses and WHEN=ANY clauses
v WHEN=NONE clauses
Processing of IFTHEN clauses continues unless one of the following occurs:
v A WHEN=(logexp) or WHEN=ANY clause is satisfied, and HIT=NEXT is
not specified.
v There are no more IFTHEN clauses to process. When processing of IFTHEN
clauses stops, the IFTHEN record created so far is used as the output record.
Example:
INREC IFTHEN=(WHEN=(12,1,BI,ALL,X3F),OVERLAY=(18:CYes)),
IFTHEN=(WHEN=(35,2,PD,EQ,+14),BUILD=(1,40,45,3,HEX),HIT=NEXT),
IFTHEN=(WHEN=(35,2,PD,GT,+17),BUILD=(1,40,41,5,HEX),HIT=NEXT),
IFTHEN=(WHEN=ANY,BUILD=(1,55,CABC,70:X)),
IFTHEN=(WHEN=(63,2,CH,EQ,CAB),OVERLAY=(18:CNo)),
IFTHEN=(WHEN=NONE,BUILD=(1,40,51,8,TRAN=LTOU))
For this example, the IFTHEN clauses are processed as follows:

v If IFTHEN clause 1 is satisfied, its overlay item is applied and IFTHEN
processing stops.
v If IFTHEN clause 1 is not satisfied, its overlay item is not applied and
IFTHEN processing continues.
v If IFTHEN clause 2 is satisfied, its build items are applied and IFTHEN
processing continues.
v If IFTHEN clause 2 is not satisfied, its build items are not applied and
processing stops.
processing stops.
processing stops.
IFTHEN processing stops.
All of the IFTHEN clauses operate sequentially on an IFTHEN record. The
IFTHEN record is created initially from the input record. Each IFTHEN clause
tests and changes the IFTHEN record, as appropriate. Thus, changes made by
earlier IFTHEN clauses are "seen" by later IFTHEN clauses. For example, if
you have a 40-byte input record and specify:
145

INREC IFTHEN=(WHEN=INIT,OVERLAY=(8:8,4,ZD,ADD,+1,TO=ZD,LENGTH=4)),
IFTHEN=(WHEN=(8,4,ZD,EQ,+27),OVERLAY=(28:CYes)),
IFTHEN=(WHEN=NONE,OVERLAY=(28:CNo))
The WHEN=INIT clause adds 1 to the ZD value and stores it in the IFTHEN
record. The WHEN=(8,4,ZD,EQ,+27) clause tests the incremented ZD value in
the IFTHEN record rather than the original ZD value in the input record.
The IFTHEN record is adjusted as needed for the records created or changed
by the IFTHEN clauses. For fixed-length records, blanks are filled in on the left
as needed. For variable-length records, the RDW length is adjusted as needed
each time the IFTHEN record is changed.
DFSORT sets an appropriate LRECL (or reformatted record length if the
INREC record is further modified) for the output records based on the build,
overlay, find/replace and group operation items specified by the IFTHEN
clauses. However, DFSORT does not analyze the possible results of
WHEN=(logexp) conditions when determining an appropriate LRECL. When
you use INREC IFTHEN clauses, you can override the INREC LRECL
determined by DFSORT with the INREC IFOUTLEN parameter.
If SEQNUM is used in multiple IFTHEN clauses, the sequence number will be
incremented for each record that satisfies the IFTHEN clause, that is, a separate
SEQNUM counter will be kept for each IFTHEN clause. For example, if your
input is:
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
A
B
B
C
A
C
B
D
1
1
2
1
2
2
3
1
and you specify:

INREC IFTHEN=(WHEN=(8,1,CH,EQ,CA),OVERLAY=(15:SEQNUM,4,ZD)),
IFTHEN=(WHEN=(8,1,CH,EQ,CB),OVERLAY=(16:SEQNUM,4,ZD)),
IFTHEN=(WHEN=NONE,OVERLAY=(17:SEQNUM,4,ZD))
your output will be:

RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
A
B
B
C
A
C
B
D
1
1
2
1
2
2
3
1
0001
0001
0002
0001
0002
0002
0003
0003
Separate SEQNUM counters are kept for the 'A' record, for the 'B' record, and
for the NONE records.
WHEN=INIT clause
See "WHEN=INIT clause" under OUTFIL IFTHEN for details. Note that /
cannot be used to create blank records or new records.
Sample Syntax:
146

INREC IFTHEN=(WHEN=INIT,
BUILD=(1,20,21:CDepartment,31:3X,21,60)),
IFTHEN=(WHEN=(5,2,CH,EQ,CD1),OVERLAY=(31:8,3)),
IFTHEN=(WHEN=(5,2,CH,EQ,CD2),OVERLAY=(31:12,3))
WHEN=GROUP clause
See "WHEN=GROUP clause" under OUTFIL IFTHEN for details.
Sample Syntax:
INREC IFTHEN=(WHEN=GROUP,
BEGIN=(1,40,SS,EQ,CJ82,OR,1,40,SS,EQ,CM72),
PUSH=(41:ID=5))
WHEN=(logexp) clause
See "WHEN=(logexp) clause" under OUTFIL IFTHEN for details. Note that
although / can be used to create blank records and new records with OUTFIL,
it cannot be used with INREC.
Sample Syntax:
INREC IFTHEN=(WHEN=(1,3,CH,EQ,CT01,AND,
18,4,ZD,LE,+2000),OVERLAY=(42:CType1 <= 2000),HIT=NEXT),
IFTHEN=(WHEN=(1,3,CH,EQ,CT01,AND,6,1,BI,BO,X03),
BUILD=(1,21,42,13)),
IFTHEN=(WHEN=(1,3,CH,EQ,CT01,AND,
18,4,ZD,GT,+2000),OVERLAY=(42:CType1 > 2000 ),HIT=NEXT),
BUILD=(1,25,42,13))
WHEN=ANY clause
See "WHEN=ANY clause" under OUTFIL IFTHEN for details. Note that
Sample Syntax:
INREC IFTHEN=(WHEN=(1,3,SS,EQ,CT01,T02,T03),
BUILD=(CGroup A,X,1,80),HIT=NEXT),
IFTHEN=(WHEN=(1,3,SS,EQ,CT04,T05,T06),
BUILD=(CGroup B,X,1,80),HIT=NEXT),
IFTHEN=(WHEN=(1,3,SS,EQ,CT07,T08,T09,T10),
BUILD=(CGroup C,X,1,80),HIT=NEXT),
IFTHEN=(WHEN=ANY,OVERLAY=(16:CGroup Found))
WHEN=NONE clause
See "WHEN=NONE clause" under OUTFIL IFTHEN for details. Note that
Sample Syntax:
INREC IFTHEN=(WHEN=INIT,BUILD=(1,20,21:CDepartment,31:3X,21,60)),
IFTHEN=(WHEN=NONE,OVERLAY=(31:C***))
Default for IFTHEN clauses: None; must be specified. See Appendix B,

"Specification/Override of DFSORT Options".
Applicable Functions: See Appendix B, "Specification/Override of DFSORT
Options".
IFOUTLEN
147

IFOUTLEN=n
Overrides the INREC LRECL (or reformatted record length if the INREC record
is further modified) determined by DFSORT from your INREC IFTHEN
clauses. DFSORT sets an appropriate LRECL for the output records based on
the build, overlay, find/replace and group operation items specified by the
IFTHEN clauses. However, DFSORT does not analyze the possible results of
WHEN=(logexp) conditions when determining an appropriate INREC LRECL.
When you use INREC IFTHEN clauses, you can override the INREC LRECL
determined by DFSORT with the INREC IFOUTLEN parameter.
Fixed-length records longer than the IFOUTLEN length are truncated to the
IFOUTLEN length. Fixed-length records shorter than the IFOUTLEN are
padded with blanks to the IFOUTLEN length. Variable-length records longer
than the IFOUTLEN length are truncated to the IFOUTLEN length.
n
specifies the length to use for the INREC LRECL (or for the reformatted
record length if the INREC record is further modified) . The value for n
must be between 1 and 32767, but must not be larger than the maximum
LRECL allowed for the RECFM, and must not conflict with the specified or
retrieved LRECL for the fixed-length output data set.
Sample Syntax:
INREC IFOUTLEN=70,
IFTHEN=(WHEN=(5,1,CH,EQ,C1,AND,8,3,ZD,EQ,+10),
BUILD=(1,40,CT01-GROUP-A,65)),
BUILD=(1,40,CT02-GROUP-B,65))
Default for IFOUTLEN: The LRECL determined from the IFTHEN clauses.
INREC statement notes

v When INREC is specified, DFSORT reformats the input records after user exit
E15 or INCLUDE/OMIT statement processing is finished. Thus, references to
fields by your E15 user exit and INCLUDE/OMIT statements are not affected,
whereas your SORT, OUTREC, and SUM statements must refer to fields in the
reformatted input records. Your E35 user exit must refer to fields in the
reformatted output record.
v In general, OUTREC should be used rather than INREC so your SORT and SUM
statements can refer to fields in the original input records.
v If you use locale processing for SORT, MERGE, INCLUDE, or OMIT fields, you
must not use INREC. Use the OUTREC statement or the OUTFIL statement
instead of INREC.
v When you specify INREC, you must be aware of the change in record size and
layout of the resulting reformatted input records.
v Performance can be improved if you can significantly reduce the length of your
records with INREC. INREC and OUTREC should not be used unless they are
actually needed to reformat your records.
v For variable-length records, the first entry in the FIELDS, BUILD, or IFTHEN
BUILD parameter must specify or include the unedited 4-byte record descriptor
word (RDW), that is, the first field must be 1,4 or 1,m with m greater than 4.
DFSORT sets the length of the reformatted record in the RDW.
If the first field in the data portion of the input record is to appear unedited in
the reformatted record immediately following the RDW, the entry in the FIELDS,
BUILD, or IFTHEN BUILD parameter can specify both RDW and data field in
148

one (1,m,...). Otherwise, the RDW must be specifically included in the
reformatted record (for example, 1,4,1,4,HEX).
v For variable-length records, OVERLAY, IFTHEN OVERLAY or IFTHEN PUSH
items must not overlay the RDW in bytes 1-4. You must ensure that 1:, 2:, 3: or
4: is not specified or defaulted for any OVERLAY or PUSH item. Note that the
default for the first OVERLAY or PUSH item is 1:, so you must override it.
v If the SORTOUT LRECL is specified or available, DFSORT will use it even if it
does not match the reformatted INREC record length; this can cause padding or
truncation of the reformatted INREC records, or termination. If the SORTOUT
LRECL is not specified or available, DFSORT can automatically use the
reformatted INREC record length as the SORTOUT LRECL, when appropriate.
See the discussion of the SOLRF and NOSOLRF options in OPTION control
For VSAM data sets, the maximum record size defined in the cluster is
equivalent to the LRECL when processing fixed-length records, and is four bytes
less than the LRECL when processing variable-length records. See VSAM
considerations on page 15 for more information.
v With FIELDS, BUILD, or IFTHEN BUILD, the variable part of the input record
(that part beyond the minimum record length) can be included in the
reformatted record, and if included, must be the last part. In this case, a value
must be specified for pn that is less than or equal to the minimum record length
(see L4 of the RECORD control statement) plus 1 byte; mn and an must be
omitted. For example:
INREC FIELDS=(1,8,20C*,9)
With OVERLAY, the variable part of the input record must not be included in
the reformatted record.
v If INREC with FIELDS or BUILD and OUTREC with FIELDS and BUILD are
specified, either both must specify position-only for the last part, or neither must
specify position-only for the last part. For example:
INREC BUILD=(1,8,20C*,9)
OUTREC BUILD=(1,4,3Z,5)
or:
INREC FIELDS=(1,40,45,5)
OUTREC FIELDS=(1,45,C****)
OVERLAY or IFTHEN, and FIELDS or BUILD, can differ with respect to

position-only. For example:
INREC BUILD=(1,24,32:25)
OUTREC IFTHEN=(WHEN=(8,1,ZD,GT,+5),
BUILD=(1,24,25:CYes,28,10)),
IFTHEN=(WHEN=NONE,
BUILD=(1,24,25:CNo ,28,10))
or:
INREC FIELDS=(1,18,8C*,23)
OUTREC OVERLAY=(24:CA)
v If the reformatted record includes only the RDW and the variable part of the
input record, null records containing only an RDW could result.
v
v The input records are reformatted before processing, as specified by INREC. The
output records are in the format specified by INREC, unless OUTREC is also
specified.
149

v Fields referenced in INREC statements can overlap each other and control fields
or both.
v If input is variable records, the output is also variable. This means that each
record is given the correct RDW by DFSORT before output.
v
v If overflow might occur during summation, INREC can be used to create a
larger SUM field in the reformatted input record (perhaps resulting in a larger
record for sorting or merging) so that overflow does not occur. Example 5 on
page 455 illustrates this technique.
v DFSORT issues a message and terminates if an INREC statement is specified for
a tape work data set sort or conventional merge application.
v If SZERO is in effect, -0 is treated as negative and +0 is treated as positive for
edited or converted input fields, decimal constants, and the results of arithmetic
expressions. If NOSZERO is in effect, -0 and +0 are treated as positive for edited
or converted input fields, decimal constants, and the results of arithmetic
expressions.
Note: OPTION SZERO or OPTION NOSZERO is ignored for INREC
IFTHEN=(WHEN=(logexp),...) unless the OPTION statement is found in a higher
source (for example, DFSPARM is a higher source than SYSIN) or before the
INREC statement in the same source. For example, NOSZERO will be used in
both of the following cases:
Case 1:
//DFSPARM DD *
OPTION COPY,NOSZERO
/*
//SYSIN DD *
INREC IFTHEN=(WHEN=(1,2,FS,EQ,+0),OVERLAY=(22:CYes)),
IFTHEN=(WHEN=NONE,OVERLAY=(22:CNo ))
/*
Case 2:
//SYSIN DD *
OPTION COPY,NOSZERO
INREC IFTHEN=(WHEN=(1,2,FS,EQ,+0),OVERLAY=(28:CA)),
IFTHEN=(WHEN=NONE,OVERLAY=(28:CB))
/*
Reformatting records before processing examples

Both INREC and OUTREC can be used to reformat records, either separately or
together. The two control statements perform similar functions, although INREC
reformats records before sort, merge, or copy processing, whereas OUTREC
reformats records after sort, merge, or copy processing. This section shows INREC
examples. See Reformatting records after processing examples on page 426 for
OUTREC examples.
Example 1
INREC method:
INCLUDE COND=(5,1,GE,CM),FORMAT=CH
INREC FIELDS=(10,3,20,8,33,11,5,1)
SORT FIELDS=(4,8,CH,A,1,3,FI,A)
SUM FIELDS=(17,4,BI)
150

OUTREC method:
INCLUDE COND=(5,1,GE,CM),FORMAT=CH
OUTREC FIELDS=(10,3,20,8,33,11,5,1)
SORT FIELDS=(20,8,CH,A,10,3,FI,A)
SUM FIELDS=(38,4,BI)
These examples illustrate how a fixed-length input data set is sorted and
reformatted for output. Unnecessary fields are eliminated from the output records
using INREC or OUTREC. The SORTIN LRECL is 80.
Records are also included or excluded by means of the INCLUDE statement, and
summed by means of the SUM statement.
The reformatted input records are fixed length with a record size of 23 bytes.
SOLRF (the IBM-supplied default) is in effect, so unless the SORTOUT LRECL is
specified or available, it will automatically be set to the reformatted record length
of 23. The reformatted records look as follows after INREC or OUTREC processing:
Position
Contents
1-3
Input positions 10 through 12
4-11
12-22
23
Input position 5
Identical results are achieved with INREC or OUTREC. However, use of OUTREC
makes it easier to code the SORT and SUM statements. In either case, the
INCLUDE COND parameters must refer to the fields of the original input records.
However, with INREC, the SUM and SORT FIELDS parameters must refer to the
fields of the reformatted input records, while with OUTREC, the SUM and SORT
FIELDS parameters must refer to the fields of the original input records.
Example 2
INREC FIELDS=(1,35,2Z,36,45)
MERGE FIELDS=(20,4,CH,D,10,3,CH,D),FILES=3
SUM FIELDS=(36,4,BI,40,8,PD)
RECORD TYPE=F,LENGTH=(80,,82)
This example illustrates how overflow of a summary field can be prevented when
three fixed-length data sets are merged and reformatted for output. The input
record size is 80 bytes. To illustrate the use of the RECORD statement, assume that
SORTIN and SORTOUT are not present (that is, all input/output is handled by
user exits).
The reformatted input records are fixed-length with a record size of 82 bytes (an
insignificant increase from the original size of 80 bytes). They look as follows:
Position
Contents
1-35
36-37
Binary zeros (to prevent overflow)
38-82
151

The MERGE and SUM statements must refer to the fields of the reformatted input
records.
The reformatted output records are identical to the reformatted input records.
Thus, the 2-byte summary field at positions 36 and 37 in the original input records
expands to a 4-byte summary field in positions 36 through 39 of the reformatted
input/output record before merging. This prevents overflow of this summary field.
Restriction: If OUTREC were used instead of INREC, the records would be
reformatted after merging, and the 2-byte summary field might overflow.
Note: This method of preventing overflow cannot be used for negative FI summary
fields because padding with zeros rather than ones would change the sign.
Example 3
INREC BUILD=(20,4,12,3)
SORT FIELDS=(1,4,D,5,3,D),FORMAT=CH
OUTREC BUILD=(5X,1,4,H,19:1,2,5,3,DATE1(-),80XFF)
This example illustrates how a fixed-length input data set can be sorted and
reformatted for output. A more efficient sort is achieved by using INREC before
sorting to reduce the input records as much as possible, and using OUTREC after
sorting to add padding, the current date and repeated fields. The SORTIN LRECL
is 80 bytes.
The reformatted input records are fixed-length, and have a record size of seven
bytes (a significant reduction from the original size of 80 bytes). They look as
follows:
Position
Contents
1-4
5-7
The SORT and OUTREC statements must refer to the fields of the reformatted
input records.
The reformatted output records are fixed length with a record size of 113 bytes.
of 113. The reformatted output records look as follows:
Position
Contents
152
1-5
EBCDIC blanks
Binary zero (for H alignment)
7-10
11-18
EBCDIC blanks
19-20
21-23
24-33
The current date in the form C'yyyy-mm-dd'

34-113 Hexadecimal FF's
Thus, the use of INREC and OUTREC allows sorting of 7-byte records rather than
80-byte records, even though the output records are 113 bytes long.
Example 4
OPTION COPY,Y2PAST=1985
INREC FIELDS=(SEQNUM,4,BI,
8,5,ZD,TO=PD,
31,2,PD,TO=FI,LENGTH=2,
15,6,Y2TP,
25,3,CHANGE=(1,CL92,X01,CM72,X02,CJ42,X03),
NOMATCH=(XFF))
This example illustrates how a sequence number can be generated, how values in
one numeric or date format can be converted to another format, and how a lookup
table can be used.
The reformatted input records will look as follows:
Position
Contents
1-4
A binary sequence number that starts at 1 and increments by 1.
57
A PD field containing the converted ZD field from input positions 8

through 12.
89
An FI field containing the converted PD field from input positions 31

through 32.
1014
A P'yyyymmdd' date field containing the C'yymmdd' date field from input
positions 15-20 transformed according to the specified century window of
1985-2084.
15
A BI field containing X'01', X'02', X'03' or X'FF' as determined by using a

lookup table for the input field in positions 25-27.
The SORT statement can now refer to the sort field in the reformatted input
records. The OUTREC statement is used to restore the records to their original
format.
Example 5
INREC OVERLAY=(61:21,11,SFF,ADD,41,11,SFF,TO=PD,LENGTH=5)
SORT FIELDS=(61,5,PD,A)
OUTREC OVERLAY=(61:61,5,PD,EDIT=(SIII,IIT.TT),SIGNS=(+,-))
This example illustrates how you can use the OVERLAY parameter with INREC
and OUTREC to change certain columns in your records without affecting other
columns.
Positions 61-65 of the reformatted input records are overlaid with a 5-byte PD
value derived from adding the sddd,ddd.dd field at positions 21-31 to the
sddd,ddd.dd field at positions 41-51. The records are then sorted by this 5-byte PD
field. Positions 61-71 of the reformatted output records are overlaid with an
sddd,ddd.dd field derived from the 5-byte PD value. The data before positions
61-71 and after positions 61-71 are not affected
153
Example 6
OPTION COPY
INREC IFTHEN=(WHEN=(5,2,CH,EQ,CGP,AND,2,1,BI,EQ,+1),
BUILD=(1,6,16,20,CT1,X0003,1,7,20C1)),
IFTHEN=(WHEN=(5,2,CH,EQ,CGP,AND,2,1,BI,EQ,+2),
BUILD=(1,6,45,20,CT2,X0008,16,7,20C2)),
IFTHEN=(WHEN=(5,2,CH,EQ,CGP,AND,2,1,BI,EQ,+3),
BUILD=(1,6,31,20,CT3,X0005,25,7,20C3)),
IFTHEN=(WHEN=NONE,OVERLAY=(27:C00,XFFFF)),
IFOUTLEN=57
This example illustrates how you can use IFTHEN clauses with INREC to reformat
different records in different ways. IFOUTLEN=57 is used to set the reformatted
record length to 57.
Records with 'GP' in positions 5-6 and X'01' in position 2 are reformatted as
follows:
Position
Contents
1-6
Input positions 1-6
7-26
Input positions 16-35
27-28
'T1'
29-30
X'0003'
31-37
Input positions 1-7
38-57
20 '1's
follows:
Position
Contents
1-6
Input positions 1-6
7-26
27-28
'T2'
29-30
X'0008'
31-37
38-57
20 '2's
follows:
Position
Contents
154
1-6
Input positions 1-6
7-26
27-28
'T3'
29-30
X'0005'
31-37
38-57
20 '3's

Records without 'GP' in positions 5-6 or without X'01', X'02', or X'03' in position 2
are reformatted as follows:
Position
Contents
1-26
27-28
'00'
29-30
X'FFFF'
31-57
Example 7
INREC OVERLAY=(16:1,15,JFY=(SHIFT=LEFT))
OUTREC BUILD=(1,15)
This example illustrates how you can left-justify characters in an input field so
they can be sorted without regard to the leading blanks.
The 15-byte input records might look like this:
CARRIE
VICKY
FRANK
SAM
DAVID
MARTIN
Note that if we sort these records using just this control statement:
The 15-byte output records are as follows:

FRANK
VICKY
SAM
MARTIN
CARRIE
DAVID
Because of the different number of leading blanks in the input records, we don't
get what we want. To fix that, while keeping the leading blanks in the original
records, we use the JFY function of INREC to make a left-justified copy of the
15-byte input field at positions 16-30, SORT on it and use OUTREC to remove the
left-justified field. With the INREC, SORT and OUTREC control statements shown
previously, the output records are:
CARRIE
DAVID
FRANK
MARTIN
SAM
VICKY
If we wanted the output to contain the sorted left-justified fields, we could use
these control statements:
INREC BUILD=(1,15,JFY=(SHIFT=LEFT))
The output records would then be:

155

CARRIE
DAVID
FRANK
MARTIN
SAM
VICKY
Example 8
INREC PARSE=(%00=(ENDBEFR=C,,FIXLEN=11),
%01=(ENDBEFR=C,,FIXLEN=5),
%02=(FIXLEN=6)),
OVERLAY=(31:%00,42:%01,47:%02)
SORT FIELDS=(31,11,CH,A,42,5,UFF,A,47,6,SFF,D)
OUTREC BUILD=(1,30)
This example illustrates how you can sort FB input records with variable
position/length fields, such as comma separated values.
Marketing,96218,+27365
Development,3807,+1275
Research,7283,+5001
Development,1700,-5316
Research,978,+13562
Research,7283,+5002
Marketing,52,-8736
Marketing,52,+1603
Research,16072,-2022
We want to sort the first field as character ascending, the second field as unsigned
numeric ascending and the third field as signed numeric descending.
Note that each record has three variable fields in comma separated value format.
The fields do not start and end in the same position in every record and do not
have the same length in every record. The first and second fields end with a
comma and the third field ends with a blank.
In order to sort variable fields like these, we use the PARSE and OVERLAY
functions of INREC to create a fixed parsed copy of each variable field. We use
%00 to create an 11-byte fixed parsed field into which we extract the value before
the first comma. We use %01 to create a 5-byte fixed parsed field into which we
extract the value after the first comma and before the second comma. We use %02
to create a 6-byte fixed parsed field into which we extract the value after the
second comma. Then we SORT on the fixed parsed fields. Finally, we use OUTREC
to remove the fixed parsed fields.
After the INREC statement is processed, the records look like this:
Research,7283,+5001
Research,978,+13562
Research,7283,+5002
Marketing,52,-8736
Marketing,52,+1603
156
Marketing 96218+27365
Development3807 +1275
Research
7283 +5001
Development1700 -5316
Research
978 +13562
Development3807 -158
Research
7283 +5002
Marketing 52
-8736
Development5781 +2736
Marketing 52
+1603
Research
16072-2022

Since the second fixed parsed field is unsigned and left-justified, we sort it with
the UFF format. Since the third fixed parsed field is signed and left-justified, we
sort it with the SFF format.
After the SORT and OUTREC statements are processed, the output records look
like this:
Marketing,52,+1603
Marketing,52,-8736
Research,978,+13562
Research,7283,+5002
Research,7283,+5001
Example 9
%02=(FIXLEN=6)),
BUILD=(1,4,5:%00,16:%01,21:%02,27:5)
SORT FIELDS=(5,11,CH,A,16,5,UFF,A,21,6,SFF,D)
OUTREC BUILD=(1,4,27)
This example illustrates how you can sort VB input records with variable
position/length fields, such as comma separated values. This example is very
similar to the previous example for FB records, except that with VB records we
need to copy the fixed parsed fields after the 4-byte RDW rather than at the end of
the records.
The VB input records might look like this:
Length|Data
26|Marketing,96218,+27365
26|Development,3807,+1275
23|Research,7283,+5001
26|Development,1700,-5316
23|Research,978,+13562
23|Research,7283,+5002
22|Marketing,52,-8736
22|Marketing,52,+1603
24|Research,16072,-2022
After the INREC statement is processed, the records look like this:
Length|Data
48|Marketing 96218+27365Marketing,96218,+27365
48|Development3807 +1275 Development,3807,+1275
45|Research
7283 +5001 Research,7283,+5001
48|Development1700 -5316 Development,1700,-5316
45|Research
978 +13562Research,978,+13562
47|Development3807 -158 Development,3807,-158
45|Research
7283 +5002 Research,7283,+5002
44|Marketing 52
-8736 Marketing,52,-8736
48|Development5781 +2736 Development,5781,+2736
44|Marketing 52
+1603 Marketing,52,+1603
46|Research
16072-2022 Research,16072,-2022
After the SORT and OUTREC statements are processed, the output records look
like this:
157

Length|Data
22|Marketing,52,+1603
22|Marketing,52,-8736
26|Marketing,96218,+27365
23|Research,978,+13562
23|Research,7283,+5002
23|Research,7283,+5001
24|Research,16072,-2022
Example 10
OPTION COPY
INREC FINDREP=(IN=(X00,XFF),OUT=C)
This example illustrates how you can remove characters from FB or VB records.
The VB input records might look like this in hexadecimal:
RDW----|Data
000F0000D1E4D5C500C1D7D9C9D3FF
00100000C2C5E3E3E800C4C1C9E2E8FF
We want to remove each X'00' and X'FF' character. We use IN=(X'00',X'FF') to

indicate we want to find each X'00' and X'FF' character, and OUT=C'' (null) to
indicate we want to remove it.
The output records look like this:
RDW----|Data
000D0000D1E4D5C5C1D7D9C9D3
000E0000C2C5E3E3E8C4C1C9E2E8
Note that the X'00' and X'FF' characters have been removed and the RDW length
decreased accordingly. For VB input records, FINDREP processing automatically
starts at position 5 after the RDW so the X'00' characters in the RDW are not
affected.
Example 11
OPTION COPY
INREC FINDREP=(IN=CBALANCE,
OUT=CBALANCE 1000,SHIFT=NO,DO=1)
This example illustrates how you can find a value in FB or VB records and overlay
it with a larger value without shifting other bytes in the records.
The FB input records might look like this:
CUSTOMER1 10100
MNTHLY STMT BALANCE 2000
CUSTOMER2 11100
MNTHLY STMT REQUIRES NO MODIFICATION ACCT BALANCE 5000
CUSTOMER3 11111
YOUR INFO ENCLOSED
MNTHLY STMT REQUIRES MODIFICATION ACCT BALANCE 7000
We want to replace every instance of 'BALANCE dddd' with 'BALANCE 1000'

where dddd can be any value. We use IN=C'BALANCE' to indicate we want to
find each instance of 'BALANCE'. We use OUT='BALANCE 1000' to indicate we
want to replace it with 'BALANCE 1000'. We use SHIFT=NO to do an overlay,
overriding the default of shifting bytes. Since we only have at most one instance of
158

'BALANCE dddd' in a record, we can use DO=1 to stop processing a record after
one replacement of 'BALANCE dddd' with 'BALANCE 1000'. In this case, DO=1 is
more efficient than the default of continuing to look for more instances of
'BALANCE dddd' after the first instance. We would get the same result without
DO=1, just less efficiently
The output records look like this:
CUSTOMER1 10100
MNTHLY STMT BALANCE 1000
CUSTOMER2 11100
MNTHLY STMT REQUIRES NO MODIFICATION ACCT BALANCE 1000
CUSTOMER3 11111
YOUR INFO ENCLOSED
MNTHLY STMT REQUIRES MODIFICATION ACCT BALANCE 1000
Example 12
INREC IFTHEN=(WHEN=GROUP,BEGIN=(2,4,CH,EQ,CRPT.),
PUSH=(31:6,8))
OPTION EQUALS
OUTFIL INCLUDE=(31,8,CH,EQ,CFRANK,OR,
31,8,CH,EQ,CSRIHARI),BUILD=(1,30)
This example illustrates how you can SORT and INCLUDE groups of FB records
depending on a value in the first record of each group. We propagate the value in
the first record of the group to every record of the group, SORT and INCLUDE on
the value, and then remove it.
The 30-byte FBA input records might look like this:
1RPT.SRIHARI
LINE 1 FOR
LINE 2 FOR
...
1RPT.VICKY
LINE 1 FOR
LINE 2 FOR
...
1RPT.FRANK
LINE 1 FOR
LINE 2 FOR
...
1RPT.DAVID
LINE 1 FOR
LINE 2 FOR
...
REPORT 1
REPORT 1
REPORT 2
REPORT 2
REPORT 3
REPORT 3
REPORT 4
REPORT 4
Each report starts with 'RPT.reptname' in positions 2-13. In the output data set we
only want to include records for reports with specific reptname values, and the
reptname values we want can change from run to run. We also want to sort by the
reptname values in ascending order. For this example, let's say we just want the
SRIHARI and FRANK reports.
We use an IFTHEN WHEN=GROUP clause to propagate the reptname value to
each record of the group. BEGIN indicates a group starts with 'RPT.' in positions
2-5. PUSH overlays the reptname value from the first record of the group (the
'RPT.reptname' record) at positions 31-38 (after the end of the record) in each
record of the group including the first. After the IFTHEN GROUP clause is
executed, the intermediate records look like this:
159

1RPT.SRIHARI
LINE 1 FOR
LINE 2 FOR
...
1RPT.VICKY
LINE 1 FOR
LINE 2 FOR
...
1RPT.FRANK
LINE 1 FOR
LINE 2 FOR
...
1RPT.DAVID
LINE 1 FOR
LINE 2 FOR
...
REPORT 1
REPORT 1
REPORT 2
REPORT 2
REPORT 3
REPORT 3
REPORT 4
REPORT 4
SRIHARI
SRIHARI
SRIHARI
SRIHARI
VICKY
VICKY
VICKY
VICKY
FRANK
FRANK
FRANK
FRANK
DAVID
DAVID
DAVID
DAVID
Note that the records of each group have the reptname value from the first record
of that group in positions 31-38.
We use a SORT statement to sort ascending on the reptname in positions 31-38. We
use the EQUALS option to ensure that records in the same group (that is, with the
same reptname value) are kept in their original order. After the SORT statement is
1RPT.DAVID
LINE 1 FOR
LINE 2 FOR
...
1RPT.FRANK
LINE 1 FOR
LINE 2 FOR
...
1RPT.SRIHARI
LINE 1 FOR
LINE 2 FOR
...
1RPT.VICKY
LINE 1 FOR
LINE 2 FOR
...
REPORT 4
REPORT 4
REPORT 3
REPORT 3
REPORT 1
REPORT 1
REPORT 2
REPORT 2
DAVID
DAVID
DAVID
DAVID
FRANK
FRANK
FRANK
FRANK
SRIHARI
SRIHARI
SRIHARI
SRIHARI
VICKY
VICKY
VICKY
VICKY
We use an OUTFIL statement to only INCLUDE the records with a a reptname of

FRANK or SRIHARI in positions 31-38, and to remove the reptname from positions
31-38 so the included output records will be identical to the input records. After
the OUTFIL statement is executed, the final output records look like this:
1RPT.FRANK
LINE 1 FOR
LINE 2 FOR
...
1RPT.SRIHARI
LINE 1 FOR
LINE 2 FOR
...
REPORT 3
REPORT 3
REPORT 1
REPORT 1
Example 13
INREC BUILD=(1,6,Y2W,TOJUL=Y4T,X,
1,6,Y2W,WEEKDAY=CHAR3,X,
9,7,Y4T,TOGREG=Y4T(/),X,
9,7,Y4T,WEEKDAY=DIGIT1)
160

This example illustrates how to convert an mmddyy date to a ccyyddd date and a
3-character weekday string, and how to convert a ccyyddd date to a ccyy/mm/dd
date and 1-digit weekday string.
The input records might be as follows:
120409
051895
999999
013099
1999014
2003235
0000000
1992343
The output records would be as follows:

2009338
2095138
9999999
1999030
FRI
WED
999
SAT
1999/01/14
2003/08/23
0000/00/00
1992/12/08
5
7
0
3
The Y2PAST=1996 option sets the century window to 1996-2095. The century
window is used to transform yy in the Y2W field to ccyy.
Note that date conversion is not performed for the special indicators (all 9s and all
0s); the special indicator is just used appropriately for the output date field.
Example 14
OPTION COPY
INREC OVERLAY=(11:1,8,Y4T,ADDDAYS,+50,TOGREG=Y4T,
21:1,8,Y4T,SUBMONS,+7,TOGREG=Y4T,
31:1,8,Y4T,ADDYEARS,+2,TOGREG=Y4T)
OUTFIL REMOVECC,
HEADER1=(Input
+50 days -7 months +2 years)
This example illustrates how you can add and subtract days, months and years
from date fields.
The SORTIN data set has these input records with a C'ccyymmdd' date field in
positions 1-8:
20070305
20071213
20080219
20080901
20091122
20090115
20100915
20100630
99999999
The INREC statement performs three different date field arithmetic operations on
the input date field. It adds 50 days, subtracts 7 months and adds 2 years. We use
1,8,Y4T for the input field to match the C'ccyymmdd' input date, and we use
TOGREG=Y4T to give us a C'ccyymmdd' output date. The OUTFIL statement adds
headings.
SORTOUT will have these records:
Input
+50 days
20070305 20070424
20071213 20080201
20080219 20080409
20080901 20081021
20091122 20100111
-7 months
20060805
20070513
20070719
20080201
20090422
+2 years
20090305
20091213
20100219
20100901
20111122
161

20090115
20100915
20100630
99999999
20090306
20101104
20100819
99999999
20080615
20100215
20091130
99999999
20110115
20120915
20120630
99999999
Note that the '99999999' input value is treated as a special indicator for output.
JOINKEYS control statement

Used only for a JOINKEYS application. See Chapter 4, Using a JOINKEYS
application for joining two files, on page 457 for details.
JOIN control statement

application for joining two files, on page 457for details.
MERGE control statement

,
MERGE FIELDS=
p,m,f,s
,
p,m,
FORMAT=f
f,
COPY

,
,
EQUALS
NOEQUALS
FILES=n
FILSZ=x
SIZE=y
SKIPREC=z
STOPAFT=n
Y2PAST=
s
f
The MERGE control statement must be used when a merge operation is to be

performed; this statement describes the control fields in the input records on which
the input data sets have previously been sorted.
A MERGE statement can also be used to specify a copy application. User labels
will not be copied to the output data sets.
You can merge up to 100 data sets with Blockset merge or up to 16 data sets with
Conventional merge. If Blockset merge is not selected, you can use a SORTDIAG
DD statement to force message ICE800I, which gives a code indicating why
Blockset could not be used.
162
MERGE Control Statement

The way in which DFSORT processes short MERGE control fields depends on the
setting for VLSHRT/NOVLSHRT. A short field is one where the variable-length
record is too short to contain the entire field, that is, the field extends beyond the
record. For details about merging short records, see the discussion of the VLSHRT
and NOVLSHRT options in OPTION control statement on page 173.
The options available on the MERGE statement can be specified in other sources as
well. A table showing all possible sources for these options and the order of
override are given in Appendix B, Specification/override of DFSORT options, on
page 863. When an option can be specified on either the MERGE or OPTION
statement, it is preferable to specify it on the OPTION statement.
DFSORT accepts but does not process the following MERGE operands:
WORK=value and ORDER=value.
The active locale's collating rules affect MERGE processing as follows:
v DFSORT produces merged records for output according to the collating rules
defined in the active locale. This provides merging for single- or multi-byte
character data, based on defined collating rules that retain the cultural and local
characteristics of a language.
character (CH) control fields.
173.
Note: For a merge application, records deleted during an E35 exit routine are not
sequence checked. If you use an E35 exit routine without an output data set,
sequence checking is not performed at the time the records are passed to the E35
user exit; therefore, you must ensure that input records are in correct sequence.
FIELDS
,
FIELDS= (
p,m,f,s
Is written exactly the same way for a merge as it is for a sort. The meanings of
p, m, f, and s are described in the discussion of the SORT statement. The
defaults for this and the following parameters are also given there. See SORT
FIELDS=COPY
FIELDS=COPY
See the discussion of the COPY option on the OPTION statement, in OPTION
FORMAT=f
163

FORMAT=f
See the discussion of the FORMAT option in SORT control statement on

page 442. Used the same way for a merge as for a sort.
EQUALS or NOEQUALS
EQUALS
NOEQUALS
See the discussion of these options on the OPTION statement, in OPTION

FILES=n
FILES=n
Specifies the number of input files for a merge when input is supplied through
the E32 exit.
Default: None; must be specified when an E32 exit is used.
FILSZ or SIZE
FILSZ=x
SIZE=y
See the discussion of these options on the OPTION statement, in OPTION

SKIPREC
SKIPREC=z
See the discussion of this option on the OPTION statement, in OPTION

Note: SKIPREC is used for a copy or sort application, but is not used for a
merge application.
STOPAFT
STOPAFT=n

164

Note: STOPAFT is used for a copy or sort application, but is not used for a
merge application.
Y2PAST
Y2PAST=
s
f

Note: CENTURY=value and CENTWIN=value can be used instead of
Y2PAST=value.
Specifying a MERGE or COPYexamples

Example 1
MERGE FIELDS=(2,5,CH,A),FILSZ=29483
FIELDS
The control field begins on byte 2 of each record in the input data sets. The
field is 5 bytes long and contains character (EBCDIC) data that has been
presorted in ascending order.
FILSZ
The input data sets contain exactly 29483 records.
Example 2
MERGE FIELDS=(3,8,ZD,E,40,6,CH,D)
FIELDS
The major control field begins on byte 3 of each record, is 8 bytes long, and
contains zoned decimal data that is modified by your routine before the merge
examines it.
The second control field begins on byte 40, is 6 bytes long, and contains
character data in descending order.
Example 3
MERGE FIELDS=(25,4,A,48,8,A),FORMAT=ZD
FIELDS
The major control field begins on byte 25 of each record, is 4 bytes long, and
contains zoned decimal data that has been placed in ascending sequence.
The second control field begins on byte 48, is 8 bytes long, is also in zoned
decimal format, and is also in ascending sequence. The FORMAT parameter is
used to indicate that both fields have ZD format.
Example 4
MERGE FIELDS=COPY
FIELDS
The input data set is copied to output. No merge takes place.
165
MODS Control Statement
MODS control statement

,
MODS exit= (
n,m
)
,

HILEVEL=YES
,e
The MODS statement is needed only when DFSORT passes control to your
routines at user exits. The MODS statement associates user routines with specific
DFSORT exits and provides DFSORT with descriptions of these routines. For
details about DFSORT user exits and how user routines can be used, see Chapter 5,
Using your own user exit routines, on page 487.
To use one of the user exits, you substitute its three-character name (for example,
E31) for the word exit in the MODS statement format shown previously in this
section. You can specify any valid user exit, except E32. (E32 can be used only in a
merge operation invoked from a program; its address must be passed in a
parameter list.)
exit
exit= (
n,m
,
s
,e
The values that follow the exit parameter describe the user routine. These
values are:
n
specifies the name of your routine (member name if your routine is in a

library). You can use any valid operating system name for your routine. This
allows you to keep several alternative routines with different names in the
same library.
specifies the number of bytes of main storage your routine uses.Include storage
obtained (via GETMAIN) by your routine (or, for example, by OPEN) and the
storage required to load the COBOL library subroutines.
specifies either the name of the DD statement in your DFSORT job step that
defines the library in which your routine is located or SYSIN if your routine is
in the input stream. SYSIN is not valid for copy processing.
If a value is not specified for s, DFSORT uses the following search order to
find the library in which your routine is located:
1. The libraries identified by the STEPLIB DD statement
2. The libraries identified by the JOBLIB DD statement (if there is no STEPLIB
DD statement)
3. The link library.
specifies the linkage editor requirements of your routine or indicates that your
routine is written in COBOL. The following values are allowed:
N
166
specifies that your routine has already been bound or link-edited and can
be used in the DFSORT run without further binding or link-editing. This is

the default for e. N (specified or defaulted) can be overridden by the EXEC
PARM parameters 'E15=COB' and 'E35=COB' or by the HILEVEL=YES
parameter.
N64
specifies that your routine will use a 64-bit exit parameter list, has already
been bound or link-edited and can be used in the DFSORT run without
further binding or link-editing.
Note: N64 is only allowed when DFSORT is invoked from a program with
the 64-bit invocation parameter list.
C
specifies that your E15 or E35 routine is written in COBOL. If you code C
for any other exit, it is ignored, and N is assumed. Your COBOL-written
routine must already have been link-edited. The COBEXIT option of the
OPTION statement specifies the library for the COBOL exits.
specifies that your routine must be bound or link-edited together with

other routines to be used in the same phase (for example, E1n routines) of
DFSORT. See Dynamically binding or link-editing user exit routines on
page 498 for additional information. This value is not valid for copy
processing.
specifies that your routine requires binding or link-editing but that it must
be bound or link-edited separately from the other routines (for example,
E3n routines) to be used in a particular phase of DFSORT. E11 and E31 exit
routines are the only routines eligible for separate binding or link-editing.
See Dynamically binding or link-editing user exit routines on page 498
for additional information. This value is not valid for copy processing.
If you do not specify a value for e, N is assumed.
HILEVEL=YES
HILEVEL=YES
specifies that:
v if an E15 routine is identified on the MODS statement, it is written in
COBOL
v if an E35 routine is identified on the MODS statement, it is written in
COBOL.
If you identify an E15 routine and an E35 routine on the MODS statement,
specify HILEVEL=YES only if both routines are written in COBOL. If you do
not identify an E15 or E35 routine on the MODS statement, HILEVEL=YES is
ignored.
Note: COBOL=YES can be used instead of HILEVEL=YES.
Note:
1. The s parameter must be the same or omitted for each routine with N, N64 or
C for the e parameter (library concatenation is allowed). These routines cannot
be placed in SYSIN. Each such routine must be a load module.
2. Each routine for which T or S is specified for the e parameter can be placed in
any library or in SYSIN; they do not all have to be in the same library or
SYSIN (but can be). Some routines can even be in different libraries (or the
167

same library) and the rest can be in SYSIN. Each such routine, if in a library,
can be either an object deck or a load module; if in SYSIN, it must be an
object deck.
3. If the same routine is used in both input (that is, E1n routines) and output
(that is, E3n routines) DFSORT program phases, a separate copy of the routine
must be provided for each exit.
4. HILEVEL=YES can be used instead of C as the fourth parameter, to indicate
that an E15 or E35 routine is written in COBOL. In this case, if T is specified
as the fourth parameter for E15 or E35, DFSORT terminates. If you identify an
E15 routine and an E35 routine on the MODS statement, specify
HILEVEL=YES only if both routines are written in COBOL.
5. EXEC PARM parameter E15=COB can be used instead of C as the fourth
parameter, to indicate that an E15 is written in COBOL. In this case, if T is
specified as the fourth parameter for E15, DFSORT terminates.
6. EXEC PARM parameter E35=COB can be used instead of C as the fourth
parameter, to indicate that an E35 is written in COBOL. In this case, if T is
specified as the fourth parameter for E35, DFSORT terminates.
7. If HILEVEL=YES, E15=COB, or E35=COB is used instead of C as the fourth
parameter, to indicate that an exit is written in COBOL, the fourth parameter
for that exit must be specified as N or not specified.
8. If a COBOL E15 or E35 is specified for a conventional merge or tape work
data set sort, DFSORT terminates.
9. exit=(n,m) can be used to omit both the s and e parameters.
10. exit=(n,m,,e) can be used to omit the s parameter, but not the parameter.
11. The s parameter must be specified for a conventional merge or tape work data
set sort, or when S or T is specified for the e parameter.
Default: None; must be specified if you use exit routines. N is the default for the
fourth parameter.
Applicable Functions: See Appendix B, Specification/override of DFSORT options,
on page 863.
For information on user exit routines in SYSIN, see System DD statements on
page 63.
For details on how to design your routines, refer to Summary of rules for user
exit routines on page 495.
When you are preparing your MODS statement,remember that DFSORT must
know the amount of main storage your routine needs so that it can allocate main
storage properly for its own use. If you do not know the exact number of bytes
your program requires (including requirements for system services), make a
slightly high estimate. The value of m in the MODS statement is written the same
way whether it is an exact figure or an estimate: you do not precede the value by
E for an estimate.
Identifying user exit routinesexamples

Example 1
MODS E15=(ADDREC,552,MODLIB),E35=(ALTREC,11032,MODLIB)
E15
At exit E15, DFSORT transfers control to your own routine. Your routine is in
168

the library defined by a job control statement with the ddname MODLIB. Its
member name is ADDREC and uses 552 bytes.
E35
At exit E35, DFSORT transfers control to your routine. Your routine is in the
library defined by the job control statement with the ddname MODLIB. Its
member name is ALTREC and will use 11032 bytes.
Example 2
MODS
E15=(COBOLE15,7000,,C),
E35=(COBOLE35,7000,EXITC,C)
E15
At exit E15, DFSORT transfers control to your own routine. Your routine is
written in COBOL and is in the STEPLIB/JOBLIB or link libraries. Its member
name is COBOLE15 and it uses 7000 bytes.
E35
At exit E35, DFSORT transfers control to your routine. Your routine is written
in COBOL and is in the library defined by the job control statement with the
ddname EXITC. Its member name is COBOLE35 and it uses 7000 bytes.
OMIT control statement

OMIT COND=

FORMAT=f
ALL
NONE
Use an OMIT statement if you do not want all of the input records to appear in
the output data sets. The OMIT statement selects the records you do not want to
include.
You can specify either an INCLUDE statement or an OMIT statement in the same
DFSORT run, but not both.
depends on the settings for VLSCMP/NOVLSCMP and VLSHRT/NOVLSHRT. A
short field is one where the variable-length record is too short to contain the entire
field, that is, the field extends beyond the record. For details about including or
omitting short records, see the discussion of the VLSCMP and NOVLSCMP options
A logical expression is one or more relational conditions logically combined, based
on fields in the input record, and can be represented at a high level as follows:
relational condition1

.
,
AND
OR
,relational condition2
If the logical expression is true for a given record, the record is omitted from the
output data set.
169
OMIT Control Statement

Five types of relational conditions can be used as follows:
1.
2. Comparisons:
Compare two compare fields or a compare field and a decimal, hexadecimal,
character or current, future, or past date constant.
For example, you can compare the first 6 bytes of each record with its last 6
bytes, and omit those records in which those fields are identical. Or you can
compare a date field with today's date, yesterday's date, or tomorrow's date
and omit those records accordingly.
3. Substring Comparison Tests:
Search for a constant within a field value or a field value within a constant.
For example, you can search the value in a 6-byte field for the character
constant C'OK', and omit those records for which C'OK' is found somewhere in
the field. Or you can search the character constant C'J69,L92,J82' for the value
in a 3-byte field, and omit those records for which C'J69', C'L92', or C'J82'
appears in the field.
4. Bit Logic Tests:
Test the state (on or off) of selected bits in a binary field using a bit or
hexadecimal mask or a bit constant.
For example, you can omit those records which have bits 0 and 2 on in a 1-byte
field. Or you can omit those records which have bits 3 and 12 on and bits 6
and 8 off in a 2-byte field.
5. Date Comparisons:
Compare a two-digit year date field to a two-digit year date constant, the
current two-digit year date or another two-digit year date field, using the
century window in effect.
For example, you can omit only those records for which a Z'yymm' date field is
between January 1996 and March 2005. Or you can omit only those records for
which a P'dddyy' field is less than another P'dddyy' field.
6. Numeric Tests:
Test a field for numerics or non-numerics in character, zoned decimal, or
For example, you can omit only those records in which a 5-byte field contains
only '0'-'9' characters (that is, numerics). Or you can omit only those records in
which a 9-byte field contains invalid ZD numeric data (that is, non-numerics).
Or you can omit only those records in which a 12-byte field contains valid PD
numeric data (that is, numerics).
7. Alphanumeric Tests:
Test a field for alphanumerics or non-alphanumerics in character format.
Various combinations of uppercase characters (A-Z), lowercase characters (a-z)
and numeric characters (0-9) can be used.
For example, you can omit only those records in which a 10-byte field contains
only 'A'-'Z' characters (that is, uppercase characters) or '0'-'9' characters (that is,
numeric characters). Or you can omit only those records in which a 20-byte
field contains characters other than 'a'-'z' (that is, lowercase characters).
For complete details on the parameters of the OMIT control statement, see
INCLUDE control statement on page 96.
The OMIT control statement differs from the OMIT parameter of the OUTFIL
statement in the following ways:
170

v The OMIT statement applies to all input records; the OMIT parameter applies
only to the OUTFIL input records for its OUTFIL group.
v FORMAT=f can be specified with the OMIT statement but not with the OMIT
parameter. Thus, you can use FORMAT=f and p,m or p,m,f fields with the OMIT
statement, but you must only use p,m,f fields with the OMIT parameter. For
example:
OMIT FORMAT=BI,
COND=(5,4,LT,11,4,OR,21,4,EQ,31,4,OR,
61,20,SS,EQ,CFLY)
OUTFIL OMIT=(5,4,BI,LT,11,4,BI,OR,21,4,BI,EQ,31,4,BI,OR,
61,20,SS,EQ,CFLY)
v D2 format can be specified with the OMIT statement but not with the OMIT
parameter.
See OUTFIL control statements on page 223 for more details on the OUTFIL
OMIT parameter.
COND
COND=
ALL
NONE
logical expression
specifies one or more relational conditions logically combined, based
on fields in the input record. If the logical expression is true for a
given record, the record is omitted from the output data sets.
ALL
specifies that all of the input records are to be omitted from the output
data sets.
NONE
specifies that none of the input records are to be omitted from the
output data sets.
Default: NONE. See Appendix B, Specification/override of DFSORT options,
on page 863 for full override details.
FORMAT
FORMAT=f
For details on this parameter, see INCLUDE control statement on page 96.
Omitting records from the output data setexample

Example
OMIT COND=(27,1,CH,EQ,CD,&,
(22,2,BI,SOME,XC008,|,
28,1,BI,EQ,B.1....01))
This statement omits records in which:

AND
171

v Bytes 22 through 23 have some, but not all of bits 0, 1 and 12 on OR byte 28 is
equal to the specified pattern of bit 1 on, bit 6 off and bit 7 on.
Note that the AND and OR operators can be written with the AND and OR signs,
and that parentheses are used to change the order in which AND and OR are
evaluated.
For additional examples of logical expressions, see INCLUDE control statement
on page 96.
172
OPTION Control Statement
OPTION control statement

,
OPTION
ARESALL=
ARESINV=
AVGRLEN=n
CHALT
NOCHALT
CHECK
NOCHECK
CINV
NOCINV
CKPT
COBEXIT=
COPY
DSA=n
DSPSIZE=
n
nK
nM
n
nK
nM
COB1
COB2
MAX
n
DYNALLOC
=
DYNAPCT=
d
(,n)
(d,n)
OFF
x
OLD
DYNSPC=n
EFS=
name
NONE
EQUALS
NOEQUALS
EXITCK=
STRONG
WEAK
FILSZ=
x
Ex
Ux
SIZE=
y
Ey
Uy
HIPRMAX=
OPTIMAL
n
p%
LIST
NOLIST
LISTX
NOLISTX
LOCALE=
name
CURRENT
NONE
173

MAINSIZE=
MERGEIN=
MOSIZE=
n
nK
nM
MAX
ddname
,
( ddname
MAX
n
p%
MOWRK
NOMOWRK
MSGDDN=ddname
MSGPRT=
ALL
NONE
CRITICAL
NOBLKSET
NOOUTREL
NOOUTSEC
NULLOUT=
RC0
RC4
RC16
ODMAXBF=
n
nK
nM
OVFLO=
RC0
RC4
RC16
PAD=
RC0
RC4
RC16
RESALL=
n
nK
nM
RESET
NORESET
RESINV=
n
nK
nM
SDB=
LARGE
YES
INPUT
NO
SKIPREC=z
SMF=
SHORT
FULL
NO
SOLRF
NOSOLRF
SORTDD=cccc
SORTIN=ddname
SORTOUT=ddname
174

SPANINC=
RC0
RC4
RC16
STOPAFT=n
SZERO
NOSZERO
TRUNC=
RC0
RC4
RC16
USEWKDD
VERIFY
NOVERIFY
VLLONG
NOVLLONG
VLSCMP
NOVLSCMP
VLSHRT
NOVLSHRT
VSAMEMT
NVSAMEMT
VSAMIO
NOVSAMIO
WRKREL
NOWRKREL
WRKSEC
NOWRKSEC
Y2PAST=
s
f
ZDPRINT
NZDPRINT
Note for Syntax Diagram: The keywords EFS, LIST, NOLIST, LISTX, NOLISTX,
MSGDDN, MSGPRT, SMF, SORTDD, SORTIN, SORTOUT, and USEWKDD are
used only when they are specified on the OPTION control statement passed by an
extended parameter list or when specified in the DFSPARM data set. If they are
specified on an OPTION statement read from the SYSIN or SORTCNTL data set,
the keyword is recognized, but the parameters are ignored.
The OPTION control statement allows you to override some of the options
available at installation time (such as EQUALS and CHECK) and to supply other
optional information (such as DYNALLOC, COPY, and SKIPREC).
Some of the options available on the OPTION statement are also available on the
SORT or MERGE statement (such as FILSZ and SIZE). It is preferable to specify
these options on the OPTION statement. For override rules, see Appendix B,
Details of aliases for OPTION statement options are given under the description of
individual options. Table 38 on page 219 summarizes the available aliases.
DFSORT accepts but does not process the following OPTION operands: ALGQ,
APP, APPEND, BIAS=value, BLKSET, CASCADE, DIAG, ERASE, EXCPVR=value,
MAXPFIX=value, NEW, NEWFILE, NODIAG, NOERASE, NOINC, NOSTIMER,
NOSWAP, OPT=value, REP, REPLACE, ROUTE=value, WRKADR=value,
WRKDEV=value, and WRKSIZ=value.
ARESALL
175

ARESALL=
n
nK
nM
Temporarily overrides the ARESALL installation option, which specifies the

number of bytes to be reserved above virtual for system use.
ARESALL applies only to the amount of main storage above virtual. This
option is normally not needed because of the large amount of storage available
above 16MB virtual (the default for ARESALL is 0 bytes). The RESALL option
applies to the amount of main storage below 16MB virtual.
n

Limit: 8 digits.
nK
specifies that n times 1024 bytes of storage are to be reserved.

Limit: 5 digits.
nM
specifies that n times 1048576 bytes of storage are to be reserved.

Limit: 2 digits.

Applicable Functions: SeeAppendix B, Specification/override of DFSORT
ARESINV
ARESINV=
n
nK
nM
Temporarily overrides the ARESINV installation option, which specifies the

number of bytes to be reserved for an invoking program's user exits that reside
in or use space above 16MB virtual. The reserved space is not meant to be
used for the invoking program's executable code. ARESINV is used only when
DFSORT is dynamically invoked.
ARESINV applies only to the amount of main storage above 16MB virtual. The
RESINV option applies to the amount of main storage below 16MB virtual.
n

Limit: 8 digits.

Limit: 5 digits.
nM specifies that n times 1048576 bytes of storage are to be reserved.
Limit: 2 digits.
AVGRLEN
AVGRLEN=n
Specifies the average input record length in bytes for variable-length record
176

sort applications. This value is used when necessary to determine the input file
size. The resulting value is important for sort applications, because it is used
for several internal optimizations as well as for dynamic work data set
allocation (see OPTION DYNALLOC). See Specify input/output data set
characteristics accurately on page 801 and Allocation of work data sets on
page 855 for more information on file size considerations.
n
specifies the average input record length. n must be between 4 and 32767
and must include the 4-byte record descriptor word (RDW).
Note:
1. AVGRLEN=n on the OPTION statement overrides the L5 value on the
RECORD statement (LENGTH operand) if both are specified. The L5 value
on the RECORD statement is ignored for Blockset.
2. L5=n can be used instead of AVGRLEN=n.
Default: If AVGRLEN=n is not specified, DFSORT uses one-half of the
maximum record length as the average record length. See Appendix B,
details.
CHALT or NOCHALT
CHALT
NOCHALT
Temporarily overrides the CHALT installation option, which specifies whether

format CH fields are translated by the alternate collating sequence as well as
format AQ or just the latter.
CHALT
specifies that DFSORT translates character control fields and compare fields
with formats CH and AQ using the alternate collating sequence.
NOCHALT
specifies that format CH fields are not translated.
Note: If you use locale processing for SORT, MERGE, INCLUDE, or OMIT
fields, you must not use CHALT. If you need alternate sequence processing for
a particular field, use format AQ.
CHECK or NOCHECK
CHECK
NOCHECK
Temporarily overrides the CHECK installation option, which specifies whether

the record count should be checked for applications that use the E35 user exit
routine without an output data set.
CHECK
specifies that the record count should be checked.
177

NOCHECK
specifies that the record count should not be checked.
Applicable Functions: SeeAppendix B, Specification/override of DFSORT
CINV or NOCINV
CINV
NOCINV
Temporarily overrides the CINV installation option, which specifies whether

DFSORT can use control interval access for VSAM data sets. The Blockset
technique uses control interval access for VSAM input data sets, when
possible, to improve performance.
CINV
specifies that DFSORT should use control interval access when possible for
VSAM data sets.
NOCINV
specifies that DFSORT should not use control interval access.
CKPT
CKPT
Activates the Checkpoint/Restart facility for sorts that use the Peerage or Vale
techniques.
Because CKPT is only supported in the Peerage and Vale techniques, the
Blockset technique must be bypassed for the Checkpoint/Restart facility to be
used. Installation option IGNCKPT=NO causes Blockset to be bypassed when
CKPT is specified at run-time. The NOBLKSET option can also be used to
bypass Blockset at run-time.
A SORTCKPT DD statement must be coded when you use the
Checkpoint/Restart facility (see SORTCKPT DD statement on page 75).
Note:
1. CHKPT can be used instead of CKPT.
2. Functions such as OUTFIL processing, which are supported only by the
Blockset technique, cannot be used if the Checkpoint/Restart facility is
used.
COBEXIT
178

COBEXIT=
COB1
COB2
Temporarily overrides the COBEXIT installation option, which specifies the

library for COBOL E15 and E35 routines.
COB1
specifies that COBOL E15 and E35 routines are run with the OS/VS
COBOL run-time library or, in some cases, with no COBOL run-time
library.
COBEXIT=COB1 is obsolete, but is still available for compatibility reasons.
COB2
specifies that COBOL E15 and E35 routines are run with either the VS
COBOL II run-time library or the Language Environment run-time library.
Note: The DFSORT documents only discuss the Language Environment
run-time library, and assume that COBEXIT=COB2 is in effect. Although it is
possible to run with older run-time libraries, and with COBEXIT=COB1, these
are not recommended or discussed, and are not supported by IBM service.
COPY
COPY
Causes DFSORT to copy a SORTIN data set or inserted records to the output
data sets unless all records are disposed of by an E35 exit routine. Records can
be edited by E15 and E35 exit routines; INCLUDE/OMIT, INREC, OUTREC,
and OUTFIL statements; and SKIPREC and STOPAFT parameters. E35 is
entered after each SORTIN or E15 record is copied.
The following must not be used in copy applications:
v BDAM data sets
v Dynamic binding or link-editing.
See message ICE160A in z/OS DFSORT Messages, Codes and Diagnosis Guide for
additional restrictions that apply to copy applications.
Note: User labels will not be copied to the output data sets.
DSA
DSA=n
Temporarily overrides the DSA installation option, which specifies the

maximum amount of storage available to DFSORT for dynamic storage
179

adjustment of a Blockset sort application when SIZE/MAINSIZE=MAX is in
effect. If you specify a DSA value greater than the TMAXLIM value in effect,
you allow DFSORT to use more storage than the TMAXLIM value if doing so
should improve performance. The amount of storage DFSORT uses is subject
to the DSA value as well as system limits such as region size. However,
whereas DFSORT always tries to obtain as much storage as it can up to the
TMAXLIM value, DFSORT only tries to obtain as much storage as needed to
improve performance up to the DSA value.
The performance improvement from dynamic storage adjustment usually
provides a good tradeoff against the increased storage used by DFSORT. On
storage constrained systems, however, the DSA value should be set low
enough to prevent unacceptable paging.
n
specifies that DFSORT can dynamically adjust storage to improve

performance, subject to a limit of n MB. n must be a value between 0 and
2000. If n is less than or equal to the TMAXLIM value in effect, n is set to 0
to indicate that storage will not be dynamically adjusted.

DSPSIZE
DSPSIZE=
MAX
n
Temporarily overrides the DSPSIZE installation option, which specifies the

maximum amount of data space to be used with dataspace sorting. A data
space is an area of contiguous virtual storage that is backed by real and
auxiliary storage, whichever is necessary as determined by the system. Because
DFSORT is able to sort large pieces of data using data space, CPU time and
elapsed time are reduced.
Several factors can limit the amount of data space an application uses:
v The IEFUSI exit can limit the total amount of Hiperspace, memory objects
and data space available to an application.
v DSPSIZE can limit the amount of data space available to an application, as
detailed later in this section.
v Sufficient available storage must be present to back DFSORT's data spaces.
"Available" storage is the storage used to back new data space data and
consists of the following two types:
1. Free storage. This is storage not being used by any application.
2. Old storage. This is storage used by another application whose data has
been unreferenced for a sufficiently long time so that the system
considers it eligible to be paged out to auxiliary storage to make room
for new data space data.
The amount of available storage constantly changes, depending on current
system activity. Consequently, DFSORT checks the amount of available
central storage throughout a data space sorting run and switches from using
a data space to using disk work data sets if the available central storage is
too low.
v Other concurrent Hipersorting, memory sorting and dataspace sorting
applications further limit the amount of available storage. A dataspace
sorting application knows the storage needs of every other Hipersorting,
memory object sorting and dataspace sorting application on the system. A
dataspace sorting application does not try to back its data space data with
180

storage needed by another Hipersorting, memory object sorting or dataspace
sorting application. This prevents overcommitment of storage resources if
multiple large concurrent DFSORT applications start at similar times on the
same system.
v The installation options EXPMAX, EXPOLD, and EXPRES can also be used
to further limit the amount of storage available to dataspace sorting
applications. EXPMAX limits the total amount of available storage that can
be used at any one time to back DFSORT Hiperspaces, memory objects and
data spaces. EXPOLD limits the total amount of old storage that can be used
at any one time to back DFSORT Hiperspaces, memory objects and data
spaces. EXPRES sets aside a specified amount of available storage for use by
non-Hipersorting, non-memory object sorting and non-dataspace sorting
applications.
Some of these limits depend on system and other DFSORT activity throughout
the time a dataspace sorting application runs. Consequently, the amount of
data space a dataspace sorting application uses can vary from run to run.
If the amount of data space DFSORT decides to use is sufficient, DFSORT sorts
your data in main storage and does not require additional temporary work
space. If the amount of data space is not sufficient, DFSORT uses disk as
temporary work space. Installation option DYNAUTO=NO is changed to
DYNAUTO=YES whenever dataspace sorting is possible. Hiperspace is not
used when dataspace sorting is used.
MAX
specifies that DFSORT dynamically determines the maximum amount of
data space to be used for dataspace sorting. In this case, DFSORT bases its
data space usage on the size of the file being sorted and the paging activity
of the system.
n
specifies the maximum amount, in MB, of data space to be used for

dataspace sorting. n must be a value between 0 and 9999. The actual
amount of data space used does not exceed n, but may be less than n
depending on the size of the file being sorted and the paging activity of
the system.
If n is zero, dataspace sorting is not used.

DYNALLOC
DYNALLOC

=
d
(,n)
(d,n)
Assigns DFSORT the task of dynamically allocating needed work space. You
do not need to calculate and use JCL to specify the amount of work space
needed by the program. DFSORT uses the dynamic allocation facility of the
operating system to allocate work space for you.
Refer to Appendix A, Using work space, on page 853 for guidelines on the
use of DYNALLOC.
d
specifies the device name. You can specify any IBM disk or tape device
181

supported by your operating system in the same way you would specify it
in the JCL UNIT parameter. You can also specify a group name, such as
DISK or SYSDA.
n
specifies the maximum number of requested work data sets. If you specify
more than 255, a maximum of 255 data sets is used. If you specify 1 and
the Blockset technique is selected, a maximum of 2 data sets is used. If you
specify more than 32 and the Blockset technique is not selected, a
maximum of 32 data sets is used.
Tip: For optimum allocation of resources such as virtual storage, avoid

specifying a large number of work data sets unnecessarily.
For tape work data sets, the number of volumes specified (explicitly or by
default) is allocated to the program. The program requests standard label
tapes.
DYNALLOC is not used if SORTWKdd DD statements are provided unless
installation option DYNAUTO=IGNWKDD is specified and OPTION
USEWKDD is not in effect.
If VIO=NO is in effect:
v Work space can be allocated on nontemporary data sets (DSNAME
parameter specified).
v If the device (d) you specify is a virtual device and reallocation to a real
device fails, DFSORT will ignore VIO=NO and use the virtual device.
Note: Message ICE165I gives information about work data set allocation/use.
v DYNALLOC can automatically be activated by using the DYNAUTO
installation option.
v If DYNALLOC is specified without d, the default for d is that specified (or
defaulted) by the DYNALOC installation option
v If DYNALLOC is specified without n, the default for n is that specified (or
defaulted) by the DYNALOC installation option.
You can specify DYNALLOC without n, without d, or without both. If
DYNALLOC is specified without n, and the IBMsupplied default for the n
value of the DYNALOC installation option is chosen, then:
v If one of the Blockset techniques is chosen, four work data sets will be
requested.
v If a technique other than Blockset is chosen, three work data sets will be
requested.
DYNALLOC=OFF
DYNALLOC=OFF
Directs DFSORT not to allocate work space dynamically, overriding that

function of installation option DYNAUTO=YES, or DYNAUTO=IGNWKDD, or
the run-time option DYNALLOC (without OFF). Use this option when you
know that an in-core sort can be performed, and you want to suppress
dynamic allocation of work space.
OFF
directs DFSORT not to allocate work space dynamically.
182

Note: When Hipersorting or dataspace sorting is in effect, DFSORT uses
dynamic allocation when necessary, even if DYNALLOC=OFF has been
specified.
DYNAPCT
DYNAPCT=
x
OLD
specifies additional work data sets to be dynamically allocated with zero space.
DFSORT only extends these data sets when necessary to complete a sort
application. The availability of additional work data sets can help avoid out of
space ABENDs.
x
specifies the number of additional work data sets (y) as a percentage of the
maximum number of dynamically allocated work data sets
(DYNALLOC/DYNALOC n value) in effect. y will be set to n * x%. The
total number of dynamically allocated work data sets will be n + y. For
example, if DYNALLOC=(SYSDA,20) and DYNAPCT=20 are in effect, 4
additional work data sets will be allocated for a total of 24.
The value x must be between 0 and 254. The minimum value for y is 1 and
the maximum value for y is 254. The maximum value for n + y is 255; if x
results in a value for n + y greater than 255, y will be set to 255-n.
OLD
specifies additional work data sets should only be allocated when DFSORT
cannot determine the file size. When DFSORT is able to determine the file
size, additional work data sets will not be allocated (y=0), and the total
number of work data sets will be n.
Note: When message ICE118I is issued indicating that DFSORT cannot
determine the file size, y is set as follows:
v For DYNAPCT=OLD, y is set to n * 50%
v For DYNAPCT=x with x <= 50, y is set to n * 50%
v For DYNAPCT=x with x > 50, y is set to n * x%
DYNSPC
DYNSPC=n
DYNSPC=ntemporarily overrides the DYNSPC installation option, which

specifies the total default primary space allocation for all of the dynamically
allocated work data sets when the input file size is unknown. That is, when
DFSORT cannot determine the input file size for a sort application and the
number of records is not supplied by a FILSZ or SIZE value.
183

Generally, DFSORT can automatically determine the input file size. However,
in a few cases, such as when an E15 supplies all of the input records, when
information about a tape data set is not available from a tape management
system, or when Blockset is not selected, DFSORT cannot determine an
accurate file size. In these cases, if the number of records is not supplied by the
FILSZ or SIZE run-time option, and dynamic allocation of work data sets is
used, DFSORT uses the DYNSPC value in effect as the approximate amount of
primary space. DFSORT uses 20% of the primary space as secondary space.
Although the primary space is always allocated, secondary space (up to 15
extents) is only allocated as needed.
You may want to use DYNSPC to override the installation default with a larger
or smaller value depending on the amount of disk space available for DFSORT
work data sets, and the amount of data to be sorted for this application. As a
guideline, Table 34 shows the approximate primary space in cylinders that is
allocated on a 3390 when Blockset sorts an unknown number of 6000-byte
records.
Table 34. Example of DYNSPC Primary Space
DYNSPC value (megabytes)
Primary space (cylinders)
32
48
64
93
128
183
256
366
512
732
The larger your DYNSPC value, the more data DFSORT can sort when the file
size is unknown. For example, in a test using just dynamically allocated work
space (no Hiperspace or data space) with the primary space shown in Table 34,
and all of the corresponding secondary space, Blockset is able to sort
approximately 150 megabytes with DYNSPC=32 and approximately 1200
megabytes with DYNSPC=256. If Hiperspace or data space can be used along
with dynamically allocated work space, the amount of data DFSORT can sort
will increase according to the amount of Hiperspace or data space available.
n
specifies the total default primary space, in megabytes, to be allocated

for all dynamically allocated work data sets (n is not the primary space
for each data set). n must be a value between 1 and 65535.
Do not specify a value which exceeds the available disk space, because
this causes dynamic allocation to fail for sort applications that use this
value.

EFS
EFS=
184
name
NONE

Temporarily overrides the EFS installation option, which specifies whether
DFSORT is to pass control to an Extended Function Support (EFS) program.
See Chapter 9, Using extended function support, on page 767 for more
information.
name
specifies the name of the EFS program that will be called to interface with
DFSORT.
NONE
specifies no call will be made to the EFS program.
Note:
1. EFS is processed only if it is passed on the OPTION control statement in an
extended parameter list or in DFSPARM.
2.
3. If you use locale processing for SORT, MERGE, INCLUDE, or OMIT fields,
you must not use an EFS program. DFSORT's locale processing may
eliminate the need for an EFS program. See the LOCALE option later in
this section for information related to locale processing.
EQUALS or NOEQUALS
EQUALS
NOEQUALS
Temporarily overrides the EQUALS installation option, which specifies

whether the original sequence of records that collate identically for a sort or a
merge should be preserved from input to output.
EQUALS
specifies that the original sequence must be preserved.
NOEQUALS
specifies that the original sequence need not be preserved.
For sort applications, the sequence of the output records depends upon the
order of:
v The records from the SORTIN file
v The records inserted by an E15 user exit routine
v The E15 records inserted within input from SORTIN.
For merge applications, the sequence of the output records depends upon the
order of:
v The records from a SORTINnn file. Records that collate identically are
output in the order of their file increments. For example, records from
SORTIN01 are output before any records that collate identically from
SORTIN02.
v The records from an E32 user exit routine for the same file increment
number. Records that collate identically from E32 are output in the order of
185

their file increments. For example, records from the file with increment 0 are
output before any records that collate identically from the file with
increment 4.
Note:
1. When EQUALS is in effect, the total number of bytes occupied by all
control fields must not exceed 4088.
2. Using EQUALS can degrade performance.
3. When EQUALS is in effect with SUM, the first record of summed records is
kept. When NOEQUALS is in effect with SUM, the record to be kept is
unpredictable.
If a technique other than Blockset is selected, NOEQUALS is forced if SUM
is specified.
4. Do not specify EQUALS if variable-length records are sorted using tape
work files and the RDW is part of the control field.
5. The number of records to be sorted cannot exceed 4294967295
(4 gigarecords minus 1); if the number of records exceeds this number,
message ICE121A is issued and DFSORT terminates.
EXITCK
EXITCK
STRONG
WEAK
Temporarily overrides the EXITCK installation option, which specifies whether

DFSORT terminates or continues when it receives certain invalid return codes
from E15 or E35 user exit routines. For full details of the return codes affected
by this parameter, see E15/E35 return codes and EXITCK on page 537.
STRONG
specifies that DFSORT issues an error message and terminates when it
receives an invalid return code from an E15 or E35 user exit routine.
WEAK
specifies that DFSORT interprets certain invalid return codes from E15 and
E35 user exit routines as valid and continues processing. Use of
EXITCK=WEAK can make it difficult to detect errors in the logic of E15
and E35 user exit routines.
Note: EXITCK=WEAK is treated like EXITCK=STRONG when:
v Tape work data sets are specified for a sort application.
v The Blockset technique is not selected for a merge application.
FILSZ or SIZE
186

FILSZ=
SIZE=
x
Ex
Ux
y
Ey
Uy
The FILSZ parameter specifies either the exact number of records to be sorted
or merged, or an estimate of the number of records to be sorted. The SIZE
parameter specifies either the exact number of records in the input data sets, or
an estimate of the number of records in the input data sets. The supplied
record count is used by DFSORT for two purposes:
1. To check that the actual number of records sorted or merged or the number
of records in the input data sets is equal to the exact number of records
expected. FILSZ=x or SIZE=y causes this check to be performed and results
in termination with message ICE047A if the check fails.
2. To determine the input file size for a sort application. DFSORT performs
calculations based on the user supplied record count and other parameters
(such as AVGRLEN) to estimate the total number of bytes to be sorted. This
value is important for sort runs, because it is used for several internal
optimizations as well as for dynamic work data set allocation (see OPTION
DYNALLOC). If no input record count (or only an estimate) is supplied for
the sort run, DFSORT attempts to automatically compute the file size to be
used for the optimizations and allocations.
The type of FILSZ or SIZE value specified (x/y, Ux/Uy, Ex/Ey, or none)
controls the way DFSORT performs the previous two functions, and can have a
significant effect on performance and work data set allocation. See Chapter 10,
Improving efficiency, on page 799 and File size and dynamic allocation on
page 857 for more information on file size considerations.
x or y
specifies the exact number of records to be sorted or merged (x) or the
exact number of records in the input data sets (y). This value is always
used for both the record check and the file size calculations. FILSZ=x or
SIZE=y can be used to force DFSORT to perform file size calculations
based on x or y, and to cause DFSORT to terminate the sort or merge
application if x or y is not exact.
If installation option FSZEST=NO is in effect and either FILSZ=x or SIZE=y
is specified, DFSORT terminates if the actual number of records is different
from the specified exact value (x or y). In this case, the actual number of
records is placed in the IN field of message ICE047A (or message ICE054I
in some cases) before termination. However, if installation option
FSZEST=YES is in effect, DFSORT treats FILSZ=x or SIZE=y like FILSZ=Ex
or SIZE=Ey, respectively; it does not terminate when the actual number of
records does not equal x or y.
FILSZ=0 causes Hipersorting, dataspace sorting, and dynamic allocation of
work space not to be used, and results in termination with message
ICE047A unless the number of records sorted or merged is 0. If no E15
user exit is present, SIZE=0 has the same effect in terms of Hipersorting
and dynamic allocation of work space, and results in termination with
message ICE047A unless the number of records in the input data sets is 0.
x
specifies the exact number of records to be sorted or merged; it must

take into account the number of records in the input data sets, records
to be inserted or deleted by E15 or E32, and records to be deleted by
187

the INCLUDE/OMIT statement, SKIPREC, and STOPAFT. x must be
changed whenever the number of records to be sorted or merged
changes in any way.
y
specifies the exact number of records in the input data sets; it must
take into account the number of records to be deleted by STOPAFT. y
must be changed whenever the number of records in the input data
sets changes in any way.
Ex or Ey
specifies an estimate of the number of records to be sorted (x) or an
estimate of the number of records in the input data sets (y). This value is
not used for the record check. It is used for the file size calculations, but
only if DFSORT could not reasonably estimate the input file size itself. In
all other cases, this value is ignored by DFSORT. See Dynamic allocation
of work data sets on page 856 for details on exactly when an estimated
record count is used and when it is ignored by DFSORT.
FILSZ=E0 or SIZE=E0 is always ignored.
x
specifies an estimate of the number of records to be sorted; it should

take into account the number of records in the input data sets, records
to be inserted or deleted by E15, and records to be deleted by the
INCLUDE/OMIT statement, SKIPREC, and STOPAFT. x should be
changed whenever the number of records to be sorted changes
significantly.
specifies an estimate of the number of records in the input data sets; it

should take into account the number of records to be deleted by
STOPAFT. y should be changed whenever the number of records in the
input data sets changes significantly.
Ux or Uy
specifies the number of records to be sorted (x) or the number of records in
the input data sets (y). This value is not used for the record check, but is
always used for the file size calculations. FILSZ=Ux or SIZE=Uy can be
used to force DFSORT to perform file size calculations based on x or y,
while avoiding termination if x or y is not exact.
The FSZEST installation option has no effect on FILSZ=Ux or SIZE=Uy
processing.
FILSZ=U0 causes Hipersorting, dataspace sorting, and dynamic allocation
of work space not to be used, and may cause degraded performance or
termination with message ICE046A, if the actual number of records to be
sorted is significantly larger than 0. If no E15 user exit is present, SIZE=U0
has the same effect in terms of Hipersorting, dataspace sorting,and
dynamic allocation of work space, and may cause degraded performance
or termination with message ICE046A, if the actual number of records in
the input data sets is significantly larger than 0.
x
188
specifies the number of records to be sorted; it should take into

account the number of records in the input data sets, records to be
inserted or deleted by E15, and records to be deleted by the
INCLUDE/OMIT statement, SKIPREC, and STOPAFT. x should be
changed whenever the number of records to be sorted changes
significantly.

y
specifies the number of records in the input data sets; it should take
into account the number of records to be deleted by STOPAFT. y
should be changed whenever the number of records in the input data
sets changes significantly.
Table 35 summarizes the differences for the three FILSZ variations:

Table 35. FILSZ Variations Summary
FILSZ=n is equivalent to
FILSZ=En if installation
option FSZEST=YES is
specified.Conditions
FILSZ=n
FILSZ=Un
FILSZ=En
Number of records
Exact
Estimate
Estimate
Applications
Sort, merge
Sort
Sort
Terminate if n wrong?
Yes
No
No
Use for file size calculation? Yes
Yes
When DFSORT cannot

compute file size
n includes records:
In input data sets
Yes
Yes
Yes
Yes
Yes
Yes
Inserted by E32
Yes
No
No
Deleted by
INCLUDE/OMIT
Yes
Yes
Yes
Deleted by SKIPREC
Yes
Yes
Yes
Deleted by STOPAFT
Yes
Yes
Yes
Update n when number of

records changes:
In any way
Significantly
Significantly
Effects of n=0
Hipersorting and
DYNALLOC not used
Hipersorting and
DYNALLOC not used
None
Table 36 summarizes the differences for the three SIZE variations:

Table 36. SIZE Variations Summary
SIZE=n is equivalent to
SIZE=En if installation
SIZE=n
SIZE=Un
SIZE=En
Number of records
Exact
Estimate
Estimate
Applications
Sort, merge
Sort
Sort
Terminate if n wrong?
Yes
No
No
Use for file size calculation? Yes
Yes
When DFSORT cannot

compute file size
n includes records:
In input data sets
Yes
Yes
Yes
No
No
No
Inserted by E32
No
No
No
Deleted by
INCLUDE/OMIT
No
No
No
189

Table 36. SIZE Variations Summary (continued)
SIZE=n is equivalent to
SIZE=En if installation
SIZE=n
SIZE=Un
SIZE=En
Deleted by SKIPREC
No
No
No
Deleted by STOPAFT
Yes
Yes
Yes
Update n when number of

records changes:
In any way
Significantly
Significantly
Effects of n=0
Hipersorting and
DYNALLOC not used
Hipersorting and
DYNALLOC not used
None
Attention: Using the SIZE or FILSZ parameter to supply inaccurate

information to DFSORT can negatively affect DFSORT performance, and, when
work space is dynamically allocated, can result in wasted disk space or
termination with message ICE083A or ICE046A. Therefore, it is important to
update the record count value whenever the number of records to be sorted
changes significantly.
HIPRMAX
HIPRMAX=
OPTIMAL
n
p%
Temporarily overrides the HIPRMAX installation option, which specifies the

maximum amount of Hiperspace to be used for Hipersorting. Hiperspace is a
high-performance data space that resides in central storage and is backed by
auxiliary storage (if necessary). Because I/O processing is reduced for
Hipersorting, elapsed time, EXCP counts, and channel usage are also reduced.
Several factors can limit the amount of Hiperspace an application uses:
v The IEFUSI exit can limit the total amount of Hiperspace, memory objects
and data space available to an application.
v HIPRMAX can limit the amount of Hiperspace available to an application,
as detailed later in this section.
v Sufficient available storage must be present to back DFSORT's Hiperspaces.
"Available" storage is the storage used to back new Hiperspace data and
consists of the following two types:
1. Free storage. This is storage not being used by any application.
2. Old storage. This is storage used by another application whose data has
been unreferenced for a sufficiently long time so that the system
considers it eligible to be paged out to auxiliary storage to make room
for new Hiperspace data.
The amount of available storage constantly changes, depending upon current
system activity. Consequently, DFSORT checks the available storage level
throughout a Hipersorting application and switches from Hiperspace to work
data sets if the available storage level gets too low.
v Other concurrent Hipersorting, memory sorting and dataspace sorting
applications further limit the amount of available storage. A Hipersorting
application knows the storage needs of every other Hipersorting, memory
object sorting and dataspace sorting application on the system. A
190

Hipersorting application does not try to back its Hiperspace data with
storage needed by another Hipersorting, memory object sorting, or
dataspace sorting application. This prevents overcommitment of storage
resources if multiple large concurrent DFSORT applications start at similar
times on the same system.
v The installation options EXPMAX, EXPOLD, and EXPRES can also be used
to further limit the amount of storage available to Hipersorting applications.
EXPMAX limits the total amount of available storage that can be used at any
one time to back DFSORT Hiperspaces, memory objects and data spaces.
EXPOLD limits the total amount of old storage that can be used at any one
time to back DFSORT Hiperspaces, memory objects and data spaces.
EXPRES sets aside a specified amount of available storage for use by
non-Hipersorting, non-memory object sorting, and non-dataspace sorting
applications.
Some of these limits depend on system and other DFSORT activity throughout
the time a Hipersorting application runs. Consequently, the amount of
Hiperspace a Hipersorting application uses can vary from run to run.
HIPRMAX=n specifies a fixed value for HIPRMAX. HIPRMAX=p% specifies a
value for HIPRMAX that varies as a percentage of an appropriate portion of
central storage. If the storage on a system changes, HIPRMAX=p% will cause a
corresponding change in the HIPRMAX value selected by DFSORT, whereas
HIPRMAX=n will not. When sharing DFSORT installation options between
systems, such as in a sysplex, HIPRMAX=p% can be used to tailor the
HIPRMAX value to the system selected for the application, providing a more
dynamic HIPRMAX value than HIPRMAX=n.
If the amount of Hiperspace available for Hipersorting is insufficient for
temporary storage of the records, intermediate disk storage is used along with
Hiperspace. If the amount of Hiperspace is too small to improve performance,
Hipersorting is not used. DYNAUTO=NO is changed to DYNAUTO=YES for
Hipersorting.
Hipersorting can cause a small CPU time increase. When CPU optimization is
a concern, you can use HIPRMAX=0 to suppress Hipersorting.
Note: HIPRLIM=OPTIMAL can be used instead of HIPRMAX=OPTIMAL.
HIPRLIM=m can be used instead of HIPRMAX=n. HIPRLIM=m specifies a
Hiperspace limit of m times 4096 bytes rounded up to the nearest megabyte. m
must be a value between 0 and 2559744. If m is 0, Hipersorting is not used.
OPTIMAL
Hiperspace to be used for Hipersorting.
n

Hiperspace to be used for Hipersorting, subject to a limit of nMB. n must
be a value between 0 and 32767. If n is 0, Hipersorting is not used.
p% specifies that DFSORT determines dynamically the maximum amount of

hiperspace to be used for Hipersorting, subject to a limit of p percent of an
appropriate portion of central storage. p must be a value between 0 and
100. If p is 0, Hipersorting is not used. The value calculated for p% is
limited to 32767MB, and is rounded down to the nearest MB.
191

LIST or NOLIST
LIST
NOLIST
Temporarily overrides the LIST installation option, which specifies whether

DFSORT program control statements should be written to the message data
set. See z/OS DFSORT Messages, Codes and Diagnosis Guide for details on use of
LIST
specifies that DFSORT control statements are printed to the message data
set.
NOLIST
specifies that DFSORT control statements are not printed to the message
data set.
Note: LIST or NOLIST are processed only if they are passed on the OPTION
control statement in an extended parameter list or in DFSPARM.
LISTX or NOLISTX
LISTX
NOLISTX
Temporarily overrides the LISTX installation option, which specifies whether

DFSORT writes to the message data set program control statements that are
returned by an EFS program. See z/OS DFSORT Messages, Codes and Diagnosis
Guide for details on use of the message data set.
LISTX
specifies that control statements returned by an EFS program are printed to
NOLISTX
specifies that control statements returned by an EFS program are not
printed to the message data set.
Note:
1. LISTX or NOLISTX are processed only if they are passed on the OPTION
control statement in an extended parameter list or in DFSPARM.
2. If EFS=NONE is in effect after final override rules have been applied,
NOLISTX is in effect.
3. LISTX and NOLISTX can be used independently of LIST and NOLIST.
4. For more information on printing EFS control statements, see z/OS DFSORT
Messages, Codes and Diagnosis Guide
LOCALE
192

LOCALE=
name
CURRENT
NONE
Temporarily overrides the LOCALE installation option, which specifies

whether locale processing is to be used and, if so, designates the active locale.
environment. Your cultural environment is defined to DFSORT using the
X/Open locale model. A locale is a collection of data grouped into categories
that describes the information about your cultural environment.
The collate category of a locale is a collection of sequence declarations that
defines the relative order between collating elements (single character and
multi-character collating elements). The sequence declarations define the
collating rules.
If locale processing is to be used, the active locale will affect the behavior of
DFSORT's SORT, MERGE, INCLUDE, and OMIT functions. For SORT and
MERGE, the active locale will only be used to process character (CH) control
fields. For INCLUDE and OMIT, the active locale will only be used to process
character (CH) compare fields, and character and hexadecimal constants
compared to character (CH) compare fields.
Note: Locale processing is not used for IFTRAIL TRLID or IFTHEN WHEN,
name
specifies that locale processing is to be used and designates the name

of the locale to be made active during DFSORT processing.
The locales are designated using a descriptive name. For example, to
set the active locale to represent the French language and the cultural
conventions of Canada, specify LOCALE=FR_CA. You can specify up
to 32 characters for the descriptive locale name. The locale names
themselves are not case-sensitive. See Using Locales for complete locale
naming conventions.
You can use IBM-supplied and user-defined locales.
The state of the active locale prior to DFSORT being entered will be
restored on DFSORT's completion.
CURRENT
specifies that locale processing is to be used, and the current locale
active when DFSORT is entered will remain the active locale during
DFSORT processing.
NONE
specifies that locale processing is not to be used. DFSORT will use the
binary encoding of the code page defined for your data for collating
and comparing.
Note:
1. LOCALE is processed only if it is passed on the OPTION control statement
in an extended parameter list or in DFSPARM.
2. To use an IBM-supplied locale, DFSORT must have access to the Language
Environment run-time library. For example, this library might be called
SYS1.SCEERUN. If you are unsure of the name of this library at your
location, contact your system administrator. To use a user-defined locale,
DFSORT must have access to the load library containing it.
3. If you use locale processing for SORT, MERGE, INCLUDE, or OMIT fields:
v VLSHRT is not used for SORT or MERGE
193

v INREC or an E61 user exit must not be usedfor SORT or MERGE.
v CHALT or an EFS program must not be used.
4. Locale processing is not used for IFTRAIL TRLID or IFTHEN WHEN,
5. Locale processing for DFSORT's SORT, MERGE, INCLUDE, and OMIT
functions can improve performance relative to applications that must
perform pre-processing or post-processing of data to produce the desired
collating results. However, locale processing should be used only when
required, because it can show degraded performance relative to collating,
using character encoding values.
6. DFSORT locale processing may require an additional amount of storage
that depends on the environment supporting the locale as well as the locale
itself. It may be necessary to specify a REGION of several MB or more for
DFSORT applications that use locale processing.
Applicable Functions:; See Appendix B, Specification/override of DFSORT
MAINSIZE
MAINSIZE=
n
nK
nM
MAX
Temporarily overrides the SIZE installation option, which specifies the amount
of main storage available to DFSORT.
MAINSIZE applies to the total amount of main storage above and below 16MB
virtual. DFSORT determines how much storage to allocate above and below
16MB virtual, but the total amount of storage cannot exceed MAINSIZE.
Storage used for OUTFIL processing will be adjusted automatically, depending
upon several factors, including:
v Total available storage
v Non-OUTFIL processing storage requirements
v Number of OUTFIL data sets and their attributes (for example, block size).
OUTFIL processing is subject to the ODMAXBF limit and your system storage
limits (for example, IEFUSI) but not to DFSORT storage limits, that is,
SIZE/MAINSIZE, MAXLIM, and TMAXLIM. DFSORT attempts to use storage
above 16MB virtual for OUTFIL processing whenever possible.
For details on main storage allocation, see Tuning main storage on page 807.
n
specifies that n bytes of storage are to be allocated. If you specify more that
2097152000, 2097152000 is used.
Limit: 10 digits
nK specifies that n times 1024 bytes of storage are to be allocated. If you

specify more than 2048000K, 2048000K is used.
Limit: 7 digits
nM specifies that n times 1048576 bytes of storage are to be allocated. If you
Limit: 4 digits.
194

MAX
instructs DFSORT to calculate the amount of virtual storage available and
allocate an amount of storage up to the TMAXLIM or DSA value when
Blockset is selected, or up to the MAXLIM value when Blockset is not
selected.
Note: CORE=value can be used instead of MAINSIZE=value.
MERGEIN
MERGEIN=
ddname
,
( ddname
Specifies up to 100 ddnames to be be used for a MERGE application instead of

the SORTINnn ddnames. This allows you to use any valid alternate ddnames
for the MERGE data sets.
If EQUALS is in effect, records that collate identically are output in the order
of their ddnames in the MERGEIN list.
Each ddname can be 1 through 8 characters. If a ddname is specified more
than once in the MERGEIN operand, it will only be used once. If more than
100 unique ddnames are specified in the MERGEIN operand, only the first 100
will be used. Do not use ddnames reserved for use by DFSORT, such as
SYSOUT, ccccOUT, ccccWKd, ccccWKdd, ccccDKd, or ccccDKdd , where cccc is
the specified or defaulted value for the SORTDD operand and d is any
character.
Note:
1. MERGEIN is processed only if it is passed on the OPTION control
statement in an extended parameter list, or in DFSPARM.
2. If both MERGEIN=(ddname1,ddname2,...) and SORTDD=cccc are specified,
ddname1, ddname2 and so on are used for the input files. The same
ddname cannot be specified for MERGEIN and SORTOUT, or for
MERGEIN and OUTFIL.
Default: SORTINnn, unless SORTDD=cccc is specified in an extended parameter
list or in DFSPARM, in which case ccccINnn is the default.
MOSIZE
MOSIZE=
MAX
n
p%
Temporarily overrides the MOSIZE installation option, which specifies the

maximum amount of memory object storage to be used for memory object
sorting in 64-bit virtual storage. A memory object is a data area in virtual
storage that is allocated above the bar and backed by central storage. Because
I/O processing is reduced for memory object sorting, elapsed time, EXCP
counts, and channel usage are also reduced.
195

Note: The "bar" refers to the 2-gigabyte address within the 64-bit address
space. The bar separates storage below the 2-gigabyte address called "below
the bar", from storage above the 2-gigabyte address called "above the bar".
There are several factors that can limit the total memory object storage used by
an application:
1. The MEMLIMIT parameter on the JOB or EXEC JCL statement can limit the
total number of usable virtual pages above the bar in a single address
space.
2. In addition to limiting the total amount of Hiperspace and data space
available to an application, the IEFUSI exit can also limit the total number
of usable virtual pages above the bar in a single address space.
3. MOSIZE can limit the total amount of memory object storage available to
an application, as detailed later in this section.
4. Sufficient available central storage must be present to back DFSORT's
memory object storage. The amount of available central storage changes
constantly, depending on current system activity. Consequently, DFSORT
checks the amount of available central storage throughout a memory object
sorting run and switches from using a memory object to using disk work
data sets if the available central storage is too low.
5. Other concurrent Hipersorting, memory object sorting, and dataspace
sorting applications further limit the amount of available storage. A
memory object sorting application keeps track of the storage needs of all
other Hipersorting, memory object sorting, and dataspace sorting
applications on the system, and does not attempt to back its memory object
storage with storage needed by another Hipersorting, memory object
sorting, or dataspace sorting application. This prevents overcommitment of
storage resources in the event of multiple large concurrent Hipersorting,
memory object sorting, and dataspace sorting applications starting at
similar times on the same system.
6. The installation options EXPMAX, EXPOLD, and EXPRES can also be used
to further limit the amount of storage available to memory object sorting
applications. EXPMAX limits the total amount of available storage that can
be used at any one time to back DFSORT Hiperspaces, memory objects, and
data spaces. EXPOLD limits the total amount of old storage that can be
used at any one time to back DFSORT Hiperspaces, memory objects, and
data spaces. EXPRES sets aside a specified amount of available storage for
use by non-Hipersorting, non-memory object sorting, and non-dataspace
sorting applications.
Some of these limits depend on system and other DFSORT activity during the
time that a memory object sorting application runs. Consequently, the total
amount of memory object storage that a memory object sorting application
uses can vary from run to run.
MOSIZE=n specifies a fixed value for MOSIZE. MOSIZE=p% specifies a value
for MOSIZE that varies as a percentage of the available central storage on the
system at run-time. If the available central storage on a system changes,
MOSIZE=p% will cause a corresponding change in the MOSIZE value selected
by DFSORT, whereas MOSIZE=n will not. When sharing DFSORT installation
options between systems, such as in a sysplex, MOSIZE=p% can be used to
tailor the MOSIZE value to the system selected for the application, providing a
more dynamic MOSIZE value than MOSIZE=n.
If the total memory object storage available for memory object sorting is
insufficient for temporary storage of the records, intermediate disk storage can
be used along with the memory object. When memory object sorting is
enabled, DFSORT changes DYNAUTO=NO to DYNAUTO=YES in some cases.
196

MAX
specifies that DFSORT determines dynamically the maximum size of a
memory object to be used for memory object sorting. In this case, DFSORT
bases its memory object usage on the size of the file being sorted and the
central storage usage activity.
n
specifies that DFSORT determines dynamically the maximum size of a

memory object to be used for memory object sorting, subject to a limit of n
MB. n must be a value between 0 and 2147483646. The actual size used
does not exceed n, but may be less, depending on the size of the file being
sorted and the central storage usage activity on the system. If n is 0,
memory object sorting is not used
p% specifies that DFSORT determines dynamically the maximum size of a

memory object to be used for memory object sorting, subject to a limit of p
percent of the available central storage. p must be a value between 0 and
100. If p is 0, memory object sorting is not used. The value calculated for
p% is limited to 2147483646 MB, and is rounded down to the nearest MB.
MOWRK or NOMOWRK
MOWRK
NOMOWRK
Temporarily overrides the MOWRK installation option, which specifies

whether the memory object storage available to DFSORT for memory object
sorting can be used as intermediate work space. DFSORT has the capability of
using memory object storage as intermediate work space (similar to the way
Hiperspace is used but more efficient), or as an extension of main storage.
Using memory object storage as intermediate work space is the preferred and
recommended choice, but can be disabled, if appropriate.
MOWRK
specifies that memory object storage can be used as intermediate work
space, or as an extension of main storage, as appropriate.
NOMOWRK
specifies that memory object storage can only be used as an extension of
main storage.
MSGDDN
MSGDDN=ddname
Temporarily overrides the MSGDDN installation option, which specifies an

alternate ddname for the message data set. MSGDDN must be in effect if:
v A program that invokes DFSORT uses SYSOUT (for instance, COBOL uses
SYSOUT) and you do not want DFSORT messages intermixed with the
program messages.
197

v Your E15 and E35 routines are written in COBOL and you do not want
DFSORT messages intermixed with the program messages.
v A program invokes DFSORT more than once and you want separate
messages for each invocation of DFSORT.
The ddname can be any 1- through 8- character name but must be unique
within the job step; do not use a name that is used by DFSORT (for example,
SORTIN). If the ddname specified is not available at run-time, SYSOUT is used
instead. For details on use of the message data set, see z/OS DFSORT Messages,
Codes and Diagnosis Guide
Note: MSGDDN is processed only if it is passed on the OPTION control
statement in an extended parameter list or in DFSPARM.
MSGPRT
MSGPRT=
ALL
CRITICAL
NONE
Temporarily overrides the MSGPRT installation option, which specifies the

class of messages to be written to the message data set. For details on use of
the message data set, see z/OS DFSORT Messages, Codes and Diagnosis Guide.
ALL
specifies that all messages except diagnostic messages (ICE800I to ICE999I)
are to be printed. Control statements print only if LIST is in effect.
CRITICAL
specifies that only critical messages will be printed. Control statements
print only if LIST is in effect.
NONE
specifies that no messages and control statements will be printed.
Note:
1. MSGPRT is processed only if it is passed on the OPTION control statement
in an extended parameter list or in DFSPARM.
2. PRINT=value can be used instead of MSGPRT=value.
NOBLKSET
NOBLKSET
Causes DFSORT to bypass the Blockset technique normally used for a sort or
merge application. Using this option generally results in degraded
performance.
Note: Functions such as OUTFIL processing, which are supported only by the
Blockset technique, cause the NOBLKSET option to be ignored.
198

NOOUTREL
NOOUTREL
Temporarily overrides the OUTREL installation option, which specifies whether

unused temporary output data set space is released. NOOUTREL means that
unused temporary output data set space is not released.
NOOUTSEC
NOOUTSEC
Temporarily overrides the OUTSEC installation option, which specifies whether

automatic secondary allocation is used for temporary or new output data sets.
NOOUTSEC means that automatic secondary allocation for output data sets is
not used.
NULLOUT
NULLOUT=
RC0
RC4
RC16
Temporarily overrides the NULLOUT installation option, which specifies the

action to be taken by DFSORT when there are no records for the SORTOUT
data set, as indicated by an OUT count of 0 in message ICE054I
RC0
data set.
RC4
data set.
RC16
a return code of 16 when there are no records for the SORTOUT data set.
Note:
1. The return code of 0 or 4 set when there are no records for the SORTOUT
data set can be overridden by a higher return code set for some other
reason.
2. NULLOUT does not apply when SORTOUT is not present, when tape work
data sets are specified for a sort application, or when the Blockset technique
199

is not selected for a merge application. DFSORT does not check if there are
no records for the SORTOUT data set in these cases.
3. NULLOUT applies to the SORTOUT data set. NULLOFL on the OUTFIL
statement applies to OUTFIL data sets.
4. For an ICEGENER application, NULLOUT applies to the SYSUT2 data set
if DFSORT copy is used. Note that ICEGENER passes back return code 12
instead of return code 16.
ODMAXBF
ODMAXBF=
n
nK
nM
Temporarily overrides the ODMAXBF installation option, which specifies the

maximum buffer space DFSORT can use for each OUTFIL data set. The actual
amount of buffer space used for a particular OUTFIL data set will not exceed
the ODMAXBF limit, but can be less than the limit. OUTFIL processing is
supported by the Blockset technique for sort, copy, and merge applications.
The storage used for OUTFIL processing is adjusted automatically according to
the total storage available, the storage needed for non-OUTFIL processing, and
the number of OUTFIL data sets and their attributes (for example, block size).
OUTFIL processing is subject to the ODMAXBF limit in effect and the system
storage limits (for example, IEFUSI), but not to the DFSORT storage limits (that
is, SIZE, MAXLIM, and TMAXLIM). DFSORT attempts to use storage above
16MB virtual for OUTFIL processing whenever possible.
Lowering ODMAXBF below 2M can cause performance degradation for the
application, but may be necessary if you consider the amount of storage used
for OUTFIL processing to be a problem. Raising ODMAXBF can improve
EXCPs for the application but can also increase the amount of storage needed.
n
specifies that a maximum of n bytes of buffer space is to be used for each

OUTFIL data set. If you specify less than 262144, 262144 is used. If you
specify more than 16777216, 16777216 is used.
Limit: 8 digits
nK specifies that a maximum of n times 1024 bytes of buffer space is to be

used for each OUTFIL data set. If you specify less than 256K, 256K is used.
If you specify more than 16384K, 16384K is used.
Limit: 5 digits
nM specifies that a maximum of n times 1048576 bytes of buffer space is to be
used for each OUTFIL data set. If you specify 0M, 256K is used. If you
Limit: 2 digits
OVFLO
200

OVFLO=
RC0
RC4
RC16
Temporarily overrides the OVFLO installation option, which specifies the

action to be taken by DFSORT when BI, FI, PD or ZD summary fields
overflow.
RC0
code of 0 and continue processing when summary fields overflow. The pair
of records involved in a summary overflow is left unsummed and neither
record is deleted. Summary overflow does not prevent further summation.
RC4
code of 4 and continue processing when summary fields overflow. The pair
of records involved in a summary overflow is left unsummed and neither
record is deleted. Summary overflow does not prevent further summation.
RC16
a return code of 16 when summary fields overflow.
Note: The return code of 0 or 4 set for summary overflow can be overridden
by a higher return code set for some other reason.
PAD
PAD=
RC0
RC4
RC16
Temporarily overrides the PAD installation option, which specifies the action to
be taken by DFSORT when the SORTOUT LRECL is larger than the
padding.
RC0
RC4
RC16
a return code of 16 when the SORTOUT LRECL is larger than the
Note:
1. The return code of 0 or 4 set for LRECL padding can be overridden by a
higher return code set for some other reason.
201

2. For an ICEGENER application, the GNPAD value is used and the PAD
value is ignored.
3. For some LRECL padding situations (for example, a tape work data set
sort), DFSORT issues ICE043A and terminates with a return code of 16. The
PAD value has no effect in these cases.
4. DFSORT does not check for LRECL padding if:
a. A SORTIN DD (sort/copy), SORTINnn DD (merge) or SORTOUT DD is
not present
b. A SORTIN DD (sort/copy), SORTINnn DD (merge) or SORTOUT DD
specifies a VSAM data set.
5. DFSORT does not check OUTFIL data sets for LRECL padding.
RESALL
RESALL=
n
nK
nM
Temporarily overrides the RESALL installation option, which specifies the

number of bytes to be reserved in a REGION for system use. Usually, only 4K
bytes (the standard default) of main storage must be available in a region for
system use. However, in some cases, this may not be enough; for example, if
your installation does not have BSAM/QSAM modules resident, you have user
exits that open data sets, or you have COBOL exits. RESALL is used only
when MAINSIZE/SIZE=MAX is in effect.
RESALL applies only to the amount of main storage below 16MB virtual. The
ARESALL option applies to the amount of main storage above 16MB virtual.
n
specifies that n bytes of storage are to be reserved. If you specify less than
4096, 4096 is used.
Limit: 8 digits.
nK specifies that n times 1024 bytes of storage are to be reserved. If you

specify less than 4K, 4K is used.
Limit: 5 digits.
nM specifies that n times 1048576 bytes of storage are to be reserved. If you
specify 0M, 4K is used.
Limit: 2 digits.
Tip: A better way to reserve the required storage for user exits activated by
the MODS statement is to use the m parameter of the MODS statement.
RESET or NORESET
202
RESET
NORESET

Temporarily overrides the RESET installation option, which specifies whether
DFSORT should process a VSAM output data set defined with REUSE as a
NEW or MOD data set.
RESET
REUSE as a NEW data set. The high-used RBA is reset to zero and the
output data set is effectively treated as an initially empty cluster.
NORESET
REUSE as a MOD data set. The high-used RBA is not reset and the output
data set is effectively treated as an initially non-empty cluster.
Note: A VSAM output data set defined without REUSE is processed as a MOD
data set.
RESINV
RESINV=
n
nK
nM
Temporarily overrides the RESINV installation option, which specifies the

number of bytes to be reserved in a REGION for the invoking program.
RESINV is used only when DFSORT is dynamically invoked and
MAINSIZE/SIZE=MAX is in effect.
RESINV applies only to the amount of main storage below 16MB virtual. The
ARESINV option applies to the amount of main storage above 16MB virtual.
This extra space is usually required for data handling by the invoking program
or user exits while DFSORT is running (as is the case with some PL/I- and
COBOL- invoked sort applications). Therefore, if your invoking program's user
exits do not perform data set handling, you do not need to specify this
parameter. The reserved space is not meant to be used for the invoking
program's executable code.
The amount of space required depends upon what routines you have, how the
data is stored, and which access method you use.
n

Limit: 8 digits

Limit: 5 digits
nM specifies n times 1048576 bytes of main storage are to be reserved.
Limit: 2 digits.
Tip: A better way to reserve the required storage for user exits activated by
the MODS statement is to use the m parameter of the MODS statement.
SDB
203

SDB=
LARGE
YES
INPUT
NO
Temporarily overrides the SDB installation option, which specifies whether

DFSORT should use the system-determined optimum block size for output
data sets when the block size is specified as zero or defaulted to zero.
System-determined block size applies to both SMS-managed and
non-SMS-managed data sets and results in the most efficient use of space for
the device on which the output data set resides.
DFSORT can select system-determined optimum block sizes greater than 32760
bytes for tape output data sets.
If you want DFSORT to use system-determined block sizes for disk and tape
output data sets, specify one of the following values:
v SDB=LARGE if you want to allow DFSORT to select tape output block sizes
greater than 32760 bytes.
v SDB=YES (or its alias SDB=SMALL) if you want DFSORT to select tape
output block sizes less than or equal to 32760 bytes.
v SDB=INPUT if you want to allow DFSORT to select tape output block sizes
greater than 32760 bytes only when tape input data sets with block sizes
greater than 32760 bytes are used.
DFSORT will not select a tape output block size greater than the BLKSZLIM in
effect. In particular, if a default BLKSZLIM of 32760 is in effect, DFSORT will
not select a tape output block size greater than 32760 bytes. Therefore, in order
to allow DFSORT to select tape output block sizes greater than 32760 bytes for
particular jobs, you may need to ensure that your JCL or data class supplies
appropriately large BLKSZLIM values (for example, 1GB) for those jobs.
If you don't want DFSORT to use system-determined block sizes, specify
SDB=NO (not recommended as an installation option).
LARGE
for an output data set when its block size is zero. SDB=LARGE allows
DFSORT to select a block size greater than 32760 bytes for a tape output
data set, when appropriate. A larger tape block size can improve elapsed
time and tape utilization, but you must ensure that applications which
subsequently use the resulting tape data set can handle larger block sizes.
DFSORT selects the system-determined optimum block size as follows:
v For a disk output data set, the optimum block size for the device used is
selected based on the obtained or derived RECFM and LRECL for the
output data set. The maximum block size for disk output data sets is
32760 bytes.
v For a tape output data set, the optimum block size is selected based on
the obtained or derived RECFM and LRECL for the output data set, as
shown in Table 37.
Table 37. SDB=LARGE Block Sizes for Tape Output Data Sets
204
RECFM
BLKSIZE is set to:
F or FS
LRECL
FB or FBS
Highest possible multiple of LRECL that is

less than or equal to the optimum block size
for the device, subject to the BLKSZLIM in
effect.

Table 37. SDB=LARGE Block Sizes for Tape Output Data Sets (continued)
RECFM
BLKSIZE is set to:
V, D, VS, or DS
LRECL + 4
VB, DB, VBS, or DBS
Optimum block size for the device, subject

to the BLKSZLIM in effect.
DFSORT uses the system-determined optimum block size for the output
data set in most cases when the block size is zero. However, the following
conditions prevent DFSORT from using the system-determined block size:
v Output data set block size is available (that is, non-zero) in the JFCB
(disk or tape) or format 1 DSCB (disk) or tape label (only for
DISP=MOD with AL, SL, or NSL label, when appropriate)
v Output is a spool, dummy, VSAM, or unmovable data set, or a z/OS
UNIX file.
v The output data set is on tape with a label type of AL
v DFSORT's Blockset technique is not selected.
In the previous cases, DFSORT uses the specified block size, or determines
an appropriate (though not necessarily optimum) block size for the output
data set. The selected block size is limited to 32760 bytes.
YES
selected block size to a maximum of 32760 bytes. See the discussion of
SDB=LARGE for more information; the only difference between
SDB=LARGE and SDB=YES is that SDB=LARGE allows block sizes greater
than 32760 bytes for tape output data sets, whereas SDB=YES does not.
INPUT
selected block size to a maximum of 32760 bytes if the input block size is
less than or equal to 32760 bytes. Thus, SDB=INPUT works like
SDB=LARGE if the input block size is greater than 32760 bytes (only
possible for tape input data sets) and works like SDB=YES if the input
block size is less than or equal to 32760 bytes. See the discussions of
SDB=LARGE and SDB=YES for more information.
NO specifies that DFSORT is not to use the system-determined optimum block
size. When the output data set block size is zero, DFSORT selects an
appropriate (though not necessarily optimum) block size for the output
data set based on the obtained or derived output or input attributes.
SDB=NO limits the selected block sizes to a maximum of 32760 bytes.
SDB=NO works like SDB=YES if the input block size is greater than 32760
bytes (only possible for tape input data sets). See the discussion of
SDB=YES for more information.
Note:
1. SDB=NO does not prevent the use of system-determined block size for the
output data set at allocation or in other cases where the output data set
block size is set before DFSORT gets control.
2. When DFSORT uses system-determined block size, the selected output data
set block size may be different from the block size selected previously.
205

Applications that require a specific output data set block size should be
changed to specify that block size explicitly.
3. SDB and SDB=SMALL can be used instead of SDB=YES. NOSDB can be
used instead of SDB=NO.
SIZE
See FILSZ.
SKIPREC
SKIPREC=z
Specifies the number of records z you want to skip (delete) before starting to
sort or copy the input data set. SKIPREC is usually used if, on a preceding
DFSORT run, you have processed only part of the input data set.
An application with an input data set that exceeds intermediate storage
capacity usually terminates unsuccessfully. However, for a tape work data set
sort, you can use a routine at E16 (as described in Chapter 5, Using your own
user exit routines, on page 487) to instruct the program to sort only those
records already read in. It then prints a message giving the number of records
sorted. You can use SKIPREC in a subsequent sort run to bypass the
previously-sorted records, sort only the remaining records, and then merge the
output from different runs to complete the application.
z
specifies the number of records to be skipped.

Note:
1. SKIPREC applies only to records read from SORTIN (not from E15
routines). (See Figure 2 on page 9.)
2. If SKIPREC=0 is in effect, SKIPREC is not used.
3. You may want to consider using the STARTREC parameter of the OUTFIL
statement as an alternative to using SKIPREC.
SMF
SMF=
SHORT
FULL
NO
Temporarily overrides the SMF installation option, which specifies whether a

DFSORT SMF record is to be produced as described in z/OS DFSORT
Installation and Customization.
SHORT
specifies that DFSORT is to produce a short SMF type-16 record for a
successful run. The short SMF record does not contain record-length
distribution statistics or data set sections.
206

FULL
specifies that DFSORT is to produce a full SMF type-16 record for a
successful run. The full SMF record contains the same information as the
short record, as well as record-length distribution and data set sections, as
appropriate.
NO specifies that DFSORT is not to produce an SMF type-16 record for this
run.
Note:
1. SMF is processed only if it is passed on the OPTION control statement in
an extended parameter list or in DFSPARM.
2. SMF=FULL can degrade performance for a variable-length record
application.
3. The DFSORT SVC is called to write SMF type-16 records. If SMF=SHORT
or SMF=FULL is in effect, the correct DFSORT SVC for this release must be
loaded in LPA or MLPA.
SOLRF or NOSOLRF
SOLRF
NOSOLRF
Temporarily overrides the SOLRF installation option, which specifies whether

DFSORT should set the SORTOUT LRECL to the reformatted record length
when the SORTOUT LRECL is unknown.
SOLRF
specifies that DFSORT should use the reformatted record length for the
SORTOUT LRECL when the SORTOUT LRECL is not specified or
available. DFSORT will use one of the following for the SORTOUT LRECL,
in the order listed:
1. The SORTOUT LRECL if available from the JFCB, format 1 DSCB,
DFSMSrmm, ICETPEX, or tape label
2. The L3 length if specified in the RECORD statement
3. The OUTREC length if the OUTREC statement is specified
4. The INREC length if the INREC statement is specified
5. The L2 length if specified in the RECORD statement providing an E15
user exit is present
6. The SORTIN or SORTINnn LRECL if available from the JFCB, format 1
DSCB, DFSMSrmm, ICETPEX, or tape label
7. The L1 length in the RECORD statement
NOSOLRF
specifies that DFSORT should not use the reformatted record length for the
SORTOUT LRECL. DFSORT will use one of the following for the
SORTOUT LRECL, in the order listed:
1. The SORTOUT LRECL if available from the JFCB, format 1 DSCB,
DFSMSrmm, ICETPEX, or tape label
exit, OUTREC statement or INREC statement is present
207

user exit is present
4. The SORTIN or SORTINnn LRECL if available from the JFCB, format 1
DSCB, DFSMSrmm, ICETPEX, or tape label
5. The L1 length in the RECORD statement
Note:
1. With SOLRF in effect (the IBM-supplied default), DFSORT sets the
SORTOUT LRECL to the INREC or OUTREC record length when
appropriate, which is usually what you want when you use INREC or
OUTREC. If you want DFSORT to use the input length for the SORTOUT
LRECL even when INREC or OUTREC is present, you can use NOSOLRF,
but be aware that this can cause padding or truncation of the reformatted
records, or termination.
2. CAOUTREC can be used instead of SOLRF.
SORTDD
SORTDD=cccc
Specifies a four-character prefix for the ddnames to be used when you

dynamically invoke DFSORT more than once in a program step. The four
characters replace SORT in the following ddnames: SORTIN, SORTOUT,
SORTINn, SORTINnn, SORTOFd, SORTOFdd, SORTWKd, SORTWKdd,
SORTJNF1, SORTJNF2 and SORTCNTL. This allows you to use a different set
of ddnames for each call to DFSORT.
cccc
Specifies a four-character prefix. The four characters must all be
alphanumeric or national ($, #, or @). The first character must be
alphabetic. The first three characters must not be SYS.
For example, if you use ABC# as replacement characters, DFSORT uses DD
statements ABC#IN, ABC#CNTL, ABC#WKdd, and ABC#OUT instead of
SORTIN, SORTCNTL, SORTWKdd, and SORTOUT.
Note:
1. SORTDD is processed only if it is passed on the OPTION control statement
in an extended parameter list, or in DFSPARM.
2. If both SORTIN=ddname and SORTDD=cccc are specified, ddname is used
for DFSORT input.
3. If both SORTOUT=ddname and SORTDD=cccc are specified, ddname is
used for DFSORT output.
4. If both MERGEIN=(ddname1,ddname2,...) and SORTDD=cccc are specified,
ddname1, ddname2, and so on are used for DFSORT input.
Default: SORT. See Appendix B, Specification/override of DFSORT options,
on page 863 for full override details.
SORTIN
208

SORTIN=ddname
Specifies a ddname to be associated with the SORTIN data set. This allows you
to dynamically invoke DFSORT more than once in a program step, passing a
different ddname for each input data set.
The ddname can be 1 through 8 characters, but must be unique within the job
step. Do not use ddnames reserved for use by DFSORT, such as ccccWKd,
ccccWKdd, ccccDKd, or ccccDKdd , where cccc is the specified or defaulted
value for the SORTDD operand and d is any character.
Note:
1. SORTIN is processed only if it is passed on the OPTION control statement
in an extended parameter list, or in DFSPARM.
2. If both SORTIN=ddname and SORTDD=cccc are specified, ddname is used
for the input file. The same ddname cannot be specified for SORTIN and
SORTOUT, or for MERGEIN and SORTOUT.
3. If SORTIN is used for a tape work data set sort, DFSORT terminates.
Default: SORTIN, unless SORTDD=cccc is specified in which case ccccIN is the
default. See Appendix B, Specification/override of DFSORT options, on page
863 for full override details.
SORTOUT
SORTOUT=ddname
Specifies a ddname to be associated with the SORTOUT data set or with an

OUTFIL data set for which FNAMES and FILES are not specified. This allows
you to dynamically invoke DFSORT more than once in a program step, passing
a different ddname for each output data set.
The ddname can be 1 through 8 characters, but must be unique within the job
step. Do not use ddnames reserved for use by DFSORT, such as ccccWKd,
ccccWKdd, ccccDKd, or ccccDKdd , where cccc is the specified or defaulted
value for the SORTDD operand and d is any character.
Note:
1. SORTOUT is processed only if it is passed on the OPTION control
2. If both SORTOUT=ddname and SORTDD=cccc are specified, ddname is
used for the output file. The same ddname cannot be specified for SORTIN
and SORTOUT, or for MERGEIN and SORTOUT.
3. If SORTOUT is specified for a conventional merge or for a tape work data
set sort, DFSORT terminates.
Default: SORTOUT, unless SORTDD=cccc is specified, in which case ccccOUT is
the default. See Appendix B, Specification/override of DFSORT options, on
page 863 for full override details.
SPANINC
SPANINC=
RC0
RC4
RC16
209

Temporarily overrides the SPANINC installation option, which specifies the
action to be taken by DFSORT when one or more incomplete spanned records
are detected in a variable spanned input data set.
RC0
code of 0 and eliminate all incomplete spanned records it detects. Valid
records will be recovered.
RC4
code of 4 and eliminate all incomplete spanned records it detects. Valid
records will be recovered.
RC16
a return code of 16 when an incomplete spanned record is detected.
Note:
1. The return code of 0 or 4 set for incomplete spanned records can be
overridden by a higher return code set for some other reason.
2. In cases where a spanned record cannot be properly assembled (for
example, it has a segment length less than 4 bytes), DFSORT issues
ICE141A and terminates with a return code of 16. The SPANINC value has
no effect in these cases.
STOPAFT
STOPAFT=n
Specifies the maximum number of records (n) you want accepted for sorting or
copying (that is, read from SORTIN or inserted by E15 and not deleted by
SKIPREC, E15, or the INCLUDE/OMIT statement). When n records have been
accepted, no more records are read from SORTIN; E15 continues to be entered
as if EOF were encountered until a return code of 8 is sent, but no more
records are inserted. If end-of-file is encountered before n records are accepted,
only those records accepted up to that point are sorted or copied.
n
specifies the maximum number of records to be accepted.


SZERO or NOSZERO
SZERO
NOSZERO
Temporarily overrides the SZERO installation option, which specifies whether

DFSORT should treat numeric -0 and +0 values as signed (that is, different) or
unsigned (that is, the same) for collation, comparisons, editing and
conversions, minimums and maximums. The following DFSORT control
210

statements are affected by this option: INCLUDE, INREC, MERGE, OMIT,
OUTFIL, OUTREC and SORT.
SZERO
specifies that DFSORT should treat numeric zero values as signed. -0 and
+0 are treated as different values, that is, -0 is treated as a negative value
and +0 is treated as a positive value. SZERO affects DFSORT processing of
numeric values as follows:
v For collation of SORT and MERGE fields, -0 collates before +0 in
ascending order and after +0 in descending order.
v For comparisons of INCLUDE, OMIT, and OUTFIL compare fields and
constants, -0 compares as less than +0.
v For editing and conversions of INREC, OUTREC, and OUTFIL
reformatting fields, decimal constants, and the results of arithmetic
expressions, -0 is treated as negative and +0 is treated as positive.
v For minimums and maximums of OUTFIL TRAILERx fields, -0 is treated
as negative and +0 is treated as positive.
NOSZERO
specifies that DFSORT should treat numeric zero values as unsigned. -0
and +0 are treated as the same value, that is, -0 and +0 are both treated as
positive values. NOSZERO affects DFSORT processing of numeric values
as follows:
v For collation of SORT and MERGE fields, -0 collates equally with +0.
v For comparisons of INCLUDE, OMIT and OUTFIL compare fields and
constants, -0 compares as equal to +0.
v For editing and conversions of INREC, OUTREC, and OUTFIL
reformatting fields, decimal constants, and the results of arithmetic
expressions, -0 and +0 are treated as positive.
v For minimums and maximums of OUTFIL TRAILERx fields, -0 and +0
are treated as positive.
Note: OPTION SZERO or OPTION NOSZERO is ignored for OUTFIL
INCLUDE and OMIT, and INREC, OUTREC, and OUTFIL WHEN, BEGIN
and END, if the OPTION statement is "found" after the INREC, OUTREC,
or OUTFIL statement. To avoid this, specify SZERO or NOSZERO as an
EXEC/DFSPARM PARM option, or in an OPTION statement before the
INREC, OUTREC, or OUTFIL statement in the same source, for example:
//SYSIN DD *
OPTION NOSZERO,COPY
OUTFIL INCLUDE=(...)
/*
Default: Usually, the installation default. See Appendix B, Specification/

TRUNC
TRUNC=
RC0
RC4
RC16
Temporarily overrides the TRUNC installation option, which specifies the

action to be taken by DFSORT when the SORTOUT LRECL is smaller than the
211

truncation.
RC0
RC4
RC16
a return code of 16 when the SORTOUT LRECL is smaller than the
Note:
1. The return code of 0 or 4 set for LRECL truncation can be overridden by a
higher return code set for some other reason.
2. For an ICEGENER application, the GNTRUNC value is used and the
TRUNC value is ignored.
3. For some LRECL truncation situations (for example, a tape work data set
sort), DFSORT issues ICE043A and terminates with a return code of 16. The
TRUNC value has no effect in these cases.
4. DFSORT does not check for LRECL truncation if:
a. A SORTIN DD (sort/copy), SORTINnn DD (merge) or SORTOUT DD is
not present
b. A SORTIN DD (sort/copy), SORTINnn DD (merge) or SORTOUT DD
specifies a VSAM data set.
5. DFSORT does not check OUTFIL data sets for LRECL truncation.
USEWKDD
USEWKDD
Temporarily overrides the DYNAUTO=IGNWKDD option, which specifies that

dynamic work data sets are used even if SORTWKdd DD statements are
present. This option allows JCL SORTWKdd data sets to be used rather than
deallocated.
Note: USEWKDD is processed only if it is passed on the OPTION control
Default: None, optional. See Appendix B, Specification/override of DFSORT
Applicable Function: See Appendix B, Specification/override of DFSORT
VERIFY or NOVERIFY
212

VERIFY
NOVERIFY
Temporarily overrides the VERIFY installation option, which specifies whether

sequence checking of the final output records must be performed.
VERIFY
specifies that sequence checking is performed.
NOVERIFY
specifies that sequence checking is not performed.
Note:
1. Using VERIFY can degrade performance.
2. SEQ=YES can be used instead of VERIFY, SEQ=NO can be used instead of
NOVERIFY.
VLLONG or NOVLLONG
VLLONG
NOVLLONG
Temporarily overrides the VLLONG installation option, which specifies

whether DFSORT is to truncate "long" variable-length output records. A long
output record is one whose length is greater than the LRECL of the SORTOUT
or OUTFIL data set it is to be written to.
VLLONG is not meaningful for fixed-length output record processing.
VLLONG
specifies that DFSORT truncates long variable-length output records to the
LRECL of the SORTOUT or OUTFIL data set.
NOVLLONG
specifies that DFSORT terminates if a long variable-length output record is
found.
Note:
1. VLLONG should not be used unless you want the data at the end of long
variable-length output records to be truncated for your DFSORT
application; inappropriate use of VLLONG can result in unwanted loss of
data.
2. VLLONG can be used to truncate long OUTFIL data records, but has no
effect on long OUTFIL header or trailer records.
VLSCMP or NOVLSCMP
VLSCMP
NOVLSCMP
Temporarily overrides the VLSCMP installation option, which specifies

213

whether DFSORT is to pad "short" variable-length INCLUDE/OMIT compare
fields with binary zeros. A short field is one where the variable-length record is
too short to contain the entire field, that is, the field extends beyond the record.
VLSCMP and NOVLSCMP apply to the INCLUDE and OMIT statements and
to the INCLUDE and OMIT parameters of the OUTFIL statement.
The compare fields are only padded temporarily for testing; they are not
actually changed for output.
VLSCMP is not meaningful for fixed-length record processing.
The settings for VLSCMP/NOVLSCMP and VLSHRT/NOVLSHRT provide
three levels of processing for short INCLUDE/OMIT fields in the following
hierarchy:
1. VLSCMP allows all of the INCLUDE/OMIT comparisons to be performed
even if some fields are short. Because short fields are padded with binary
zeros, comparisons involving short fields are false (unless a test against
binary zero is relevant, as discussed later in this section). Comparisons
involving non-short fields can be true or false.
2. NOVLSCMP and VLSHRT treat the entire INCLUDE/OMIT logical
expression as false if any field is short. Thus comparisons involving
non-short fields are ignored if any comparison involves a short field.
3. NOVLSCMP and NOVLSHRT result in termination if any field is short.
To illustrate how this works, suppose the following INCLUDE statement is
used:
INCLUDE COND=(6,1,CH,EQ,C1,OR,70,2,CH,EQ,CT1)
If a variable-length input record has a length less than 71 bytes, the field at
bytes 70-71 is short and the following occurs:
v With VLSCMP, the record is included if byte 6 of the input record is C'1' or
omitted if byte 6 is not C'1'. The comparison of bytes 70-71 equal to C'T1' is
false because bytes 70-71 contain either X'hh00' (for a record length of 70
bytes) or X'0000' (for a record length of less than 70 bytes). The comparison
involving the non-short field is performed even though a short field is
present.
v With NOVLSCMP and VLSHRT, the record is omitted because any short
field makes the entire logical expression false. The comparison involving the
non-short field is not performed because a short field is present.
v With NOVLSCMP and NOVLSHRT, DFSORT terminates because any short
field results in termination.
In general, comparisons involving short fields are false with VLSCMP.
However, if a binary zero value is relevant to the comparison, the use of binary
zeros for padding might make the comparison true. For example, suppose the
following INCLUDE statement is used:
INCLUDE COND=(21,2,CH,EQ,CJX,OR,
(55,2,CH,EQ,58,2,CH,AND,
70,1,BI,LT,X08))
If a variable-length input record has a length less than 70 bytes, the field at
byte 70 is short and is padded to X'00'. This makes the comparison of byte 70
less than X'08' true even though byte 70 is a short field and so probably
irrelevant.
Likewise, if a variable-length record has a length less than 55 bytes, the fields
at bytes 55-56 and 58-59 are short and are each padded to X'0000', and the field
at byte 70 is short and is padded to X'00'. This makes the comparison of bytes
55-56 equal to 58-59 true and the comparison of byte 70 less than X'08' true
even though all three fields are short and probably irrelevant.
214

In such cases where padding of short fields with binary zeros may result in
unwanted true comparisons, you can get the result you want by adding an
appropriate check of the record length to the INCLUDE/OMIT logical
expression, such as:
INCLUDE COND=(21,2,CH,EQ,CJX,OR,
(1,2,BI,GE,X0046,AND,
55,2,CH,EQ,58,2,CH,AND,
70,1,BI,LT,X08))
Now the comparisons involving bytes 55-56, 58-59 and 70 can only be true for
records that are 70 bytes (X'0046') or longer. Thus, the irrelevant comparisons
involving short fields are eliminated.
Keep in mind that short compare fields are padded with zeros when VLSCMP
is in effect and code your INCLUDE/OMIT logical expressions to allow for
that or even take advantage of it.
VLSCMP
specifies that short variable-length compare fields are padded with binary
zeros.
NOVLSCMP
specifies that short variable-length compare fields are not padded.
VLSHRT or NOVLSHRT
VLSHRT
NOVLSHRT
Temporarily overrides the VLSHRT installation option, which specifies whether

DFSORT is to continue processing if a "short" variable-length SORT/MERGE
control field, INCLUDE/OMIT compare field, or SUM summary field is found.
A short field is one where the variable-length record is too short to contain the
entire field, meaning that the field extends beyond the record. VLSHRT applies
to the SORT, MERGE, INCLUDE, OMIT and SUM statements, and to the
INCLUDE and OMIT parameters of the OUTFIL statement.
VLSHRT processing is not meaningful for fixed-length record processing.
depends on the settings for VLSCMP/NOVLSCMP and VLSHRT/NOVLSHRT.
For details, see the discussion of the VLSCMP and NOVLSCMP options.
VLSHRT
specifies that DFSORT continues processing if a short control field,
compare field or summary field is found.
NOVLSHRT
specifies that DFSORT terminates if a short control field, compare field or
summary field is found.
Note:
1. VLSHRT is not used if an INREC or OUTREC statement is specified, if you
have an EFS01 or EFS02 routine, or if locale processing is used for SORT or
MERGE fields. Note that none of these situations prevents the use of
VLSCMP.
215

2. Unlike the OUTREC statement, the OUTREC , BUILD, OVERLAY,
FINDREP or IFTHEN parameter of the OUTFIL statement does not force
NOVLSHRT. Thus, you can use VLSHRT with OUTFIL to eliminate records
with the INCLUDE or OMIT parameter and reformat the remaining records
with the OUTREC, , BUILD, OVERLAY, FINDREP or IFTHEN parameter. If
a short OUTFIL OUTREC or BUILD field is found, DFSORT terminates
(even if VLSHRT is in effect) unless the VLFILL=byte parameter of OUTFIL
is specified. If a short OUTFIL OVERLAY, FINDREP or IFTHEN field is
found, DFSORT pads the missing bytes with blanks so it can be processed.
3. If VLSHRT is in effect and Blockset is selected:
v DFSORT pads short SORT or MERGE control fields with binary zeros,
thus making the order predictable for records with equal control fields of
different lengths. The control fields are only padded temporarily for
collation; they are not actually changed for output. Padding may increase
the amount of work space required.
v Records with short SUM summary fields are excluded from summation;
that is, if either one of a pair of records being summed has a short SUM
field, the records are left unsummed and neither record is deleted.
4. If VLSHRT is in effect and Blockset is not selected:
v DFSORT terminates if the first byte of the first (major) SORT or MERGE
control field is not included in the record.
v DFSORT does not pad short SORT or MERGE control fields, thus making
the order unpredictable for records with equal control fields of different
lengths.
v In certain cases, VLSHRT is not used because of the number and position
of the SORT or MERGE control fields.
v EQUALS is not used.
Tip: You can use a SORTDIAG DD statement to force message ICE800I,
which gives a code indicating why Blockset could not be used.
VSAMEMT or NVSAMEMT
VSAMEMT
NVSAMEMT
Temporarily overrides the VSAMEMT installation option, which specifies

whether DFSORT should accept an empty VSAM input data set.
VSAMEMT
specifies that DFSORT accepts an empty VSAM input data set and
processes it as having zero records.
NVSAMEMT
specifies that DFSORT terminates if an empty VSAM input data set is
found.
216

VSAMIO or NOVSAMIO
VSAMIO
NOVSAMIO
Temporarily overrides the VSAMIO installation option, which specifies

whether DFSORT should allow a VSAM data set defined with REUSE to be
sorted in-place.
VSAMIO
specifies that DFSORT can use the same VSAM data set for input and
output when all of the following conditions are met:
v The application is a sort.
v RESET is in effect.
v The VSAM data set was defined with REUSE.
These conditions ensure that the VSAM data set is processed as NEW for
output and will contain the sorted input records, that is, it will be sorted
in-place.
DFSORT terminates if the same VSAM data set is specified for input and
output and any of the previous conditions are not met.
NOVSAMIO
specifies that DFSORT terminates if the same VSAM data set is used for
input and output.
WRKREL or NOWRKREL
WRKREL
NOWRKREL
Temporarily overrides the WRKREL installation option, which specifies

whether unused temporary SORTWKdd data set space will be released.
WRKREL
specifies that unused space is released.
NOWRKREL
specifies that unused space is not released.
Note:
1. If you have dedicated certain volumes for SORTWKdd data sets, and you
do not want unused temporary space to be released, you should specify
NOWRKREL.
2. If WRKREL is in effect, DFSORT releases space for the SORTWKdd data
sets just prior to termination. Space is released only for those SORTWKdd
data sets that were used for the sort application.
3. RLS=0 can be used instead of NOWRKREL. RLS=n (n greater than 0) can
be used instead of WRKREL.
217

WRKSEC or NOWRKSEC
WRKSEC
NOWRKSEC
Temporarily overrides the WRKSEC installation option, which specifies

whether DFSORT uses automatic secondary allocation for temporary JCL
SORTWKdd data sets.
WRKSEC
SORTWKdd data sets is used and that 25 percent of the primary allocation
will be used as the secondary allocation.
NOWRKSEC
SORTWKdd data sets is not used.
Note: SEC=0 can be used instead of NOWRKSEC. SEC=n (n greater than 0)
can be used instead of WRKSEC.
Y2PAST
Y2PAST=
s
f
Temporarily overrides the Y2PAST installation option, which specifies the

sliding (s) or fixed (f) century window. The century window is used with
DFSORT's Y2 formats to correctly interpret two-digit year data values as
four-digit year data values.
s
specifies the number of years DFSORT is to subtract from the current year
to set the beginning of the sliding century window. Because the Y2PAST
value is subtracted from the current year, the century window slides as the
current year changes. For example, Y2PAST=81 would set a century
window of 1925-2024 in 2006 and 1926-2025 in 2007. s must be a value
between 0 and 100.
specifies the beginning of the fixed century window. For example,

Y2PAST=1962 would set a century window of 1962-2061. f must be a value

Y2PAST=value.
ZDPRINT or NZDPRINT
218

ZDPRINT
NZDPRINT
Temporarily overrides the ZDPRINT installation option, which specifies

whether positive zoned-decimal (ZD) fields resulting from summing must be
converted to printable numbers (that is, whether the zone of the last digit
should be changed from a hexadecimal C to a hexadecimal F). See SUM
control statement on page 451 for further details on the use of ZDPRINT and
NZDPRINT.
ZDPRINT
means convert positive ZD summation results to printable numbers. For
example, change hexadecimal F3F2C5 (prints as 32E) to F3F2F5 (prints as
325).
NZDPRINT
means do not convert positive ZD summation results to printable numbers.
Note: ZDPRINT=YES can be used instead of ZDPRINT. ZDPRINT=NO can be
used instead of NZDPRINT.
Applicable Function: See Appendix B, Specification/override of DFSORT
Aliases for OPTION statement options

For compatibility reasons, the following OPTION statement options can be
specified by using the aliases listed in Table 38. See the indicated OPTION
statement options for complete details.
Table 38. Aliases for OPTION Statement Options
OPTION Statement Option
CAOUTREC
SOLRF
CENTURY=value
Y2PAST=value
CENTWIN=value
Y2PAST=value
CHKPT
CKPT
CORE=value
MAINSIZE=value
HIPRLIM=value
HIPRMAX=value
L5=value
AVGRLEN=value
NOSDB
SDB=NO
PRINT=value
MSGPRT=value
RLS=n
WRKREL
RLS=0
NOWRKREL
SDB
SDB=YES
SDB=SMALL
SDB=YES
SEC=n
WRKSEC
SEC=0
NOWRKSEC
SEQ=YES
VERIFY
SEQ=NO
NOVERIFY
ZDPRINT=YES
ZDPRINT
219

Table 38. Aliases for OPTION Statement Options (continued)
OPTION Statement Option
ZDPRINT=NO
NZDPRINT
Specifying DFSORT options or COPYexamples

Example 1
OPTION SIZE=50000,SKIPREC=5,EQUALS,DYNALLOC
FIELDS
The control field begins on the first byte of each record in the input data set, is
20 bytes long, contains character data, and is to be sorted in ascending order.
SIZE
The data set to be sorted contains 50000 records.
SKIPREC
Five records are skipped (deleted) before starting to process the input data set.
EQUALS
The sequence of records that collate identically is preserved from input to
output.
DYNALLOC
Two data sets (by default) are allocated on SYSDA (by default). The space on
the data set is calculated using the SIZE value in effect.
Example 2
SORT
FIELDS=(1,2,CH,A),CKPT
OPTION EQUALS,NOCHALT,NOVERIFY,CHECK
FIELDS
CKPT
DFSORT takes checkpoints during this run.
Note: CKPT is ignored if the Blockset technique is used. If checkpoints are
required, you must bypass the Blockset technique by specifying the
NOBLKSET option, or by specifying installation option IGNCKPT=NO.
However, functions such as OUTFIL, which are supported only by the Blockset
technique, cannot be used if the Checkpoint/Restart facility is used.
EQUALS
The sequence of records that collate identically is preserved from input to
output.
NOCHALT
Only AQ fields are translated through the ALTSEQ translate table. If
installation option CHALT=YES was specified, then NOCHALT temporarily
overrides it.
NOVERIFY
No sequence check is performed on the final output records.
CHECK
The record count is checked at the end of program processing.
220
Example 3
OPTION FILSZ=50,SKIPREC=5,DYNALLOC=3390
SORT FIELDS=(1,2,CH,A),SKIPREC=1,SIZE=200,DYNALLOC=(3380,5)
This example shows how parameters specified on the OPTION control statement
override those specified on the SORT control statement, regardless of the order of
the two statements.
FILSZ
DFSORT expects 50 records on the input data set. (Note that there is a
difference in meaning between FILSZ and SIZE and that the OPTION
specification of FILSZ is used in place of SIZE.)
SKIPREC
DFSORT causes five records from the beginning of the input file to be skipped.
(SKIPREC=1 on the SORT statement is ignored.)
DYNALLOC
DFSORT allocates two work data sets (by default) on an IBM 3390.
FIELDS
Example 4
OPTION NOBLKSET
NOBLKSET
DFSORT does not use the Blockset technique for a sort or merge.
Example 5
OPTION STOPAFT=100
STOPAFT
DFSORT accepts 100 records before sorting or copying.
Example 6
OPTION RESINV=32000,MSGPRT=NONE,
MSGDDN=SORTMSGS,SORTDD=ABCD,SORTIN=MYINPUT,
SORTOUT=MYOUTPUT,NOLIST
This example illustrates the parameters RESINV, MSGPRT, MSGDDN, SORTDD,

SORTIN, SORTOUT, and NOLIST, and the actions taken when these parameters
are supplied on an OPTION statement read from the SYSIN data set or the
SORTCNTL data set. The parameters are recognized, but not used.
RESINV
32000 bytes of storage are reserved for the user.
MSGPRT=NONE
The keyword is ignored, and messages are printed according to the installation
default.
MSGDDN=SORTMSGS
The keyword is ignored, and all messages are written to the SYSOUT data set.
SORTDD=ABCD
The keyword is ignored, and the standard prefix SORT is used.
SORTIN=MYINPUT
The keyword is ignored, and the ddname SORTIN is used to reference the
input data set.
221

SORTOUT=MYOUTPUT
The keyword is ignored, and the ddname SORTOUT is used to reference the
output data set.
NOLIST
The keyword is ignored, and control statements are printed according to the
installation defaults.
Example 7
OPTION RESINV=32000,MSGPRT=CRITICAL
MSGDDN=SORTMSGS,SORTDD=ABCD,SORTIN=MYINPUT,
SORTOUT=MYOUTPUT,NOLIST
This example illustrates keywords RESINV, MSGPRT, MSGDDN, SORTDD,

SORTIN, SORTOUT, and NOLIST and the actions taken when these keywords are
supplied on the OPTION control statement passed by DFSPARM. These options
can also be passed in an extended parameter list, but must be coded as one
contiguous statement without continuation lines.
RESINV
32000 bytes of storage are reserved for the user.
MSGPRT=CRITICAL
Only critical messages are printed on the message data set.
MSGDDN=SORTMSGS
Messages are written to the SORTMSGS data set.
SORTDD=ABCD
SORT uses ABCD as a prefix for all sort names.
SORTIN=MYINPUT
The ddname MYINPUT is used to reference the input data set.
SORTOUT=MYOUTPUT
The ddname MYOUTPUT is used to reference the output data set.
NOLIST
Control statements are not printed.
Example 8
SORT
FIELDS=(3,4,CH,A)
OPTION COPY,SKIPREC=10,CKPT
MODS E15=(E15,1024,MODLIB),E35=(E35,1024,MODLIB)
SORT
The sort statement is ignored because the COPY option has been specified.
COPY
The copy processing is always done on a record-by-record basis. Each record is
therefore read from SORTIN, passed to the E15 exit, passed to the E35 exit, and
written to SORTOUT. (Contrast this with a sort, where all the records are read
from SORTIN and passed to the E15 exit before any records are passed to the
E35 exit and written to SORTOUT.)
SKIPREC
Ten records are skipped before copying starts.
CKPT
The checkpoint option is not used for copy applications.
222
Example 9
SORT
FIELDS=(5,4,CH,A)
SUM FIELDS=(12,5,ZD,25,6,ZD)
OPTION ZDPRINT
ZDPRINT
The positive summed ZD values are printable because DFSORT uses an F sign
for the last digit.
Example 10
//S1 EXEC PGM=SORT
//SYSOUT
DD SYSOUT=*
//SARA DD *
AAA FROM SARA
CCC FROM SARA
DDD FROM SARA
//MOLLY DD *
AAA FROM MOLLY
BBB FROM MOLLY
DDD FROM MOLLY
//NORA DD *
AAA FROM NORA
BBB FROM NORA
CCC FROM NORA
//SORTOUT DD SYSOUT=*
//DFSPARM DD
*
OPTION EQUALS,MERGEIN=(NORA,SARA,MOLLY)
MERGE FIELDS=(1,3,CH,A)
/*
This example illustrates the use of the alternate ddnames NORA, SARA and
MOLLY for a MERGE application instead of SORTINnn ddnames. Since EQUALS
is specified, equally collating records will be from NORA, then SARA, then
MOLLY, that is, in the order specified in the MERGEIN list. Thus, SORTOUT
contains these records:
AAA
AAA
AAA
BBB
BBB
CCC
CCC
DDD
DDD
FROM
FROM
FROM
FROM
FROM
FROM
FROM
FROM
FROM
NORA
SARA
MOLLY
NORA
MOLLY
NORA
SARA
SARA
MOLLY
If MERGEIN=(SARA,MOLLY,NORA) was specified in the MERGEIN list in

DFSPARM, SORTOUT would contain these records:
AAA
AAA
AAA
BBB
BBB
CCC
CCC
DDD
DDD
FROM
FROM
FROM
FROM
FROM
FROM
FROM
FROM
FROM
SARA
MOLLY
NORA
MOLLY
NORA
SARA
NORA
SARA
MOLLY
OUTFIL control statements
223
OUTFIL Control Statements

,
OUTFIL
FNAMES=
FILES=
ddname
,
( ddname
suffix
,
( suffix )
STARTREC=n
ENDREC=n
SAMPLE=
n
(n,m)
INCLUDE= ( logical expression )
ALL
NONE
OMIT=
( logical expression )
ALL
NONE
SAVE
ACCEPT=n
,
PARSE=( definition
,
OUTREC=
BUILD=
( item
)
,VTOF
,CONVERT
,
OVERLAY= ( item
,
FINDREP= ( item
,
IFTHEN=(clause)
IFOUTLEN=n
FTOV
VLTRIM=byte
VLTRAIL=string
REPEAT=n
SPLIT
SPLITBY=n
SPLIT1R=n
NULLOFL= RC0
RC4
RC16
LINES=n
,
HEADER1= ( item
,
TRAILER1= ( item
,
HEADER2= ( item
,
TRAILER2= ( item )
Additional Parameters for OUTFIL
224
,VLFILL=byte

Additional Parameters for OUTFIL
,
SECTIONS= ( item
NODETAIL
BLKCCH1
BLKCCH2
BLKCCT1
REMOVECC
IFTRAIL=(updates)
OUTFIL control statements allow you to create one or more output data sets for a
sort, copy, or merge application from a single pass over one or more input data
sets. You can use multiple OUTFIL statements, with each statement specifying the
OUTFIL processing to be performed for one or more output data sets. OUTFIL
processing begins after all other processing ends (that is, after processing for exits,
options, and other control statements).
OUTFILE can be used as an alias for OUTFIL.
OUTFIL statements support a wide variety of output data set tasks, including:
v Creation of multiple output data sets containing unedited or edited records from
a single pass over one or more input data sets.
v Creation of multiple output data sets containing different ranges or subsets of
records from a single pass over one or more input data sets. In addition, records
that are not selected for any subset can be saved in a separate output data set.
v Conversion of variable-length record data sets to fixed-length record data sets.
v Conversion of fixed-length record data sets to variable-length record data sets.
v A wide variety of parsing, editing, and reformatting tasks including:
The use of fixed position/length fields or variable position/length fields. For
fixed fields, you specify the starting position and length of the field directly.
For variable fields, such as delimited fields, comma separated values (CSV),
tab separated values, blank separated values, keyword separated fields,
Insertion of blanks, zeros, strings, current date, future date, past date, current
records.
Sophisticated conversion capabilities, such as find and replace, hexadecimal
display, bit display, translation of EBCDIC letters from lowercase to uppercase
or uppercase to lowercase, translation of characters from EBCDIC to ASCII
and from ASCII to EBCDIC, translation of characters using the ALTSEQ
translation table, conversion of numeric values from one format to another,
left justify or left-squeeze (remove leading blanks or all blanks and shift left),
and right-justify or right-squeeze (remove trailing blanks or all blanks and
shift right).
Sophisticated editing capabilities, such as control of the way numeric fields
are presented with respect to length, leading or suppressed zeros, thousands
separators, decimal points, leading and trailing positive and negative signs,
and so on.
225

forms.
year, Julian, Gregorian) to corresponding output date fields of another type or
Selection of a character constant, hexadecimal constant, or input field from a
lookup table for output, based on a character, hexadecimal, or bit string as
input (that is, lookup and change).
v Creation of the reformatted records in one of the following ways:
By building the entire record one item at a time.
By only overlaying specific columns.
By performing find and replace operations.
By using sophisticated conditional logic or group operations to choose how
v Highly detailed three-level (report, page, and section) reports containing a
variety of report elements you can specify (for example, current date, current
time, edited or converted page numbers, character strings, and blank lines) or
derive from the input records (for example, character fields; unedited, edited or
converted numeric input fields; edited or converted record counts; and edited or
converted totals, maximums, minimums, and averages for numeric input fields).
v Creation of multiple output records from each input record, with or without
v Repetition and sampling of data records.
v Splitting of data records in rotation among a set of output data sets.
v Updating counts and totals in an existing trailer (last) record based on the
current data records.
The parameters of OUTFIL are grouped by primary purpose as follows:
v
v FNAMES and FILES specify the ddnames of the OUTFIL data sets to be
created. Each OUTFIL data set to be created must be specifically identified using
FNAMES or FILES in an OUTFIL statement. By contrast, the SORTOUT data set
is created by default if a DD statement for it is present. The term SORTOUT
data set denotes the single non-OUTFIL output data set, but in fact, the
SORTOUT ddname can be used for an OUTFIL data set either explicitly or by
default.
If SORTOUT is identified as an OUTFIL ddname, either explicitly (for example,
via FILES=OUT) or by default (OUTFIL without FILES or FNAMES), the data
set associated with the SORTOUT ddname will be processed as an OUTFIL data
set rather than as the SORTOUT data set.
OUTFIL data sets have characteristics and requirements similar to those for the
SORTOUT data set, but there are differences in the way each is processed. The
major differences are that an E39 exit routine is not entered for OUTFIL data
sets, and that OUTFIL processing does not permit the use of the LRECL value to
pad fixed-format OUTFIL records. (DFSORT will automatically determine and
226

set an appropriate RECFM, LRECL, and BLKSIZE for each OUTFIL data set for
which these attributes are not specified or available.)
For a single DFSORT application, OUTFIL data sets can be intermixed with
respect to VSAM and non-VSAM, tape and disk, and so on. All of the data sets
specified for a particular OUTFIL statement are processed in a similar way and
thus are referred to as an OUTFIL group. (That is, you group OUTFIL data sets
that use the same operands by specifying them on a single OUTFIL statement.)
For example, the first OUTFIL statement might have an INCLUDE operand that
applies to an OUTFIL group of one non-VSAM data set on disk and another on
tape; a second OUTFIL statement might have OMIT and OUTREC operands that
apply to an OUTFIL group of one non-VSAM data set on disk and two VSAM
data sets.
Records are processed for OUTFIL as they are for SORTOUT, after all other
DFSORT processing is complete. Conceptually, you can think of an OUTFIL
input record as being intercepted at the point between being passed from an E35
exit and written to SORTOUT, although neither an E35 exit nor SORTOUT need
actually be specified with OUTFIL processing. With that in mind, see Figure 2 on
page 9 for details on the processing that occurs prior to processing the OUTFIL
input record. In particular:
Records deleted by an E15 or E35 exit, an INCLUDE, OMIT or SUM
statement, or the SKIPREC or STOPAFT parameter are not available for
OUTFIL processing
If records are reformatted by an E15 exit, an INREC or OUTREC statement, or
an E35 exit, the resulting reformatted record is the OUTFIL input record to
which OUTFIL fields must refer.
v STARTREC starts processing for an OUTFIL group at a specific OUTFIL input
record. ENDREC ends processing for an OUTFIL group at a specific OUTFIL
input record. SAMPLE selects a sample of OUTFIL input records for an OUTFIL
group using a specific interval and number of records in that interval. Separately
or together, STARTREC, ENDREC, and SAMPLE select a range of records to
which subsequent OUTFIL processing will apply.
v INCLUDE, OMIT, and SAVE select the records to be included in the data sets
of an OUTFIL group. INCLUDE and OMIT operate against the specified fields of
each OUTFIL input record to select the output records for their OUTFIL group
(all records are selected by default). SAVE selects the records that are not
selected for any other OUTFIL group.
Whereas the INCLUDE and OMIT statements apply to all input records, the
INCLUDE and OMIT parameters apply only to the OUTFIL input records for
their OUTFIL group. The INCLUDE and OMIT parameters have all of the logical
expression capabilities of the INCLUDE and OMIT statements.
v ACCEPT limits the number of OUTFIL input records accepted for processing for
an OUTFIL group. A record is accepted if it is not deleted by STARTREC,
ENDREC, SAMPLE, INCLUDE or OMIT processing. ACCEPT limits the number
of records to which subsequent OUTFIL processing will apply.
v
v PARSE, OUTREC, BUILD, OVERLAY, FINDREP, or IFTHEN reformat the
output records for an OUTFIL group. These parameters allow you to rearrange,
edit, replace, remove and change fixed position/length fields or variable
position/length fields from the OUTFIL input records and to insert blanks,
zeros, strings, current date, future date, past date, current time, sequence
numbers, decimal constants, and the results of arithmetic expressions.
OVERLAY allows you to change specific existing columns without affecting the
entire record.
227

FINDREP allows you to do various find and replace operations.
IFTHEN clauses allow you to reformat different records in different ways
according to the criteria you specify. IFOUTLEN can be used with IFTHEN
clauses to set the output LRECL.
OUTREC or BUILD gives you complete control over the items in your
reformatted records and the order in which they appear, and also allows you to
produce multiple reformatted output records from each input record, with or
without intervening blank output records.
VTOF or CONVERT can be used with OUTREC or BUILD to convert
variable-length input records to fixed-length output records.
VLFILL can be used to allow processing of variable-length input records which
are too short to contain all specified OUTREC or BUILD fields.
Whereas the FIELDS, BUILD, OVERLAY, FINDREP, and IFTHEN parameters of
the OUTREC statement apply to all input records, the OUTREC, BUILD,
OVERLAY, FINDREP, and IFTHEN parameters of the OUTFIL statement apply
only to the OUTFIL input records for its OUTFIL group. In addition, the
OUTREC and BUILD parameters of the OUTFIL statement support the forward
slash (/) separator for creating blank records and new records, whereas the
FIELDS and BUILD parameters of the OUTREC statement do not.
v
v FTOV can be used to convert fixed-length input records to variable-length
output records. FTOV can be used with or without OUTREC , BUILD,
OVERLAY, FINDREP, or IFTHEN.
v VLTRIM can be used to remove the trailing bytes with a specified value, such
as blanks, binary zeros or asterisks, from variable-length records. VLTRIM can be
used with or without FTOV.
v VLTRAIL can be used to insert a character string or hexadecimal string at the
end of variable-length records. VLTRAIL can be used with or without FTOV or
VLTRIM.
v REPEAT can be used to repeat each output record a specified number of times.
v SPLIT, SPLITBY, or SPLIT1R splits the output records in rotation among the
data sets of an OUTFIL group.
With SPLIT, the first output record is written to the first OUTFIL data set in the
group, the second output record is written to the second data set, and so on.
When each OUTFIL data set has one record, the rotation starts again with the
first OUTFIL data set.
SPLITBY can be used to rotate by a specified number of records rather than by
one record, for example, records 1-10 to the first OUTFIL data set, records 11-20
to the second OUTFIL data set, and so on. When each OUTFIL data set has n
records, the rotation starts again with the first OUTFIL data set.
SPLIT1R can be used to rotate once by a specified number of records. SPLIT1R
continues with the last OUTFIL data set instead of rotating back to the first
OUTFIL data set. When each OUTFIL data set has n records, the last OUTFIL
data set will have the remainder of the records even if that is more than n.
v
v LINES, HEADER1, TRAILER1, HEADER2, TRAILER2, SECTIONS, and
NODETAIL indicate that a report is to be produced for an OUTFIL group, and
specify the details of the report records to be produced for the report. Reports
can contain report records for a report header (first page), report trailer (last
page), page header and page trailer (at the top and bottom of each page,
respectively), and section headers and trailers (before and after each section,
respectively).
228

Data records for the report result from the inclusion of OUTFIL input records.
All of the capabilities of the OUTREC, BUILD, OVERLAY, FINDREP, or IFTHEN
parameters are available to create reformatted data records from the OUTFIL
input records. Each set of sequential OUTFIL input records, with the same
binary value for a specified field, results in a corresponding set of data records
that is treated as a section in the report.
The length for the data records must be equal to or greater than the maximum
report record length. OUTFIL data sets used for reports must have or will be
given ANSI control character format ('A' as in, for example, RECFM=FBA or
RECFM=VBA), and must allow an extra byte in the LRECL for the carriage
control character that DFSORT will add to each report and data record. DFSORT
uses these carriage control characters to control page ejects and the placement of
the lines in your report according to your specifications. DFSORT uses
appropriate carriage controls (for example, C'-' for triple space) in header and
trailer records when possible, to reduce the number of report records written.
DFSORT always uses the single space carriage control (C' ') in data records.
Although these carriage control characters may not be shown when you view an
OUTFIL data set (depending upon how you view it), they will be used if you
print the report.
v BLKCCH1, BLKCCT1 or BLKCCH2 can be used avoid forcing a page eject for
the report header, report trailer or first page header, respectively, by using a
blank instead of a '1' for the ANSI control carriage character of its first line.
v REMOVECC can be used to remove the ANSI control characters from a report.
In this case, an 'A' is not added to or required for the RECFM and an extra byte
is not added to or required for the LRECL.
v IFTRAIL can be used to update count and total values in an existing trailer
(last) record based on the current data records.
v Figure 7 on page 230 illustrates the order in which OUTFIL records and
parameters are processed.
229
Figure 7. OUTFIL Processing Order
Note:
1. DFSORT accepts but does not process the following OUTFIL operands:
BLKSIZE=value, BUFLIM=value, BUFOFF=value, CARDS=value,
CLOSE=value, DISK, ESDS, EXIT, FREEOUT, KSDS, LRECL=value, NOTPMK,
OPEN=value, OUTPUT, PAGES=value, PRINT, PUNCH, REUSE, RRDS, SPAN,
SYSLST, TAPE, and TOL.
2. Sample syntax is shown throughout this section. Complete OUTFIL statement
examples are shown and explained under OUTFIL featuresexamples on
page 377.
FNAMES
FNAMES=
ddname
,
(
ddname
Specifies ddnames associated with the OUTFIL data sets for this OUTFIL
230

statement. The ddnames specified using the FNAMES and FILES parameters
constitute the output data sets for this OUTFIL group to which all of the other
parameters for this OUTFIL statement apply.
If FNAMES specifies the ddname in effect for the SORTOUT data set (that is,
whichever is in effect among ddname from SORTOUT=ddname, ccccOUT from
SORTDD=cccc, or SORTOUT) DFSORT will treat the data set associated with
that ddname as an OUTFIL data set rather than as the SORTOUT data set.
ddname
specifies a 1- through 8-character ddname. A DD statement must be
present for this ddname.
Sample Syntax:
OUTFIL FNAMES=(OUT1,OUT2,PRINTER,TAPE)
OUTFIL FNAMES=BACKUP
Default for FNAMES: If neither FNAMES nor FILES is specified for an OUTFIL
statement, the default ddname is:
v ddname if SORTOUT=ddname is in effect
v ccccOUT if SORTDD=cccc is in effect and SORTOUT=ddname is not in effect
v SORTOUT if neither SORTOUT=ddname nor SORTDD=cccc is in effect.
FILES
FILES=
d
dd
OUT
,
(
d
dd
OUT
Specifies suffixes for ddnames to be associated with the OUTFIL data sets for
this OUTFIL statement. The ddnames specified using the FNAMES and FILES
parameters constitute the output data sets for this OUTFIL group to which all
of the other parameters for this OUTFIL statement apply.
If FILES specifies the ddname in effect for the SORTOUT data set (that is,
whichever is in effect among ddname from SORTOUT=name, ccccOUT from
SORTDD=cccc, or SORTOUT), DFSORT will treat the data set associated with
that ddname as an OUTFIL data set rather than as the SORTOUT data set.
specifies the 1-character suffix to be used to form the ddname SORTOFd or
ccccOFd if SORTDD=cccc is in effect. A DD statement must be present for
this ddname.
dd specifies the 2-character suffix to be used to form the ddname SORTOFdd

or ccccOFdd if SORTDD=cccc is in effect. A DD statement must be present
for this ddname.
OUT
specifies the suffix OUT is to be used to form the ddname SORTOUT or
ccccOUT if SORTDD=cccc is in effect. A DD statement must be present for
this ddname.
Sample Syntax:
OUTFIL FILES=(1,2,PR,TP)
OUTFIL FILES=OUT
231

Default for FILES: If neither FNAMES nor FILES is specified for an OUTFIL
statement, the default ddname is:
v ddname if SORTOUT=ddname is in effect
v ccccOUT if SORTDD=cccc is in effect and SORTOUT=ddname is not in effect
v SORTOUT if neither SORTOUT=ddname nor SORTDD=cccc is in effect.
STARTREC
STARTREC=n
Specifies the OUTFIL input record at which OUTFIL processing is to start for
this OUTFIL group. OUTFIL input records before this starting record are not
included in the data sets for this OUTFIL group.
specifies the relative record number. The value for n starts at 1 (the first
record) and is limited to 28 digits (15 significant digits).
Sample Syntax:
OUTFIL FNAMES=SKIP20,STARTREC=21
Default for STARTREC: 1.

ENDREC
ENDREC=n
Specifies the OUTFIL input record at which OUTFIL processing is to end for
this OUTFIL group. OUTFIL input records after this ending record are not
included in the data sets for this OUTFIL group.
The ENDREC value must be equal to or greater than the STARTREC value if
both are specified on the same OUTFIL statement.
specifies the relative record number. The value for n starts at 1 (the first
Sample Syntax:
OUTFIL
OUTFIL
OUTFIL
OUTFIL
FNAMES=TOP10,ENDREC=10
FNAMES=FRONT,ENDREC=500
FNAMES=MIDDLE,STARTREC=501,ENDREC=2205
FNAMES=BACK,STARTREC=2206
Default for ENDREC: The last OUTFIL input record.
SAMPLE
SAMPLE=
n
(n,m)
Specifies a sample of OUTFIL input records to be processed for this OUTFIL

group. The sample consists of the first m records in every nth interval.
232
specifies the interval size. The value for n starts at 2 (sample every other
specifies the number of records to be processed in each interval. The value

for m starts at 1 (process the first record in each interval) and is limited to
28 digits (15 significant digits). If m is not specified, 1 is used for m. If m is
specified, it must be less than n.
Sample Syntax:
* PROCESS RECORDS 1, 6, 11, ...
OUTFIL FNAMES=OUT1,SAMPLE=5
* PROCESS RECORDS 1, 2, 1001, 1002, 2001, 2002
OUTFIL FNAMES=OUT2,SAMPLE=(1000,2),ENDREC=2500
* PROCESS RECORDS 23, 48, 73
OUTFIL FNAMES=OUT3,STARTREC=23,ENDREC=75,SAMPLE=25
* PROCESS RECORDS 1001, 1002, 1003, 1101, 1102, 1103, ...
OUTFIL FNAMES=OUT4,STARTREC=1001,SAMPLE=(100,3)
Default for SAMPLE: None; must be specified.

INCLUDE
INCLUDE=
ALL
NONE
Selects the records to be included in the data sets for this OUTFIL group.
The INCLUDE parameter operates in the same way as the INCLUDE
statement, except that:
v The INCLUDE statement applies to all input records; the INCLUDE
parameter applies only to the OUTFIL input records for its OUTFIL group.
v FORMAT=f can be specified with the INCLUDE statement, but not with the
INCLUDE parameter. Thus, you can use FORMAT=f and p,m or p,m,f fields
with the INCLUDE statement, but you must only use p,m,f fields with the
INCLUDE parameter. For example:
INCLUDE FORMAT=BI,
COND=(5,4,LT,11,4,OR,21,4,EQ,31,4,OR,
61,20,SS,EQ,CFLY)
OUTFIL INCLUDE=(5,4,BI,LT,11,4,BI,OR,21,4,BI,EQ,31,4,BI,OR,
61,20,SS,EQ,CFLY)
v D2 format can be specified with the INCLUDE statement, but not with the
INCLUDE parameter.
See INCLUDE control statement on page 96 for complete details.
logical expression
specifies one or more relational conditions logically combined based on
fields in the OUTFIL input record. If the logical expression is true for a
given record, the record is included in the data sets for this OUTFIL group.
ALL
specifies that all of the OUTFIL input records are to be included in the data
sets for this OUTFIL group.
NONE
specifies that none of the OUTFIL input records are to be included in the
data sets for this OUTFIL group.
Sample Syntax:
OUTFIL FNAMES=J69,INCLUDE=(5,3,CH,EQ,CJ69)
OUTFIL FNAMES=J82,INCLUDE=(5,3,CH,EQ,CJ82)
233

Default for INCLUDE: ALL.
OMIT
OMIT=
ALL
NONE
Selects the records to be omitted from the data sets for this OUTFIL group.
The OMIT parameter operates in the same way as the OMIT statement, except
that:
v The OMIT statement applies to all input records; the OMIT parameter
applies only to the OUTFIL input records for its OUTFIL group.
v FORMAT=f can be specified with the OMIT statement, but not with the
OMIT parameter. Thus, you can use FORMAT=f and p,m or p,m,f fields
with the OMIT statement, but you must only use p,m,f fields with the OMIT
parameter. For example:
OMIT FORMAT=BI,
COND=(5,4,LT,11,4,OR,21,4,EQ,31,4,OR,
61,20,SS,EQ,CFLY)
OUTFIL OMIT=(5,4,BI,LT,11,4,BI,OR,21,4,BI,EQ,31,4,BI,OR,
61,20,SS,EQ,CFLY)
v The D2 format can be specified with the OMIT statement, but not with the
OMIT parameter.
See OMIT control statement on page 169 and INCLUDE control statement
on page 96 for complete details.
logical expression
specifies one or more relational conditions logically combined based on
fields in the OUTFIL input record. If the logical expression is true for a
given record, the record is omitted from the data sets for this OUTFIL
group.
ALL
specifies that all of the OUTFIL input records are to be omitted from the
NONE
specifies that none of the OUTFIL input records are to be omitted from the
Sample Syntax:
OUTFIL FILES=01,OMIT=NONE
OUTFIL OMIT=(5,1,BI,EQ,B110.....)
OUTFIL FNAMES=(OUT1,OUT2),
OMIT=(7,2,CH,EQ,C32,OR,18,3,CH,EQ,CXYZ)
Default for OMIT: NONE.

SAVE
SAVE
Specifies that OUTFIL input records not included by STARTREC, ENDREC,

ACCEPT, SAMPLE, INCLUDE or OMIT for any other OUTFIL group are to be
included in the data sets for this OUTFIL group. SAVE operates in a global
fashion over all of the other OUTFIL statements for which SAVE is not
specified, enabling you to keep any OUTFIL input records that would not be
kept otherwise. SAVE will include the same records for each group for which it
234

is specified.
Sample Syntax:
OUTFIL INCLUDE=(8,6,CH,EQ,CACCTNG),FNAMES=GP1
OUTFIL INCLUDE=(8,6,CH,EQ,CDVPMNT),FNAMES=GP2
OUTFIL SAVE,FNAMES=NOT1OR2
Default for SAVE: None; must be specified.

ACCEPT
ACCEPT=n
Specifies the number of OUTFIL input records to be accepted for processing for
this OUTFIL group. A record is accepted if it is not deleted by STARTREC,
ENDREC, SAMPLE, INCLUDE or OMIT processing for the group. After n
OUTFIL input records have been accepted, additional OUTFIL input records
are not included in the data sets for this OUTFIL group.
n
specifies the number of records to be accepted. The value for n starts at 1

and is limited to 28 digits (15 significant digits).
ACCEPT=n operates in a similar way to the ENDREC=n option. However,

whereas ENDREC=n stops processing OUTFIL input records for a group at
relative record n, ACCEPT=n stops processing OUTFIL input records for a
group after n records have been "accepted". If both ACCEPT=n and
ENDREC=n are specified, processing will stop when the first criteria is met.
Sample Syntax:
OUTFIL FNAMES=OUT1,INCLUDE=(11,3,CH,EQ,CD51),ACCEPT=3
OUTFIL FNAMES=OUT2,STARTREC=5,SAMPLE=1000,ACCEPT=12
Default for ACCEPT: None; must be specified.

PARSE
235

,
,
PARSE=(
%n=
%nn=
%nnn=
%=
FIXLEN=m
ABSPOS=p
ADDPOS=x
SUBPOS=y
,
STARTAFT=string
STARTAFT=an
STARTAFT=BLANKS
STARTAT=string
STARTAT=an
STARTAT=BLANKS
STARTAT=NONBLANK
,
ENDBEFR=string
ENDBEFR=an
ENDBEFR=BLANKS
ENDAT=string
ENDAT=an
ENDAT=BLANKS
PAIR=APOST
PAIR=QUOTE
REPEAT=v
position/length fields (p,m) can be used in the BUILD (or OUTREC) or
want to extract.
For each %nn parsed field, you define where the data to be extracted starts
and ends, and the length (m) of the fixed field to contain the extracted data.
For example, if your input records contained CSV data as follows:
AA,BBBB,C,DDDDD
EEEEE,,F,GG
HHH,IIIII,JJ,K
236

and you wanted to reformat the data into fixed positions like this:
AA
EEEEE
HHH
BBBB
IIIII
C
F
JJ
DDDDD
GG
K
you could use this OUTFIL statement:

OUTFIL PARSE=(%01=(ENDBEFR=C,,FIXLEN=5),
%04=(FIXLEN=5)),
BUILD=(1:%01,11:%02,21:%03,31:%04)
The PARSE operand:

v assigns %01 to the first parsed field, and defines it as starting at position 1
and ending before the next (first) comma with a fixed length of 5 bytes
v assigns %02 to the second parsed field, and defines it as starting after the
(first) comma and ending before the next (second) comma with a fixed
length of 5 bytes
v assigns %03 to the third parsed field, and defines it as starting after the
(second) comma and ending before the next (third) comma with a fixed
length of 5 bytes
v assigns %04 to the fourth parsed field, and defines it as starting after the
(third) comma and ending after 5 bytes with a fixed length of 5 bytes
You can start extracting data:
v at a specific position (ABSPOS=p), a relative position (ADDPOS=x or
SUBPOS=y)
v after the start of one or more specific strings (STARTAFT=string),
alphanumeric character sets (STARTAFT=an) or blanks
(STARTAFT=BLANKS)
v at the start of one or more specific strings (STARTAT=string), alphanumeric
character sets (STARTAT=an) or blanks (STARTAT=BLANKS) or a nonblank
(STARTAT=NONBLANK).
You can end extracting data:
v before the start of one or more specific strings (ENDBEFR=string),
alphanumeric character sets (ENDBEFR=an) or blanks
(ENDBEFR=BLANKS),
v at the end of one or more specific strings (ENDAT=string), alphanumeric
character sets (ENDAT=an) or blanks (ENDAT=BLANKS)
v after a specified number of bytes (FIXLEN=m without ENDBEFR or
ENDAT).
For STARTAFT=an, STARTAT=an, ENDBEFR=an and ENDAT=an, you can
specify one of the following for an to indicate the alphanumeric character set
you want to use:
v
v
v
v
v
LC: Lowercase characters (a-z)

MC: Mixed case characters (A-Z, a-z)
v NUM: Numeric characters (0-9)
237

Multiple STARTAFT/STARTAT strings, UC, LC, MC, UN, LN, MN, NUM,
BLANKS or NONBLANK can be used for a %nn parsed field to search for
more than one string, set of characters, blanks or a nonblank. The first search
satisfied is used.
Multiple ENDBEFR/ENDAT strings, UC, LC, MC, UN, LN, MN, NUM, or
BLANKS can be used for a %nn parsed field to search for more than one
string, set of characters or blanks. The first search satisfied is used.
For each record, each %nn and % parsed field you define is processed
according to the suboperands you define, in the following order:
v ABSPOS or ADDPOS or SUBPOS
v STARTAFT, STARTAT and PAIR
v ENDBEFR, ENDAT and PAIR
v FIXLEN
As each %nn parsed field is processed, data is extracted according to the
suboperands you specify.
In addition, as each %nn or % parsed field is processed, the Start Pointer for
the next parsed field advances through the record according to the
suboperands you specify. By default, for the first parsed field, the Start Pointer
is set to position 1 for fixed-length records or to position 5 for variable-length
records, and the Start Pointer for each subsequent %nn parsed field uses the
Start Pointer as set by the previous %nn parsed field. For any parsed field, the
Start Pointer advances as follows depending on the suboperands specified, and
may advance more than once (for example, if STARTAFT and ENDBEFR are
both specified, or if STARTAT and FIXLEN are both specified):
v if ABSPOS=p is specified, the Start Pointer is set to position p.
v if ADDPOS=x is specified, the Start Pointer is incremented by x.
v if SUBPOS=y is specified, the Start Pointer is decremented by y.
v if STARTAFT=string or STARTAT=string is specified, the search for string
starts at the Start Pointer. If string is found, the Start Pointer is set to the
byte after the end of the string. If string is not found, the current parsed
field and any subsequent parsed fields contain all blanks.
v if STARTAFT=an or STARTAT=an is specified, the search for the character in
the specified alphanumeric set starts at the Start Pointer. If a character in the
set is found, the Start Pointer is set to the byte after the character. If a
character in the set is not found, the current parsed field and any
subsequent parsed fields contain all blanks.
v if STARTAFT=BLANKS or STARTAT=BLANKS is specified, the search for
blanks starts at the Start Pointer. If a blank is found, the Start Pointer is set
to the next nonblank. If a blank is not found, the current parsed field and
any subsequent parsed fields contain all blanks.
v if STARTAT=NONBLANK is specified, the search for a nonblank starts at the
Start Pointer. If a nonblank is found, the Start Pointer is set to the nonblank.
If a nonblank is not found, the current parsed field and any subsequent
parsed fields contain all blanks.
v if ENDBEFR=string or ENDAT=string is specified, the search for string starts
at the Start Pointer. If string is found, the Start Pointer is set to the byte after
the end of the string. If string is not found, data for this parsed field is
extracted up to the end of the record. Subsequent parsed fields contain all
blanks.
v if ENDBEFR=an or ENDAT=an is specified, the search for the character in
the specified alphanumeric set starts at the Start Pointer. If a character in the
238

set is found, the Start Pointer is set to the byte after the character. If a
character in the set is not found, data for this parsed field is extracted up to
the end of the record. Subsequent parsed fields contain all blanks.
v if ENDBEFR=BLANKS or ENDAT=BLANKS is specified, the search for
blanks starts at the Start Pointer. If a blank is found, the Start Pointer is set
to the next nonblank. If a blank is not found, data for this parsed field is
extracted up to the end of the record. Subsequent parsed fields contain all
blanks.
v if ENDBEFR and ENDAT are not specified and FIXLEN=m is specified, the
Start Pointer is set m bytes past the start of the extracted data.
v if the Start Pointer advances past the end of the record, PARSE processing
stops.
For example, if you had these records as input:
11*+23 NEW MAXCC=00
(Monica) 18
1*-6 OLD MAXCC=04 (Joey)
-2735
232*+86342 SHR MAXCC=04 (Rachel) 0
and you wanted this output:

Monica
Joey
Rachel
MAXCC=00 +000023 011

18
MAXCC=04 -000006 001 -2735
MAXCC=04 +086342 232
0

OUTFIL FNAMES=OUT1,
PARSE=(%00=(ENDBEFR=C*,FIXLEN=3),
BUILD=(%03,11:%02,21:%01,SFF,M26,LENGTH=7,
30:%00,UFF,M11,LENGTH=3,35:%04,JFY=(SHIFT=RIGHT))
For this example, the Start Pointer advances as follows for the first record:
v For %00: Set Start Pointer to position 1. Start searching for '*' end string at
position 1. Extract '11' (per ENDBEFR=C'*'). Set Start Pointer after '*' (at '+').
v For %01: Start searching for end blank at the Start Pointer (at '+'). Extract
'+23' (per ENDBEFR=BLANKS). Set Start Pointer after the blanks (at 'N').
v For %02: Start searching for 'MAX' start string at the Start Pointer (at 'N').
Extract 'MAXCC=00' (per FIXLEN=8). Increment Start Pointer by m (8) bytes
from the start of the extracted string (at blank after '00').
v For %03: Start searching for '(' start string at Start Pointer (at blank after '00').
Set Start Pointer after '(' (at 'M'). Start searching for ')' end string at Start
Pointer (at 'M'). Extract 'Monica' (per ENDBEFR=C')'). Set Start Pointer after
')' (at blank after ')').
v For %04: Start searching for start blank at the Start Pointer (at blank after ')').
Extract '18 ' (per FIXLEN=5).
If the Start Pointer advances past the end of the record, PARSE processing
stops. For example, if you had these input records:
First=George Last=Washington
First=John Middle=Quincy Last=Adams
and you used this OUTFIL statement:
239

OUTFIL FNAMES=OUT1,
PARSE=(%00=(STARTAFT=CFirst=,ENDBEFR=C ,FIXLEN=12),
%01=(STARTAFT=CMiddle=,ENDBEFR=C ,FIXLEN=12),
%02=(STARTAFT=CLast=,ENDBEFR=C ,FIXLEN=12)),
BUILD=(%00,%01,%02)
the %01 parsed field (middle name) is not found in the first record, so PARSE
processing stops and the %02 parsed field is set to blanks. You can handle this
kind of possible missing field situation by using IFTHEN PARSE instead of
PARSE. For example:
OUTFIL FNAMES=OUT1,
IFTHEN=(WHEN=INIT,
PARSE=(%00=(STARTAFT=CFirst=,ENDBEFR=C ,FIXLEN=12))),
IFTHEN=(WHEN=INIT,
PARSE=(%01=(STARTAFT=CMiddle=,ENDBEFR=C ,FIXLEN=12))),
IFTHEN=(WHEN=INIT,
PARSE=(%02=(STARTAFT=CLast=,ENDBEFR=C ,FIXLEN=12)),
BUILD=(%00,%01,%02))
By default, the Start Pointer is set to 1 (F) or 5 (V) for each IFTHEN PARSE, so
the missing middle name in the first record does not prevent the last name
from being extracted.
See the discussion of "IFTHEN" for more information on IFTHEN PARSE.
You must define each %nn field with PARSE before you use it in BUILD or
OVERLAY, so generally you would want to specify PARSE before BUILD or
OVERLAY. If you define a %nn field with PARSE, but do not use it in BUILD
or OVERLAY, data will not be extracted for that parsed field.
Each %nn parsed field (%0-%999) can only be defined once per run with
PARSE, but can be used as many times as needed in BUILD or OVERLAY.
The %nn parsed fields defined for a particular source (INREC, OUTREC or
OUTFIL) can only be used in the BUILD or OVERLAY operands for that
source. Generally, you will only use PARSE for one source (for example,
INREC) per run. But you could define a separate set of %nn parsed fields for
each source, if needed. For example, you could define %21 and %22 for the
INREC statement, %101, %102 and %103 for the OUTREC statement, %40 and
%45 for OUTFIL statement 1, and %30, %31 and %32 for OUTFIL statement 2.
You could then use %21 and %22 for INREC BUILD or OVERLAY, %102 and
%103 for OUTREC BUILD or OVERLAY, %40 and %45 for OUTFIL statement 1
BUILD or OVERLAY, and %30, %31 and %32 for OUTFIL statement 2 BUILD
or OVERLAY. Note that you could not, for example, use %21 and %22 for
OUTREC BUILD or OVERLAY.
If you specify PARSE=(...),IFTHEN=(...) or IFTHEN=(...),PARSE=(...), DFSORT
will terminate. If you need to specify PARSE with IFTHEN, specify it within
one or more individual IFTHEN clauses, as appropriate. For details, see the
description of "IFTHEN".
%n, %nn or %nnn
Assigns %n, %nn or %nnn to a parsed field. The variable-length data
defined for the parsed field is extracted to a fixed length area (padded
with blanks or truncated as appropriate) and can be used where a p,m
field can be used in BUILD or OVERLAY.
n can be 0 to 9, nn can be 00 to 99, and nnn can be 000 to 999 but each %n,
%nn or %nnn value can only be defined once per run. This gives you a
maximum of 1000 parsed fields per run.
240

%0 to %9 will be treated as equivalent to %00 to %09. %010 to %099 will
be treated as equivalent to %10 to %99. You can define parse field n with
%n, %0n or %00n; you can use %n, %0n and %00n interchangeably. You
can define parse field nn with %nn or %0nn; you can use %nn and %0nn
interchangeably.
Sample Syntax
OUTFIL PARSE=(%99=(ABSPOS=6,STARTAT=CKW=(,ENDAT=C),FIXLEN=20),
%10=(ENDBEFR=UC,ENDBEFR=C$,FIXLEN=5),
OVERLAY=(1:1,5,%3,C=,%99,1:1,36,SQZ=(SHIFT=LEFT),%010)
Specifies a parsed field to be ignored. No data is extracted, but the starting

point for the next parsed field advances according to the suboperands
specified. Use % when you don't need the data from a particular field, but
you do need to get to the next field. For example, if we had the four CSV
fields shown earlier as input, but we only wanted to extract the first and
fourth fields, we could use this OUTFIL statement:
%=(ENDBEFR=C,),
%=(ENDBEFR=C,),
%04=(FIXLEN=5)),
BUILD=(1:%01,11:%04,21:%01,HEX)
Data is extracted for %01 (first field) and %04 (fourth field), but not for %
(second and third fields).
FIXLEN=m
Specifies the length (m) of the fixed area to contain the extracted
variable-length data for this %nn fixed parsed field. m can be 1 to 32752.
You must specify FIXLEN=m for each %nn parsed field. FIXLEN=m is
optional for a % ignored field.
When %nn is specified in BUILD or OVERLAY, the variable-length data is
extracted to the fixed area and then used for BUILD or OVERLAY
processing. The fixed area always contains m bytes. If the extracted data is
longer than m, the fixed area contains the first m bytes of the extracted
data (the data is truncated on the right to m bytes). If the extracted data is
shorter than m, the fixed area contains the extracted data padded with
blanks on the right to m bytes. If no data is extracted, the fixed area
contains all blanks.
The length m from FIXLEN=m for fixed parsed fields (%nn) corresponds to
the length (m) for fixed fields (p,m). Keep in mind that %nn fields are
left-aligned and may be padded on the right with blanks. You must ensure
that the data in each %nn field and its length (m) are valid for the way
you want to use that %nn field in BUILD or OVERLAY. For example, if
you had these records as input:
123,8621
1,302
18345,17
and you wanted this output:

1.23
0.01
183.45
86.21
3.02
0.17
241

%01=(FIXLEN=5)),
BUILD=(%00,UFF,EDIT=(IIT.TT),2X,%01,UFF,EDIT=(IIT.TT))
Note that the 5-byte extracted parsed fields for %00 are left-aligned and
padded with blanks (b) like this:
123bb
1bbbb
18345
and the 5-byte extracted parsed fields for %01 are left-aligned and padded
with blanks (b) like this:
8621b
302bb
17bbb
Thus, you cannot use numeric formats like ZD or FS for these parsed
fields, but you can use UFF. Note also that you must ensure that m from
FIXLEN=m for each %nn parsed field is valid for the specific BUILD or
OVERLAY item in which you use %nn. For example, you could use 1-44 as
m for UFF, but you couldn't use 45 as m.
If ENDBEFR and ENDAT are not specified, the Start Pointer is set m bytes
past the start of the extracted data. For example, if you had this record as
input:
MAX=(ABCDEF)
and you specified:

OUTFIL PARSE=(%100=(STARTAT=CMAX,FIXLEN=8),
%101=(FIXLEN=3),%102=(FIXLEN=1)),
BUILD=(%100,12:%101,20:%102)
The output record would be:

MAX=(ABC
DEF
Sample Syntax
OUTFIL PARSE=(%00=(STARTAFT=CKW=,FIXLEN=12)),
BUILD=(%00)
ABSPOS=p
Sets the Start Pointer for this parsed field to p. p can be 1 to 32752. By
default, the Start Pointer for the first %nn parsed field is position 1 for
fixed-length records or position 5 for variable-length records, and the Start
Pointer for each subsequent %nn parsed field is the Start Pointer set by the
previous %nn field. You can use ABSPOS=p to set the Start Pointer to
position p to override the default Start Pointer. If the resulting Start Pointer
is less than position 5 for variable length records, it will be set to position
5.
Example
If your input is:
****|BB|CCCC|
****|EEEE|FF|
and you wanted to reformat the data into fixed positions like this:
**** BB
CCCC
**** EEEE FF
242

OUTFIL PARSE=(%01=(ABSPOS=6,ENDBEFR=C|,FIXLEN=4),
%02=(ENDBEFR=C|,FIXLEN=4)),
BUILD=(1,4,X,%01,X,%02)
The initial Start Pointer for the %01 parsed field is set to position 6 instead
of to position 1.
ADDPOS=x
Increments the Start Pointer for this parsed field by x. x can be 1 to 32752.
By default, the Start Pointer for the first %nn parsed field is position 1 for
previous %nn field. You can use ADDPOS=x to increment the Start Pointer
by x to override the default Start Pointer.
Sample Syntax
OUTFIL PARSE=(%00=(ENDAT=C||,FIXLEN=10),
%01=(ADDPOS=5,STARTAFT=C;,FIXLEN=6)),
BUILD=(%00,TRAN=ALTSEQ,%01)
SUBPOS=y
Decrements the Start Pointer for this parsed field by y. y can be 1 to 32752.
By default, the Start Pointer for the first %nn parsed field is position 1 for
previous %nn field. You can use SUBPOS=y to decrement the Start Pointer
by x to override the default Start Pointer. If the resulting Start Pointer is
less than position 1 for fixed-length records, it will be set to position 1. If
the resulting Start Pointer is less than position 5 for variable-length
records, it will be set to position 5.
Sample Syntax
OUTFIL PARSE=(%00=(ENDBEFR=C|;|,FIXLEN=10),
%01=(SUBPOS=1,STARTAT=C|,ENDAT=C|,FIXLEN=12)),
BUILD=(%00,TRAN=ALTSEQ,%01)
STARTAFT=string
Data is to be extracted for this parsed field starting after the last byte of
the specified string. The search for the string begins at the Start Pointer. If
the specified string is not found, data is not extracted for this parsed field
or any subsequent parsed fields. If the specified string is found, data is
extracted to the fixed area for this parsed field starting at the byte after the
end of the string, and the Start Pointer is set to the byte after the end of
the string.
string can be 1 to 256 characters specified using a character string constant
(C'xx...x') or a hexadecimal string constant (X'yy...yy'). See INCLUDE
control statement on page 96 for details of coding character and
hexadecimal string constants.
Example
If your input is:
MAX=25,832
MAX=1,275
MIN=17 %3
MIN=2,831
$4
and you wanted your output to contain MAX-MIN, you could use this
OUTFIL statement:
OUTFIL PARSE=(%00=(STARTAFT=CMAX=,ENDBEFR=X40,FIXLEN=6),
%01=(STARTAFT=CMIN=,ENDBEFR=X40,FIXLEN=6)),
OVERLAY=(50:CDELTA: ,%00,UFF,SUB,%01,UFF,TO=ZD,LENGTH=8)
243

STARTAFT=an
Data is to be extracted for this parsed field starting after the first character
found from the specified alphanumeric set. The search for a character in
the specified set begins at the Start Pointer. If a character in the set is not
found, data is not extracted for this parsed field or any subsequent parsed
field. If a character in the set is found, data is extracted to the fixed area
for this parsed field starting at the byte after that character, and the Start
Pointer is set to the byte after that character.
an can be one of the following alphanumeric character sets:
v LC: Lowercase characters (a-z)
v MC: Mixed case characters (A-Z, a-z)
v UN: Uppercase and numeric characters (A-Z, 0-9)
v LN: Lowercase and numeric characters (a-z, 0-9)
v MN: Mixed case and numeric characters (A-Z, a-z, 0-9)
v NUM: Numeric characters (0-9)
Note that STARTAFT=an is equivalent to specifying a STARTAFT=c
operand for each character in the set. For example:
%=(STARTAFT=NUM)
is equivalent to specifying
%=(STARTAFT=C0,STARTAFT=C1,STARTAFT=C2,STARTAFT=C3,
STARTAFT=C4,STARTAFT=C5,STARTAFT=C6,STARTAFT=C7,
STARTAFT=C8,STARTAFT=C9)
You can add more characters to a set with additional STARTAFT operands.
For example, if you wanted the lowercase characters and the $ and @
characters for the set, you could use:
%05=(STARTAFT=LC,STARTAFT=C$,STARTAFT=C@,FIXLEN=10)
Example
If your input is:
a123
1Stop
Z0056
2rest
q8
3go
and you wanted your output to be:

000123 STOP
000056 REST
000008 GO

OUTFIL PARSE=(%00=(STARTAFT=MC,ENDBEFR=C ,FIXLEN=5),
%01=(STARTAFT=NUM,FIXLEN=8)),
BUILD=(%00,UFF,EDIT=(TTTTTT),X,%01,TRAN=LTOU)
STARTAFT=BLANKS
Data is to be extracted for this parsed field starting at the first nonblank
after one or more blanks. The search for a blank begins at the Start Pointer.
If a blank is not found, data is not extracted for this parsed field or any
subsequent parsed fields. If a blank is found, data is extracted to the fixed
area for this parsed field starting at the first nonblank, and the Start
Pointer is set to the first nonblank.
Example
244

If your input is:
Frank
D28
Vicky
D52

Frank
Vicky
D28
D52

OUTFIL PARSE=(%00=(STARTAFT=BLANKS,FIXLEN=8),
BUILD=(%00,14:%01)
STARTAT=string
Data is to be extracted for this parsed field starting at the first byte of the
specified string. The search for the string begins at the Start Pointer. If the
specified string is not found, data is not extracted for this parsed field or
any subsequent parsed fields. If the specified string is found, data is
extracted to the fixed area for this parsed field starting at the first byte of
the string, and the Start Pointer is set to the byte after the end of the
string.
Example
If your input is:
MAX=58321
MIN=00273
MAX=01275
MIN=00017

MAX=58321 MIN=00273
MAX=01275 MIN=00017

OUTFIL PARSE=(%00=(STARTAT=CMAX=,FIXLEN=9),
%01=(STARTAT=CMIN=,FIXLEN=9)),
BUILD=(%00,X,%01)
STARTAT=an
Data is to be extracted for this parsed field starting at the first character
field. If a character in the set is found, data is extracted to the fixed area
for this parsed field starting at that character, and the Start Pointer is set to
the byte after that character.
See STARTAFT=an previously in this section for more information on using
the available alphanumeric character sets.
Example
If your input is:
$@$321%@%$Frank
@053$Susan
%%%$$836%$Vicky
245

and you wanted your output to contain the amount and name fields:
321 Frank
053 Susan
836 Vicky

OUTFIL PARSE=(%00=(STARTAT=NUM,FIXLEN=3),
%01=(STARTAT=UC,FIXLEN=10)),
BUILD=(%00,X,%01)
STARTAT=BLANKS
Data is to be extracted for this parsed field starting at the first blank. The
search for a blank begins at the Start Pointer. If a blank is not found, data
is not extracted for this parsed field or any subsequent parsed fields. If a
blank is found, data is extracted to the fixed area for this parsed field
starting at the first blank, and the Start Pointer is set to the first nonblank.
STARTAT=NONBLANK
Data is to be extracted for this parsed field starting at the first nonblank.
The search for a nonblank begins at the Start Pointer. If a nonblank is not
fields. If a nonblank is found, data is extracted to the fixed area for this
parsed field starting at the nonblank, and the Start Pointer is set to the
nonblank.
Example
If your input is:
Frank
D28
Victoria
Holly D15
Roberta
D52
D52

Frank
Victoria
Holly
Roberta
D28
D52
D15
D52

OUTFIL PARSE=(%00=(STARTAT=NONBLANK,ENDBEFR=BLANKS,FIXLEN=8),
%01=(FIXLEN=3)),
BUILD=(%00,14:%01)
Note: Multiple STARTAFT/STARTAT strings, UC, LC, MC, UN, LN, MN,
NUM, BLANKS or NONBLANK can be used for a %nn parsed field to
search for more than one string, set of characters, blanks or a nonblank.
The first search satisfied is used.
Example
If your input is:
12
MAXCC=08
58 MINCC=00
MAXCC=04

MAXCC=08
MINCC=00
MAXCC=04
246

OUTFIL PARSE=(%00=(STARTAT=CMAXCC=,STARTAT=CMINCC=,
FIXLEN=8)),BUILD=(%00)
ENDBEFR=string
Data is to be extracted for this parsed field ending before the first byte of
the specified string. The search for the string begins at the Start Pointer. If
the specified string is not found, data is extracted to the fixed area for this
parsed field up to the end of the record, but data is not extracted for any
subsequent parsed fields. If the specified string is found, data is extracted
to the fixed area for this parsed field up to the byte before the start of the
string, and the Start Pointer is set to the byte after the end of the string.
Example
If your input is:
Morgan Hill;California;5000:
San Jose;California;2000:
Austin;Texas;8000:

Morgan Hill
San Jose
Austin
California 5000
California 2000
Texas
8000

OUTFIL PARSE=(%00=(ENDBEFR=C;,FIXLEN=14),
%01=(ENDBEFR=C;,FIXLEN=12),
%02=(ENDBEFR=C:,FIXLEN=4)),
BUILD=(%00,%01,%02)
ENDBEFR=X'00' can be used to extract null-terminated strings to fixed

parsed fields.
Example
If your input contains null-terminated strings that look like this in hex:
F0F5F800
F1F2F3F4F500
F9F7F2F000
F500
F0F0F3F700
and you wanted to convert the null-terminated strings to right-aligned

numeric values like this:
58
12345
9720
5
37

OUTFIL PARSE=(%01=(ENDBEFR=X00,FIXLEN=5)),
BUILD=(%01,UFF,EDIT=(IIIIT))
ENDBEFR=an
Data is to be extracted for this parsed field ending before the first character
247

found, data is extracted to the fixed area for this parsed field up to the end
of the record, but data is not extracted for any subsequent parsed fields. If
a character in the set is found, data is extracted to the fixed area for this
parsed field up to the byte before that character, and the Start Pointer is set
to the byte after that character.
Example
If your input is:
$$$$Samantha
$052
$$$$$cat
$$358721
and you wanted to extract the leading $ characters in each input record,
OUTFIL PARSE=(%10=(ENDBEFR=MN,FIXLEN=10)),
BUILD=(%10)
ENDBEFR=BLANKS
Data is to be extracted for this parsed field ending before the first blank.
The search for a blank begins at the Start Pointer. If a blank is not found,
data is extracted to the fixed area for this parsed field up to the end of the
record, but data is not extracted for any subsequent parsed fields. If a
blank is found, data is extracted to the fixed area for this parsed field up to
the byte before the blank, and the Start Pointer is set to the first nonblank.
Example
If your input is:
1
Frank
2 Loretta
D28 123
No
D52
58 Yes

0123 Frank
D28
0058 Loretta D52

OUTFIL PARSE=(%00=(STARTAFT=BLANKS,ENDBEFR=BLANKS,FIXLEN=8),
%02=(ENDBEFR=C ,FIXLEN=4)),
BUILD=(%02,UFF,EDIT=(TTTT),X,%00,%01)
ENDAT=string
Data is to be extracted for this parsed field ending at the last byte of the
specified string. The search for the string begins at the Start Pointer. If the
specified string is not found, data is extracted to the fixed area for this
parsed field up to the end of the record, but data is not extracted for any
subsequent parsed fields. If the specified string is found, data is extracted
to the fixed area for this parsed field up to the last byte of the string, and
the Start Pointer is set to the byte after the end of the string
248

Example
If your input is:
12,(June)
5852,(Gracie)

(June)
(Gracie)

OUTFIL PARSE=(%00=(STARTAT=C(,ENDAT=C),FIXLEN=12)),
BUILD=(%00)
ENDAT=an
Data is to be extracted for this parsed field ending at the first character
found, data is extracted to the fixed area for this parsed field up to the end
of the record, but data is not extracted for any subsequent parsed fields. If
a character in the set is found, data is extracted to the fixed area for this
parsed field up to that character, and the Start Pointer is set to the byte
after that character.
Example
If your input is:
A0005X2
Z081789Q331
R3M05
S52J06834

A0005X
Z081789Q
R3M
S52J
2
331
05
06834

OUTFIL PARSE=(%01=(STARTAT=UC,ENDAT=UC,FIXLEN=10),
%02=(FIXLEN=5)),
BUILD=(%01,X,%02)
ENDAT=BLANKS
Data is to be extracted for this parsed field ending at the last blank before
a nonblank. The search for a blank begins at the Start Pointer. If a blank is
not found, data is extracted to the fixed area for this parsed field up to the
end of the record, but data is not extracted for any subsequent parsed
fields. If a blank is found, data is extracted to the fixed area for this parsed
field up to the byte before the first nonblank, and the Start Pointer is set to
the first nonblank.
Note: Multiple ENDBEFR/ENDAT strings, UC, LC, MC, UN, LN, MN,
NUM, or BLANKS can be used for a %nn parsed field to search for more
than one string, set of characters or blanks. The first search satisfied is
used.
249

Example
If your input is:
12;
58:
04/

12
58
04

OUTFIL PARSE=(%01=(ENDBEFR=C;,ENDBEFR=C:,ENDBEFR=C/,
FIXLEN=10)),
BUILD=(%01,UFF,TO=ZD,LENGTH=2,80:X)
PAIR=APOST
Do not search for strings or blanks between apostrophe (') pairs. Use
POST=APOST when you might have strings you are searching for within
literals that should not satisfy the search.
Once an apostrophe is found, searching is discontinued until another
apostrophe is found. Searching then continues after the second apostrophe
of the pair. If an unpaired apostrophe is found, strings or BLANKS will not
be recognized from that point on, so be careful not to set the Start Pointer
in the middle of an apostrophe pair.
Do not specify PAIR=APOST if you want to search for a string containing
an apostrophe, because the string will not be searched for within the
paired apostrophes.
Example
If your input is:
23,12,567,823,5,032
9,903,18,321,8

23 12,567,823
9,903
18,321
5,032
8

OUTFIL PARSE=(%00=(ENDBEFR=C,,FIXLEN=12,PAIR=APOST),
%01=(ENDBEFR=C,,FIXLEN=12,PAIR=APOST),
%02=(FIXLEN=12)),
BUILD=(%00,UFF,EDIT=(II,III,IIT),X,
%01,UFF,EDIT=(II,III,IIT),X,
%02,UFF,EDIT=(II,III,IIT))
With PAIR=APOST, the commas outside the apostrophe pairs satisfy the
search, but the commas within the apostrophe pairs do not satisfy the
search.
PAIR=QUOTE
Do not search for strings or blanks between quote (") pairs. Use
POST=QUOTE when you might have strings you are searching for within
literals that should not satisfy the search.
Once a quote is found, searching is discontinued until another quote is
found. Searching then continues after the second quote of the pair. If an
250

unpaired quote is found, strings or BLANKS will not be recognized from
that point on, so be careful not to set the Start Pointer in the middle of a
quote pair.
Do not specify PAIR=QUOTE if you want to search for a string containing
a quote, because the string will not be searched for within the paired
quotes.
Example
If your input is:
"23","12,567,823","5,032"
"9,903","18,321","8"

23 12,567,823
9,903
18,321
5,032
8

OUTFIL PARSE=(%00=(ENDBEFR=C,,FIXLEN=12,PAIR=QUOTE),
%01=(ENDBEFR=C,,FIXLEN=12,PAIR=QUOTE),
%02=(FIXLEN=12)),
BUILD=(%00,UFF,EDIT=(II,III,IIT),X,
%01,UFF,EDIT=(II,III,IIT),X,
%02,UFF,EDIT=(II,III,IIT))
With PAIR=QUOTE, the commas outside the quote pairs satisfy the search,
but the commas within the quote pairs do not satisfy the search.
Default for PARSE: None; must be specified.
REPEAT=v
Repeat this parsed field v times.
REPEAT=v can be used with % to specify v identically defined consecutive
parsed fields to be ignored. v can be 2 to 1000. For example, to ignore five
consecutive comma delimited fields, you can use:
%=(ENDBEFR=C,,REPEAT=5),
which is equivalent to using:

%=(ENDBEFR=C,),
%=(ENDBEFR=C,),
%=(ENDBEFR=C,),
%=(ENDBEFR=C,),
%=(ENDBEFR=C,),
REPEAT=v can be used with %nn to specify v identically defined

consecutive parsed fields for which data is to be extracted. v can be 2 to
1000. The parsed fields will start with the %nn field you select and be
incremented by 1 for each repeated parsed field.
For example, to extract four consecutive 10-byte comma delimited fields as
%11, %12, %13 and %14, you can use:
%11=(ENDBEFR=C,,FIXLEN=10,REPEAT=4),
which is equivalent to using:

251

You must ensure that the %nn fields resulting from the use of REPEAT=v
are unique for the run. For example,
%21=(ENDBEFR=C',',FIXLEN=8,REPEAT=10) defines %21-%30, so if you
used any of %21-%30 separately, DFSORT would issue an error message
and terminate; the next available %nn field would be %31.
You must also ensure that any %nnn field resulting from the use of
REPEAT=v does not exceed %999. For example,
%998=(STARTAFT=C'$',FIXLEN=4,REPEAT=3) defines %998-%1000; since
%1000 is invalid, DFSORT would issue an error message and terminate.
Example
If your input is the following period delimited values:
11.2.30.4.55.160.7.82.95.37
10.22.3.41.5.16.72.2.5.42
3.18.33.12.7.9.52.12
and you wanted to extract the first value, skip the next three values and
then extract the last six values to produce output like this:
011 055 160 007 082 095 037
010 005 016 072 002 005 042
003 007 009 052 012 000 000

OUTFIL PARSE=(%00=(ENDBEFR=C.,FIXLEN=3),
%=(ENDBEFR=C.,REPEAT=3),
%01=(ENDBEFR=C.,FIXLEN=3,REPEAT=6)),
BUILD=(%00,UFF,EDIT=(TTT),X,
%01,UFF,EDIT=(TTT),X,
%06,UFF,EDIT=(TTT))
Default for PARSE: None; must be specified.

OUTREC, BUILD, OVERLAY, FINDREP, or IFTHEN
,
OUTREC=
BUILD=
( item
,
OVERLAY=( ( item
,
FINDREP=( ( item
,
IFTHEN=(clause)
These operands allow you to reformat the OUTFIL input records in this
OUTFIL group.
You can create the reformatted OUTFIL records in one of the following ways
using unedited, edited, or converted input fields (p,m for fixed fields, or %nn
for parsed fields - see PARSE) and a variety of constants:
252

v BUILD or OUTREC: Reformat each record by specifying all of its items one
by one. Build gives you complete control over the items you want in your
reformatted OUTFIL records and the order in which they appear. You can
delete, rearrange and insert fields and constants. Examples:
OUTFIL FNAMES=OUT1,BUILD=(1,20,CABC,26:5C*,
15,3,PD,EDIT=(TTT.TT),21,30,80:X)
OUTFIL FNAMES=OUT2,
PARSE=(%00=(ENDBEFR=C,,FIXLEN=6),
%=(ENDBEFR=C,),
%02=(FIXLEN=6)),
BUILD=(C|,%01,SFF,ADD,%02,SFF,M4,LENGTH=12,
C|,%00,C|)
OUTFIL OVERLAY=(45:45,8,TRAN=LTOU)
OUTREC FINDREP=(INOUT=(CPigeon,CDove,CHawk,CEagle))
v IFTHEN clauses: Reformat different records in different ways by specifying

how build, overlay, find/replace, or group operation items are applied to
records that meet given criteria. IFTHEN clauses let you use sophisticated
conditional logic to choose how different record types are reformatted.
Example:
OUTFIL IFTHEN=(WHEN=(1,5,CH,EQ,CTYPE1),
BUILD=(1,40,C**,+1,TO=PD)),
OUTFIL records:
v Fixed position/length fields or variable position/length fields. For fixed
fields, you specify the starting position and length of the field directly. For
DFSORT to extract the relevant data into fixed parsed fields, and then use
the parsed fields as you would use fixed fields.
v Blanks, binary zeros, character strings, and hexadecimal strings.
v Current date, future date, past date, and current time in various forms.
boundaries.
v Hexadecimal or bit representations of binary input fields.
v Left-justified, right-justified, left-squeezed or right-squeezed input fields.
v Numeric input fields of various formats converted to different numeric
formats, or to character format edited to contain signs, thousands separators,
decimal points, leading zeros or no leading zeros, and so on.
253

v Decimal constants converted to different numeric formats, or to character
format edited to contain signs, thousands separators, decimal points, leading
zeros or no leading zeros, and so on.
converted to different numeric formats, or to character format edited to
contain signs, thousands separators, decimal points, leading zeros or no
v SMF, TOD, and ETOD date and time fields converted to different numeric
formats, or to character format edited to contain separators, leading zeros or
Gregorian) converted to corresponding output date fields of another type or
v A character constant, hexadecimal constant or input field selected from a
lookup table, based on a character, hexadecimal or bit constant as input.
v A zoned decimal group identifier, a zoned decimal group sequence number,
or a field propagated from the first record of a group to all of the records of
a group.
OUTREC or BUILD
254

,
OUTREC=
BUILD=

c:
s
p,m
,a
%nn
p
p,m
%nn
p
,TRAN
LTOU
UTOL
ALTSEQ
ATOE
ETOA
HEX
UNHEX
BIT
UNBIT
p,m
,HEX
%nn
p
p,m,f
,edit
%nn,f
,to
(p,m,f)
(%nn,f)
deccon
,edit
(deccon)
,to
arexp
,edit
(arexp)
,to
p,m
,Y2x
%nn
,Y4x
,edit
,to
,todate
,dateop
,Y2x(s)
,Y4x(s)
,Y2xP
p,m
,lookup
%nn
p,m
,justify
%nn
p,m
,squeeze
%nn
seqnum
Specifies all of the items in the reformatted OUTFIL record in the order in
which they are to be included. The reformatted OUTFIL record consists of the
for parsed fields - see PARSE), edited decimal constants, edited results of
For variable-length records, the first item in the BUILD or OUTREC parameter
position (p) as the last item in the BUILD or OUTREC parameter. For example:
OUTFIL OUTREC=(1,4,
C|,
5)
unedited RDW
| separator
255

You can use the BUILD or OUTREC parameter to produce multiple
reformatted output records for each OUTFIL input record, with or without
You can use the BUILD or OUTREC parameter in conjunction with the VTOF
or CONVERT parameter to convert variable-length record data sets to
fixed-length record data sets.
You can use the BUILD or OUTREC parameter with the FTOV parameter to
convert fixed-length record data sets to variable-length record data sets.
You can use the VLFILL parameter to allow processing of variable-length input
records which are too short to contain all specified BUILD or OUTREC fields.
You can use the VLTRIM parameter in conjunction with the BUILD or
OUTREC parameter to remove specified trailing bytes from the end of
variable-length records.
You can use the VLTRAIL parameter in conjunction with the BUILD or
OUTREC parameter to insert a string at the end of variable-length records.
The BUILD or OUTREC parameter can be used with any or all of the report
parameters (LINES, HEADER1, TRAILER1, HEADER2, TRAILER2, SECTIONS,
BLKCCH1, BLKCCH2, BLKCCT1, and NODETAIL) to produce reports. The
report parameters specify the report records to be produced, while the BUILD
or OUTREC parameter specifies the reformatted data records to be produced.
DFSORT uses ANSI carriage control characters to control page ejects and the
placement of the lines in your report, according to your specifications. You can
use the REMOVECC parameter to remove the ANSI carriage control characters.
When you create an OUTFIL report, the length for the longest or only data
record must be equal to or greater than the maximum report record length.
You can use the BUILD or OUTREC parameter to force a length for the data
records that is longer than any report record; you can then either let DFSORT
compute and set the LRECL, or ensure that the computed LRECL is equal to
the existing or specified LRECL. Remember to allow an extra byte in the
LRECL for the ANSI carriage control character.
For example, if your data records are 40 bytes, but your longest report record
is 60 bytes, you can use a BUILD or OUTREC parameter such as:
OUTREC=(1,40,80:X)
DFSORT will then set the LRECL to 81 (1 byte for the ANSI carriage control
character plus 80 bytes for the length of the data records), and pad the data
records with blanks on the right.
If you don't want the ANSI carriage control characters to appear in the output
data set, use the REMOVECC parameter to remove them. For example, if you
specify:
OUTREC=(1,40,80:X),REMOVECC
DFSORT will set the LRECL to 80 instead of 81 and remove the ANSI carriage
control character from each record before it is written.
The BUILD or FIELDS parameter of the OUTREC statement differs from the
BUILD or OUTREC parameter of the OUTFIL statement in the following ways:
256

v The BUILD or FIELDS parameter of the OUTREC statement applies to all
input records; the BUILD or OUTREC parameter of the OUTFIL statement
only applies to the OUTFIL input records for its OUTFIL group.
v The BUILD or OUTREC parameter of the OUTFIL statement supports the
slash (/) separator for creating blank records and new records; the BUILD or
FIELDS parameter of the OUTREC statement does not.
The reformatted OUTFIL output record consists of the separation fields, edited
and unedited input fields, (p,m for fixed fields, or %nn for parsed fields - see
PARSE), edited decimal constants, edited results of arithmetic expressions, and
sequence numbers you select, in the order in which you select them, aligned
on the boundaries or in the columns you indicate.
c: specifies the position (column) for a separation field, input field, decimal
constant, arithmetic expression, or sequence number, relative to the start of the
reformatted OUTFIL output record. Count the RDW (variable-length output
records only) but not the carriage control character (for reports) when
specifying c:. That is, 1: indicates the first byte of the data in fixed-length
output records and 5: indicates the first byte of the data in variable-length
output records.
Unused space preceding the specified column is padded with EBCDIC blanks.
The following rules apply:
v c: must be followed by a separation field, input field, decimal constant, or
arithmetic expression.
reformatted OUTFIL output record.
v For variable-length records, c: must not be specified before the first input
field (the record descriptor word) nor after the variable part of the OUTFIL
input record.
v The colon (:) is treated like the comma (,) or semicolon (;) for continuation to
another line.
See Table 29 on page 130 for examples of valid and invalid column alignment.
s
specifies that a separation field (blanks, zeros, character string, hexadecimal

string, current date, future date, past date, or current time) is to appear in the
reformatted OUTFIL output record, or that a new output record is to be
started, with or without intervening blank output records. These separation
elements (separation fields, new record indicators, and blank record indicators)
can be specified before or after any input field. Consecutive separation
elements may be specified. For variable-length records, separation elements
must not be specified before the first input field (the record descriptor word)
or after the variable part of the OUTFIL input record. Permissible values are
nX, nZ, nC'xx...x', nX'yy...yy', various date and time constants, /.../ and n/.
nX
Blank separation. n bytes of EBCDIC blanks (X'40') are to appear in the

reformatted OUTFIL output records. n can range from 1 to 4095. If n is
omitted, 1 is used.
See Table 30 on page 131 for examples of valid and invalid blank
separation.
nZ
Binary zero separation. n bytes of binary zeros (X'00') are to appear in

the reformatted OUTFIL output records. n can range from 1 to 4095. If
n is omitted, 1 is used.
257

See Table 31 on page 131 for examples of valid and invalid binary zero
separation.
nC'xx...x'
constant (C'xx...x') are to appear in the reformatted OUTFIL output
records. n can range from 1 to 4095. If n is omitted, 1 is used. x can be
any EBCDIC character. You can specify from 1 to 256 characters.
If you want to include a single apostrophe in the character string, you
must specify it as two single apostrophes:
Required:
ONEILL Specify:
CONEILL
See Table 32 on page 131 for examples of valid and invalid character
string separation.
nX'yy...yy'
Hexadecimal string separation. n repetitions of the hexadecimal string
constant (X'yy...yy') are to appear in the reformatted OUTFIL output
records. n can range from 1 to 4095. If n is omitted, 1 is used.
specify from 1 to 256 pairs of hexadecimal digits.
See Table 33 on page 132 for examples of valid and invalid
hexadecimal string separation.
DATEn, DATEN(c), DATEnP

Constant for current date. The current date of the run is to appear in
the reformatted OUTFIL output records.
DATE1, &DATE1, DATE1(c), &DATE1(c), DATE2, &DATE2, DATE2(c),
&DATE2(c), DATE3, &DATE3, DATE3(c) or &DATE3(c) can be used to
generate a character constant for the current date of the run. DATE1P,
&DATE1P, DATE2P, &DATE2P, DATE3P or &DATE3P can be used to
generate a packed decimal constant for the current date of the run.
Table 39 shows the form of the constant generated for each current
date operand and an example of the actual constant generated when
the date of the run is June 21, 2005 at 4:42:45 PM, using (/) for (c)
where relevant. yyyy represents the year, mm (for date) represents the
month (01-12), dd represents the day (01-31), ddd represents the day of
the year (001-366), hh represents the hour (00-23), mm (for time)
represents the minutes (00-59), ss represents the seconds (00-59), and c
can be any character except a blank.
Table 39. Current date constants
Format of
Operand
Format of Constant
Length (bytes)
Example of Constant
DATE1
C'yyyymmdd'
C'20050621'
DATE1(c)
C'yyyycmmcdd'
10
C'2005/06/21'
DATE1P
P'yyyymmdd'
P'20050621'
DATE2
C'yyyymm'
C'200506'
DATE2(c)
C'yyyycmm'
C'2005/06'
DATE2P
P'yyyymm'
P'200506'
DATE3
C'yyyyddd'
C'2005172'
258

Table 39. Current date constants (continued)
Format of
Operand
Format of Constant
Length (bytes)
Example of Constant
DATE3(c)
C'yyyycddd'
C'2005/172'
DATE3P
P'yyyyddd'
P'2005172'
DATE4
19
C'2005-06-21-16.52.45'
DATE5
C'yyyy-mm-dd-hh.mm.ss.nnnnnn'
26
C'2005-06-21-16.52.45.602837'
Note: You can precede each of the operands in the table with an &
with identical results.
&DATEn, &DATEN(c), &DATEnP
Can be used instead of DATEn, DATEn(c) and DATEnP, respectively
Constant for future date. A future date relative to the current date of
the run is to appear in the reformatted OUTFIL output records.
DATE1+d, &DATE1+d, DATE1(c)+d, &DATE1(c)+d, DATE2+m,
&DATE2+m, DATE2(c)+m, &DATE2(c)+m, DATE3+d, &DATE3+d,
DATE3(c)+d or &DATE3(c)+d can be used to generate a character
constant for a future date relative to the current date of the run.
DATE1P+d, &DATE1P+d, DATE2P+m, &DATE2P+m, DATE3P+d or
&DATE3P+d can be used to generate a packed decimal constant for a
future date relative to the current date of the run. d is days in the
Table 40 shows the form of the constant generated for each future date
operand and an example of the actual constant generated when the
date of the run is June 21, 2005, using (/) for (c) where relevant. yyyy
represents the year, mm (for date) represents the month (01-12), dd
represents the day (01-31), ddd represents the day of the year (001-366),
and c can be any character except a blank.
Table 40. Future Date Constants
Format of
Operand
Format of
Constant
Length (bytes)
Example of
Operand
Example of
Constant
DATE1+d
C'yyyymmdd'
DATE1+11
C'20050702'
DATE1(c)+d
C'yyyycmmcdd'
10
DATE1(/)+90
C'2005/09/19'
DATE1P+d
P'yyyymmdd'
DATE1P+11
P'20050702'
DATE2+m
C'yyyymm'
DATE2+2
C'200508'
DATE2(c)+m
C'yyyycmm'
DATE2(.)+25
C'2007.07'
DATE2P+m
P'yyyymm'
DATE2P+2
P'200508'
DATE3+d
C'yyyyddd'
DATE3+200
C'2006007'
DATE3(c)+d
C'yyyycddd'
DATE3(-)+1
C'2005-171'
DATE3P+d
P'yyyyddd'
DATE3P+200
P'2006007'
259

Can be used instead of DATEn+r, DATEn(c)+r and DATEnP+r,
respectively.
Constant for past date. A past date relative to the current date of the
run is to appear in the reformatted OUTFIL output records.
DATE1-d, &DATE1-d, DATE1(c)-d, &DATE1(c)-d, DATE2-m,
&DATE2-m, DATE2(c)-m, &DATE2(c)-m, DATE3-d, &DATE3-d,
DATE3(c)-d or &DATE3(c)-d can be used to generate a character
constant for a past date relative to the current date of the run.
DATE1P-d, &DATE1P-d, DATE2P-m, &DATE2P-m, DATE3P-d or
&DATE3P-d can be used to generate a packed decimal constant for a
past date relative to the current date of the run. d is days in the past
and m is months in the past. d and m can be 0 to 9999.
Table 41 shows the form of the constant generated for each past date
operand and an example of the actual constant generated when the
date of the run is June 21, 2005, using (/) for (c) where relevant. yyyy
represents the year, mm (for date) represents the month (01-12), dd
represents the day (01-31), ddd represents the day of the year (001-366),
and c can be any character except a blank.
Table 41. Past Date Constants
Format of
Operand
Format of
Constant
Length (bytes)
Example of
Operand
Example of
Constant
DATE1-d
C'yyyymmdd'
DATE1-1
C'20050620'
DATE1(c)-d
C'yyyycmmcdd'
10
DATE1(-)-60
C'2005-04-22'
DATE1P-d
P'yyyymmdd'
DATE1P-30
P'20050522'
DATE2-m
C'yyyymm'
DATE2-6
C'200412'
DATE2(c)-m
C'yyyycmm'
DATE2(/)-1
C'2005/05'
DATE2P-m
P'yyyymm'
DATE2P-12
P'200406'
DATE3-d
C'yyyyddd'
DATE3-300
C'2004238'
DATE3(c)-d
C'yyyycddd'
DATE3(.)-21
C'2005.151'
DATE3P-d
P'yyyyddd'
DATE3P-172
P'2004366'
Can be used instead of DATEn-r, DATEn(c)-r and DATEnP-r,
respectively.
reformatted OUTFIL output records. Table 42 on page 261 shows the
constant generated for each separation field you can specify along with
its length and an example using (:) for (c) where relevant. hh
represents the hour (00-23), mm represents the minutes (00-59), ss
represents the seconds (00-59), and c can be any character except a
blank.
260

Table 42. Current time constants
Separation
Field
Constant
Length (bytes) 01:55:43 PM
TIME1
C'hhmmss'
C'135543'
TIME1(c)
C'hhcmmcss'
C'13:55:43'
TIME1P
P'hhmmss'
P'135543'
TIME2
C'hhmm'
C'1355'
TIME2(c)
C'hhcmm'
C'13:55'
TIME2P
P'hhmm'
P'1355'
TIME3
C'hh'
C'13'
TIME3P
P'hh'
P'13'

Can be used instead of TIMEn, TIMEn(c), and TIMEnP, respectively.
DATE specifies that the current date is to appear in the reformatted OUTFIL
output records in the form 'mm/dd/yy', where mm represents the
month (01-12), dd represents the day (01-31), and yy represents the last
two digits of the year (for example, 04).
&DATE
can be used instead of DATE.
DATE=(abcd)
specifies that the current date is to appear in the reformatted OUTFIL
output records in the form 'adbdc', where a, b, and c indicate the order
in which the month, day, and year are to appear and whether the year
is to appear as two or four digits, and d is the character to be used to
separate the month, day and year.
For a, b, and c, use M to represent the month (01-12), D to represent
the day (01-31), Y to represent the last two digits of the year (for
example, 04), or 4 to represent the four digits of the year (for example,
2004). M, D, and Y or 4 can each be specified only once. Examples:
DATE=(DMY.) would produce a date of the form 'dd.mm.yy', which on
March 29, 2004, would appear as '29.03.04'. DATE=(4MD-) would
produce a date of the form 'yyyy-mm-dd', which on March 29, 2004,
would appear as '2004-03-29'.
a, b, c, and d must be specified.
&DATE=(abcd)
can be used instead of DATE=(abcd).
DATENS=(abc)
output record in the form 'abc', where a, b and c indicate the order in
which the month, day, and year are to appear and whether the year is
to appear as two or four digits.
For a, b and c, use M to represent the month (01-12), D to represent the
day (01-31), Y to represent the last two digits of the year (for example,
04), or 4 to represent the four digits of the year (for example, 2004). M,
D, and Y or 4 can each be specified only once. Examples:
DATENS=(DMY) would produce a date of the form 'ddmmyy', which
261

on March 29, 2004, would appear as '290304'. DATENS=(4MD) would
produce a date of the form 'yyyymmdd', which on March 29, 2004,
would appear as '20040329'.
a, b and c must be specified.
&DATENS=(abc)
can be used instead of DATENS=(abc).
YDDD=(abc)
output records in the form 'acb', where a and b indicate the order in
which the year and day of the year are to appear and whether the year
is to appear as two or four digits, and c is the character to be used to
separate the year and day of the year.
For a and b, use D to represent the day of the year (001-366), Y to
represent the last two digits of the year (for example, 04), or 4 to
represent the four digits of the year (for example, 2004). D, and Y or 4
can each be specified only once. Examples: YDDD=(DY-) would
produce a date of the form 'ddd-yy', which on April 7, 2004, would
appear as '098-04'. YDDD=(4D/) would produce a date of the form
'yyyy/ddd', which on April 7, 2004, would appear as '2004/098'.
&YDDD=(abc)
can be used instead of YDDD=(abc).
YDDDNS=(ab)
output records in the form 'ab', where a and b indicate the order in
which the year and day of the year are to appear and whether the year
is to appear as two or four digits.
represent the last two digits of the year (for example, 04), or 4 to
represent the four digits of the year (for example, 2004). D, and Y or 4
can each be specified only once. Examples: YDDDNS=(DY) would
produce a date of the form 'dddyy', which on April 7, 2004, would
appear as '09804'. YDDDNS=(4D) would produce a date of the form
'yyyyddd', which on April 7, 2004, would appear as '2004098'.
a and b must be specified.
&YDDDNS=(ab)
TIME specifies that the current time is to appear in the reformatted OUTFIL
output records in the form 'hh:mm:ss', where hh represents the hour
(00-23), mm represents the minutes (00-59), and ss represents the
seconds (00-59).
&TIME
can be used instead of TIME.
TIME=(abc)
specifies that the current time is to appear in the reformatted OUTFIL
output records in the form 'hhcmmcss' (24-hour time) or 'hhcmmcss xx'
(12-hour time).
If ab is 24, the time is to appear in the form 'hhcmmcss' (24-hour time)
where hh represents the hour (00-23), mm represents the minutes
262

(00-59), ss represents the seconds (00-59), and c is the character used to
separate the hours, minutes, and seconds. Example: TIME=(24.) would
produce a time of the form 'hh.mm.ss', which at 08:25:13 pm would
appear as '20.25.13'.
If ab is 12, the time is to appear in the form 'hhcmmcss xx' (12-hour
time) where hh represents the hour (01-12), mm represents the minutes
(00-59), ss represents the seconds (00-59), xx is am or pm, and c is the
character used to separate the hours, minutes, and seconds. Example:
TIME=(12.) would produce a time of the form 'hh.mm.ss xx', which at
08:25:13 pm would appear as '08.25.13 pm'.
ab and c must be specified.
&TIME=(abc)
can be used instead of TIME=(abc).
TIMENS=(ab)
specifies that the current time is to appear in the reformatted OUTFIL
output record in the form 'hhmmss' (24-hour time) or 'hhmmss xx'
(12-hour time).
If ab is 24, the time is to appear in the form 'hhmmss' (24-hour time)
where hh represents the hour (00-23), mm represents the minutes
(00-59), and ss represents the seconds (00-59). Example: TIMENS=(24)
would produce a time of the form 'hhmmss', which at 08:25:13 pm
would appear as '202513'.
If ab is 12, the time is to appear in the form 'hhmmss xx' (12-hour
time) where hh represents the hour (01-12), mm represents the minutes
(00-59), and ss represents the seconds (00-59). Example: TIMENS=(12)
would produce a time of the form 'hhmmss xx', which at 08:25:13 pm
would appear as '082513 pm'.
ab must be specified.
&TIMENS=(ab)
can be used instead of TIMENS=(ab).
/.../ or n/
Blank records or a new record. A new output record is to be started
with or without intervening blank output records. If /.../ or n/ is
specified at the beginning or end of OUTREC, n blank output records
are to be produced. If /.../ or n/ is specified in the middle of
OUTREC, n-1 blank output records are to be produced (thus, / or 1/
indicates a new output record with no intervening blank output
records).
At least one input field or separation field must be specified if you use
/.../ or n/. For example, OUTREC=(//) is not allowed, whereas
OUTREC=(//X) is allowed.
Either n/ (for example, 5/) or multiple /'s (for example, /////) can be
used. n can range from 1 to 255. If n is omitted, 1 is used.
As an example, if you specify:
OUTFIL OUTREC=(2/,CField 2 contains ,4,3,/,
CField 1 contains ,1,3)
an input record containing:

111222
263

would produce the following four output records:
Blanks
Blanks
Field 2 contains 222
Field 1 contains 111
Note that four OUTFIL output records are produced for each OUTFIL
input record.
p,m,a
specifies that an unedited input field is to appear in the reformatted OUTFIL
output record.
p
specifies the first byte of the input field relative to the beginning of the
OUTFIL input record. The first data byte of a fixed-length record has
relative position 1. The first data byte of a variable-length record has
relative position 5, because the first four bytes are occupied by the RDW.
All fields must start on a byte boundary, and no field can extend beyond
byte 32752. See OUTFIL statements notes on page 374 for special rules
concerning variable-length records.
specifies the length in bytes of the input field.
specifies the alignment (displacement) of the input field in the reformatted

OUTFIL output record relative to the start of the reformatted OUTFIL
output record.
The permissible values of a are:
H
Halfword aligned. The displacement (p-1) of the field from the

beginning of the reformatted OUTFIL input record, in bytes, is a
multiple of 2 (that is, position 1, 3, 5, and so forth).
Fullword aligned. The displacement is a multiple of 4 (that is,

position 1, 5, 9, and so forth).
Doubleword aligned. The displacement is a multiple of 8 (that is,

Alignment can be necessary if, for example, the data is used in a COBOL
application program where items are aligned through the
SYNCHRONIZED clause. Unused space preceding aligned fields are
always padded with binary zeros.
%nn
OUTFIL output record. See PARSE for details of parsed fields. See p,m,a for
further details. Note that alignment (H, F, D) is not permitted for %nn fields
(for example, %nn,F results in an error message and termination).
p
specifies the unedited variable part of the OUTFIL input record (that part
beyond the minimum record length) is to appear in the reformatted OUTFIL
output record as the last field. p without m can only be used for
variable-length records; not for fixed-length records.
Attention: If 1,4,p is specified (only RDW and variable part of record), "null"
records containing only an RDW will result if the record length is less than p.
OUTFIL input record length plus 1 byte.
264

p,m,TRAN=keyword
Specifies that an input field is to be translated as indicated by the keyword.
Attention: If TRAN=keyword is used for non-character data (for example,
PD), it may have unintended consequences.
p
See p under p,m,a
specifies the length in bytes of the numeric field and depends on the
keyword used as indicated.
TRAN=keyword
specifies the type of translation to be performed. The keyword can be one
of the following:
TRAN=LTOU
translates lowercase EBCDIC letters (that is, a-z) to uppercase EBCDIC
letters (that is, A-Z). Other characters are not changed. For example,
the characters 'Vicky-123,x' would be replaced by 'VICKY-123,X'.
m can be 1 to 32752. The output length will be m bytes.
TRAN=UTOL
translates uppercase EBCDIC letters (that is, A-Z) to lowercase EBCDIC
letters (that is, a-z). Other characters are not changed. For example, the
characters 'CARRIE-005, CA' would be replaced by 'carrie-005, ca'.
TRAN=ALTSEQ
translates characters according to the ALTSEQ translation table in
effect. For example, the characters X'5B5C' would be replaced by
X'4040' if ALTSEQ CODE=(5B40,5C40) was in effect.
TRAN=ATOE
translates characters from ASCII to EBCDIC using the default standard
TCP/IP service ASCII-to-EBCDIC translation table. For example, with
TRAN=ATOE, the ASCII characters aB2 (X'614232') would be replaced
by the EBCDIC characters aB2 (X'81C2F2').
Table 43 on page 266 is the default standard TCP/IP table used for
TRAN=ATOE. Each column shows the ASCII (A) character value on
the left and its equivalent EBCDIC (E) character value on the right.
265

Table 43. TRAN=ATOE ASCII-to-EBCDIC translation
Part 1 - TRAN=ATOE ASCII-to-EBCDIC Translation
| A
|-|00
|01
|02
|03
|04
|05
|06
|07
|08
|09
|0A
|0B
|0C
|0D
|0E
|0F
E| A
--|-00|10
01|11
02|12
03|13
37|14
2D|15
2E|16
2F|17
16|18
05|19
25|1A
0B|1B
0C|1C
0D|1D
0E|1E
0F|1F
E| A
--|-10|20
11|21
12|22
13|23
3C|24
3D|25
32|26
26|27
18|28
19|29
3F|2A
27|2B
22|2C
1D|2D
35|2E
1F|2F
E| A
--|-40|30
5A|31
7F|32
7B|33
5B|34
6C|35
50|36
7D|37
4D|38
5D|39
5C|3A
4E|3B
6B|3C
60|3D
4B|3E
61|3F
E| A
--|-F0|40
F1|41
F2|42
F3|43
F4|44
F5|45
F6|46
F7|47
F8|48
F9|49
7A|4A
5E|4B
4C|4C
7E|4D
6E|4E
6F|4F
E| A
--|-7C|50
C1|51
C2|52
C3|53
C4|54
C5|55
C6|56
C7|57
C8|58
C9|59
D1|5A
D2|5B
D3|5C
D4|5D
D5|5E
D6|5F
E| A
--|-D7|60
D8|61
D9|62
E2|63
E3|64
E4|65
E5|66
E6|67
E7|68
E8|69
E9|6A
AD|6B
E0|6C
BD|6D
5F|6E
6D|6F
E| A
--|-79|70
81|71
82|72
83|73
84|74
85|75
86|76
87|77
88|78
89|79
91|7A
92|7B
93|7C
94|7D
95|7E
96|7F
E|
--|
97|
98|
99|
A2|
A3|
A4|
A5|
A6|
A7|
A8|
A9|
C0|
4F|
D0|
A1|
07|
Part 2 - TRAN=ATOE ASCII-to-EBCDIC Translation

| A
|-|80
|81
|82
|83
|84
|85
|86
|87
|88
|89
|8A
|8B
|8C
|8D
|8E
|8F
E| A
--|-00|90
01|91
02|92
03|93
37|94
2D|95
2E|96
2F|97
16|98
05|99
25|9A
0B|9B
0C|9C
0D|9D
0E|9E
0F|9F
E| A
--|-10|A0
11|A1
12|A2
13|A3
3C|A4
3D|A5
32|A6
26|A7
18|A8
19|A9
3F|AA
27|AB
22|AC
1D|AD
35|AE
1F|AF
E| A
--|-40|B0
5A|B1
7F|B2
7B|B3
5B|B4
6C|B5
50|B6
7D|B7
4D|B8
5D|B9
5C|BA
4E|BB
6B|BC
60|BD
4B|BE
61|BF
E| A
--|-F0|C0
F1|C1
F2|C2
F3|C3
F4|C4
F5|C5
F6|C6
F7|C7
F8|C8
F9|C9
7A|CA
5E|CB
4C|CC
7E|CD
6E|CE
6F|CF
E| A
--|-7C|D0
C1|D1
C2|D2
C3|D3
C4|D4
C5|D5
C6|D6
C7|D7
C8|D8
C9|D9
D1|DA
D2|DB
D3|DC
D4|DD
D5|DE
D6|DF
E| A
--|-D7|E0
D8|E1
D9|E2
E2|E3
E3|E4
E4|E5
E5|E6
E6|E7
E7|E8
E8|E9
E9|EA
AD|EB
E0|EC
BD|ED
5F|EE
6D|EF
E| A
--|-79|F0
81|F1
82|F2
83|F3
84|F4
85|F5
86|F6
87|F7
88|F8
89|F9
91|FA
92|FB
93|FC
94|FD
95|FE
96|FF
E|
--|
97|
98|
99|
A2|
A3|
A4|
A5|
A6|
A7|
A8|
A9|
C0|
4F|
D0|
A1|
07|
TRAN=ETOA
translates characters from EBCDIC to ASCII using the default standard
TCP/IP service EBCDIC-to-ASCII translation table. For example, with
TRAN=ETOA, the EBCDIC characters aB2 (X'81C2F2') would be
replaced by the ASCII characters aB2 (X'614232').
Table 44 on page 267is the default standard TCP/IP table used for
TRAN=ETOA. Each column shows the EBCDIC (E) character value on
the left and its equivalent ASCII (A) character value on the right.
266

Table 44. TRAN=ETOA ASCII-to-EBCDIC translation
Part 1 - TRAN=ETOA ASCII-to-EBCDIC Translation
| E
|-|00
|01
|02
|03
|04
|05
|06
|07
|08
|09
|0A
|0B
|0C
|0D
|0E
|0F
A| E
--|-00|10
01|11
02|12
03|13
1A|14
09|15
1A|16
7F|17
1A|18
1A|19
1A|1A
0B|1B
0C|1C
0D|1D
0E|1E
0F|1F
A| E
--|-10|20
11|21
12|22
13|23
1A|24
0A|25
08|26
1A|27
18|28
19|29
1A|2A
1A|2B
1C|2C
1D|2D
1E|2E
1F|2F
A| E
--|-1A|30
1A|31
1C|32
1A|33
1A|34
0A|35
17|36
1B|37
1A|38
1A|39
1A|3A
1A|3B
1A|3C
05|3D
06|3E
07|3F
A| E
--|-1A|40
1A|41
16|42
1A|43
1A|44
1E|45
1A|46
04|47
1A|48
1A|49
1A|4A
1A|4B
14|4C
15|4D
1A|4E
1A|4F
A| E
--|-20|50
A6|51
E1|52
80|53
EB|54
90|55
9F|56
E2|57
AB|58
8B|59
9B|5A
2E|5B
3C|5C
28|5D
2B|5E
7C|5F
A| E
--|-26|60
A9|61
AA|62
9C|63
DB|64
A5|65
99|66
E3|67
A8|68
9E|69
21|6A
24|6B
2A|6C
29|6D
3B|6E
5E|6F
A| E
--|-2D|70
2F|71
DF|72
DC|73
9A|74
DD|75
DE|76
98|77
9D|78
AC|79
BA|7A
2C|7B
25|7C
5F|7D
3E|7E
3F|7F
A|
--|
D7|
88|
94|
B0|
B1|
B2|
FC|
D6|
FB|
60|
3A|
23|
40|
27|
3D|
22|
Part 2 - TRAN=ETOA ASCII-to-EBCDIC Translation

| E
|-|80
|81
|82
|83
|84
|85
|86
|87
|88
|89
|8A
|8B
|8C
|8D
|8E
|8F
A| E
--|-F8|90
61|91
62|92
63|93
64|94
65|95
66|96
67|97
68|98
69|99
96|9A
A4|9B
F3|9C
AF|9D
AE|9E
C5|9F
A| E
--|-8C|A0
6A|A1
6B|A2
6C|A3
6D|A4
6E|A5
6F|A6
70|A7
71|A8
72|A9
97|AA
87|AB
CE|AC
93|AD
F1|AE
FE|AF
A| E
--|-C8|B0
7E|B1
73|B2
74|B3
75|B4
76|B5
77|B6
78|B7
79|B8
7A|B9
EF|BA
C0|BB
DA|BC
5B|BD
F2|BE
F9|BF
A| E
--|-B5|C0
B6|C1
FD|C2
B7|C3
B8|C4
B9|C5
E6|C6
BB|C7
BC|C8
BD|C9
8D|CA
D9|CB
BF|CC
5D|CD
D8|CE
C4|CF
A| E
--|-7B|D0
41|D1
42|D2
43|D3
44|D4
45|D5
46|D6
47|D7
48|D8
49|D9
CB|DA
CA|DB
BE|DC
E8|DD
EC|DE
ED|DF
A| E
--|-7D|E0
4A|E1
4B|E2
4C|E3
4D|E4
4E|E5
4F|E6
50|E7
51|E8
52|E9
A1|EA
AD|EB
F5|EC
F4|ED
A3|EE
8F|EF
A| E
--|-5C|F0
E7|F1
53|F2
54|F3
55|F4
56|F5
57|F6
58|F7
59|F8
5A|F9
A0|FA
85|FB
8E|FC
E9|FD
E4|FE
D1|FF
A|
--|
30|
31|
32|
33|
34|
35|
36|
37|
38|
39|
B3|
F7|
F0|
FA|
A7|
FF|
TRAN=HEX
translates binary values to their equivalent EBCDIC hexadecimal
values. For example, with TRAN=HEX, X'C1F1' (C'A1') would be
replaced by C'C1F1'.
m can be 1 to 16376. The output length will be m*2 bytes
TRAN=UNHEX
translates EBCDIC hexadecimal values to their equivalent binary
values. For example, with TRAN=UNHEX, C'C1F1' would be replaced
by X'C1F1' (C'A1').
m can be 1 to 32752. The output length will be (m+1)/2 bytes.
If m is odd, the last nibble will be 0. As an example, C'C1F1F' would
translate to X'C1F1F0' (C'A10').
If an input character is not 0-9 or A-F, the equivalent nibble will be set
to 0. As an example, C'G2' will be translated to X'02'.
267

TRAN=BIT
translates binary values to their equivalent EBCDIC bit values. For
example, with TRAN=BIT, X'C1F1' (C'A1') would be replaced by
C'1100000111110001'.
m can be 1 to 4094. The output length will be m*8 bytes
TRAN=UNBIT
translates EBCDIC bit values to their equivalent binary values. For
example, with TRAN=UNBIT, C'1100000111110001' would be replaced
by X'C1F1' (C'A1').
m can be 1 to 32752. The output length will be (m+7)/8 bytes
If m is not a multiple of 8, the missing bits on the right will be set to 0.
As an example, C'1100000111111' would translate to X'C1F8'.
If an input character is not 0 or 1, the equivalent bit will be set to 0. As
an example, C'11100005' will be translated to X'E0'.
Sample Syntax:
OUTFIL FNAMES=OUT1,
BUILD=(5,6,TRAN=UNHEX,2X,20,15,TRAN=UTOL)
%nn,TRAN=keyword
keyword. See PARSE for details of PARSED fields. See p,m,TRAN=keyword for
details of the translation functions.
p,TRAN=keyword
indicated by the keyword. p,TRAN=keyword must be the last item and can
only be used for variable-length records; not for fixed-length records. A value
must be specified for p that is less than or equal to the minimum input record
length plus 1 byte. See p,m,TRAN=keyword for details of the translation
functions.
Attention: If 1,4,p,TRAN=keyword is specified (only RDW and variable part
of record), "null" records containing only an RDW will result if the record
length is less than p.
Sample Syntax:
OUTFIL FNAMES=OUT2,BUILD=(1,4,5,TRAN=ETOA)
p,m,HEX
Can be used instead of p,m,TRAN=HEX. See p,m,TRAN=HEX for details.
%nn,HEX
Can be used instead of %nn,TRAN=HEX. See %nn,TRAN=HEX for details.
p,HEX
Can be used instead of p,TRAN=HEX. See p,TRAN=HEX for details.
OUTFIL output record. You can edit BI, FI, PD, PD0, ZD, FL,CSF, FS, UFF, SFF,
DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3, TC4, TE1,
TE2, TE3, TE4, TM1, TM2, TM3 or TM4 fields using either pre-defined edit
masks (M0-M26) or specific edit patterns you define. You can control the way
the edited fields look with respect to length, leading or suppressed zeros,
thousands separators, decimal points, leading and trailing positive and
negative signs, and so on.
268

p
See p under p,m,a.
specifies the length in bytes of the numeric field. The length must include
the sign, if the data is signed. See Table 45 for permissible length values.
specifies the format of the numeric field:
Table 45. Edit Field Formats and Lengths

Format Code
Length
Description
BI
1 to 8 bytes
Unsigned binary
FI
1 to 8 bytes
Signed fixed-point
PD
1 to 16 bytes
PD0
2 to 8 bytes
Packed decimal with sign and first

digit ignored
ZD
1 to 31 bytes
FL
4 or 8 bytes
Signed hexadecimal floating-point

converted to signed integer
CSF or FS
1 to 32 bytes (31 digit limit)
Signed numeric with optional leading

floating sign
UFF
SFF
DT1
4 bytes
SMF date interpreted as

Z'yyyymmdd'
DT2
4 bytes
SMF date interpreted as Z'yyyymm'
DT3
4 bytes
SMF date interpreted as Z'yyyyddd'
DC1
8 bytes
TOD date interpreted as

Z'yyyymmdd'
DC2
8 bytes
TOD date interpreted as Z'yyyymm'
DC3
8 bytes
TOD date interpreted as Z'yyyyddd'
DE1
8 bytes
ETOD date interpreted as

Z'yyyymmdd'
DE2
8 bytes
ETOD date interpreted as Z'yyyymm'
DE3
8 bytes
ETOD date interpreted as Z'yyyyddd'
TM1
4 bytes
SMF time interpreted as Z'hhmmss'
TM2
4 bytes
SMF time interpreted as Z'hhmm'
TM3
4 bytes
SMF time interpreted as Z'hh'
TM4
4 bytes
SMF time interpreted as Z'hhmmssxx'
TC1
8 bytes
TOD time interpreted as Z'hhmmss'
TC2
8 bytes
TOD time interpreted as Z'hhmm'
TC3
8 bytes
TOD time interpreted as Z'hh'
TC4
8 bytes
TOD time interpreted as

Z'hhmmssxx'
TE1
8 bytes
ETOD time interpreted as Z'hhmmss'
TE2
8 bytes
ETOD time interpreted as Z'hhmm'
TE3
8 bytes
ETOD time interpreted as Z'hh'
TE4
8 bytes
ETOD time interpreted as

Z'hhmmssxx'
269

Table 45. Edit Field Formats and Lengths (continued)
Format Code
Length
Description
Note: See Appendix C, "Data Format

Descriptions" for detailed format
descriptions.
For a CSF or FS format field:

v A maximum of 31 digits is allowed. If a CSF or FS value with 32 digits is found,
the leftmost digit will be treated as a positive sign indicator.
For a UFF or SFF format field:
v A maximum of 31 digits is allowed. If a UFF or SFF value with more than 31
digits is found, the leftmost digits will be ignored.
For a ZD or PD format field:
v An invalid digit results in a data exception (0C7 ABEND) or incorrect numeric
output; A-F are invalid digits. ICETOOL's VERIFY or DISPLAY operator can be
used to identify decimal values with invalid digits.
v A value is treated as positive if its sign is F, E, C, A, 8, 6, 4, 2, or 0.
v A value is treated as negative if its sign is D, B, 9, 7, 5, 3, or 1.
For a PD0 format field:
v The first digit is ignored.
v An invalid digit other than the first results in a data exception (0C7 ABEND) or
incorrect numeric output; A-F are invalid digits.
v The sign is ignored and the value is treated as positive.
For an FL format field:
v The normalized or unnormalized FL (hexadecimal floating-point) value is
converted to a signed integer in the range -9223372036854775808 to
9223372036854775807. The fractional part of the FL value is lost, and in some
cases the signed integer may be one of a number of possible signed integers for
the FL value depending on its precision. Converted values less than
-9223372036854775808 are set to -9223372036854775808. Converted values greater
than 9223372036854775807 are set to 9223372036854775807.
v If you are not running in z/Architecture mode, specifying an FL format field
results in an error message and termination.
For a DT1, DT2, or DT3 format field:
v An invalid SMF date can result in a data exception (0C7 ABEND) or an incorrect
ZD date.
v SMF date values are always treated as positive.
For a DC1, DC2, DC3, DE1, DE2, or DE3 format field:
v TOD and ETOD date values are always treated as positive.
For a TM1, TM2, TM3, or TM4 format field:
v An invalid SMF time can result in an incorrect ZD time.
v SMF time values are always treated as positive.
For a TC1, TC2, TC3, TC4, TE1, TE2, TE3, or TE4 format field:
270

v TOD and ETOD time values are always treated as positive.
edit
Mn
EDIT=
EDxy=

(pattern)
('pattern')

,
SIGNS=
SIGNz=
(lp,ln,tp,tn)
,LENGTH=n
Specifies how the numeric field is to be edited for output. If an Mn, EDIT, or
EDxy parameter is not specified:
v a DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3, TC4,
TE1, TE2, TE3, TE4, TM1, TM2, TM3, or TM4 field is edited using the M11
edit mask.
v a BI, FI, PD, PD0, ZD, FL, CSF, FS, UFF, or SFF field is edited using the M0
edit mask.
Mn specifies one of twenty-seven pre-defined edit masks (M0-M26) for
presenting numeric data. If these pre-defined edit masks are not suitable
for presenting your numeric data, the EDIT parameter gives you the
flexibility to define your own edit patterns.
The twenty-seven pre-defined edit masks can be represented as follows:
Table 46. Edit Mask Patterns
Mask
M0
M1
M2
M3
Pattern
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIITS
TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS
II,III,III,III,III,III,III,III,III,IIT.TTS
II,III,III,III,III,III,III,III,III,IIT.TTCR
Examples
Value
Result
+01234
1234
-00001
1-
-00123
00123-
+00123
00123
+123450
1,234.50
-000020
0.20-
-001234
12.34CR
+123456
M4
M5
SII,III,III,III,III,III,III,III,III,IIT.TT
SII,III,III,III,III,III,III,III,III,IIT.TTS
+0123456
+1,234.56
-1234567
-12,345.67
-001234
(12.34)
+123450
M6
M7
M8
III-TTT-TTTT
TTT-TT-TTTT
IT:TT:TT
1,234.56
1,234.50
00123456
012-3456
12345678
1-234-56788
00123456
000-12-3456
12345678
012-34-5678
030553
3:05:53
121736
12:17:36
271

Table 46. Edit Mask Patterns (continued)
Mask
M9
M10
M11
M12
M13
M14
Pattern
Examples
IT/TT/TT
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIT
TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT
SI,III,III,III,III,III,III,III,III,III,IIT
SI.III.III.III.III.III.III.III.III.III.IIT
SI III III III III III III III III III IITS
Value
Result
123004
12/30/04
083104
8/31/04
01234
1234
00000
00010
00010
01234
01234
+1234567
1,234,567
-0012345
-12,345
+1234567
1.234.567
-0012345
-12.345
+1234567
-0012345
M15
M16
M17
M18
M19
M20
M21
M22
M23
M24
M25
M26
I III III III III III III III III III IITS
SI III III III III III III III III III IIT
SI'III'III'III'III'III'III'III'III'III'IIT
SII,III,III,III,III,III,III,III,III,IIT.TT
SII.III.III.III.III.III.III.III.III.IIT,TT
SI III III III III III III III III IIT,TTS
II III III III III III III III III IIT,TTS
SI III III III III IIII III III III IIT,TT
SII'III'III'III'III'III'III'III'III'IIT.TT
SII'III'III'III'III'III'III'III'III'IIT,TT
SIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIT
STTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT
+1234567
1 234 567
(12 345)
1 234 567
-0012345
12 345-
+1234567
1 234 567
-0012345
-12 345
+1234567
1'234'567
-0012345
-12'345
+0123456
1,234.56
-1234567
-12,345.67
+0123456
1.234,56
-1234567
-12.345,67
+0123456
1 234,56
-1234567
(12 345,67)
+0123456
1 234,567
-1234567
12 345,67-
+0123456
1 234,56
-1234567
-12 345,67
+0123456
1'234.56
-1234567
-12'345.67
+0123456
1'234,56
-1234567
-12'345,67
+01234
1234
-00001
-1
1234
+01234
-1
-00001
The elements used in the representation of the edit masks in Table 46 on page 271
are as follows:
272

v I indicates a leading insignificant digit. If zero, this digit will not be shown.
v T indicates a significant digit. If zero, this digit will be shown.
v CR (in M3) is printed to the right of the digits if the value is negative;
otherwise, two blanks are printed to the right of the digits.
v S indicates a sign. If it appears as the first character in the pattern, it is a leading
sign. If it appears as the last character in the pattern, it is a trailing sign. If S
appears as both the first and last characters in a pattern (example: M5), the first
character is a leading sign and the last character is a trailing sign. Four different
sign values are used: leading positive sign (lp), leading negative sign (ln),
trailing positive sign (tp) and trailing negative sign (tn). Their applicable values
for the Mn edit masks are:
Table 47. Edit Mask Signs
Mask
lp
ln
tp
tn
M0
none
none
blank
M1
none
none
blank
M2
none
none
blank
M3
none
none
none
none
M4
none
none
M5
blank
blank
M6
none
none
none
none
M7
none
none
none
none
M8
none
none
none
none
M9
none
none
none
none
M10
none
none
none
none
M11
none
none
none
none
M12
blank
none
none
M13
blank
none
none
M14
blank
blank
M15
none
none
blank
M16
blank
none
none
M17
blank
none
none
M18
blank
none
none
M19
blank
none
none
M20
blank
blank
M21
none
none
blank
M22
blank
none
none
M23
blank
none
none
M24
blank
none
none
M25
blank
none
none
M26
none
none
v any other character (for example, /) will be printed as shown, subject to certain
rules to be subsequently discussed.
273

The implied length of the edited output field depends on the number of digits and
characters needed for the pattern of the particular edit mask used. The LENGTH
parameter can be used to change the implied length of the edited output field.
The number of digits needed depends on the format and length of the numeric
field as follows:
Table 48. Digits Needed for Numeric Fields
Format
Input Length
Digits Needed
ZD
PD
2m-1
PD0
2m-2
BI, FI
BI, FI
BI, FI
BI, FI
10
BI, FI
13
BI, FI
15
BI, FI
17
BI, FI
20
FL
4 or 8
20
CSF or FS
32
31
CSF or FS
m (less than 32)
UFF, SFF
32 to 44
31
UFF, SFF
m (less than 32)
The length of the output field can be represented as follows for each pattern,
where d is the number of digits needed, as shown in Table 48, and the result is
rounded down to the nearest integer:
Table 49. Edit Mask Output Field Lengths
Mask
Output Field Length
Example
Input (f,m)
Output Length
M0
d+1
ZD,3
M1
d+1
PD,10
20
M2
d + 1 + d/3
BI,4
14
M3
d + 2 + d/3
UFF,20
28
M4
d + 1 + d/3
PD,8
21
M5
d + 2 + d/3
FI,3
12
M6
12
ZD,10
12
M7
11
PD,5
11
M8
ZD,6
M9
PD,4
M10
BI,6
15
M11
PD,5
274

Table 49. Edit Mask Output Field Lengths (continued)
Mask
Output Field Length
Example
Input (f,m)
Output Length
M12
d + 1 + (d - 1)/3
PD,3
M13
d + 1 + (d - 1)/3
FS,5
M14
d + 2 + (d - 1)/3
ZD,5
M15
d + 1 + (d - 1)/3
FI,3
11
M16
d + 1 + (d - 1)/3
SFF,41
42
M17
d + 1 + (d - 1)/3
FI,4
14
M18
d + 1 + d/3
BI,4
14
M19
d + 1 + d/3
PD,8
21
M20
d + 2 + d/3
FI,3
12
M21
d + 1 + d/3
ZD,3
M22
d + 1 + d/3
BI,2
M23
d + 1 + d/3
PD,6
15
M24
d + 1 + d/3
ZD,21
29
M25
d+1
CSF,16
17
M26
d+1
FL,4
21
To illustrate conceptually how DFSORT produces the edited output from the
numeric value, consider the following example:
OUTFIL OUTREC=(5,7,ZD,M5)
with ZD values of C0123456(+0123456)
and C000302J (-0003021)
As shown in the preceding tables, it is determined that:

v The general pattern for M5 is: SI,III,...,IIT.TTS
v The signs to be used are blank for leading positive sign, C'(' for leading negative
sign, blank for trailing positive sign and C')' for trailing negative sign
v The number of digits needed is 7
v The length of the output field is 11 (7 + 2 + 7/3)
v The specific pattern for the output field is thus: C'SII,IIT.TTS'
The digits of C'0123456' are mapped to the pattern, resulting in C'S01,234.56S'.
Because the value is positive, the leading sign is replaced with blank and the
trailing sign is replaced with blank, resulting in C' 01,234.56 '. Finally, all digits
before the first non-zero digit (1 in this case), are replaced with blanks, resulting in
the final output of C' 1,234.56 '.
The digits of C'000302J' are mapped to the pattern, resulting in C'S00,030.21S'.
Because the value is negative, the leading sign is replaced with C'(' and the trailing
sign is replaced with C')' resulting in C'(00,030.21)'. All digits before the first
non-zero digit (3 in this case), are replaced with blanks, resulting in C'(
30.21)'.
Finally, the leading sign is "floated" to the right, next to the first non-zero digit,
resulting in the final output of C'
(30.21)'.
275

To state the rules in more general terms, the steps DFSORT takes conceptually to
produce the edited output from the numeric value are as follows:
v Determine the specific pattern and its length, using the preceding tables.
v Map the digits of the numeric value to the pattern.
v If the value is positive, replace the leading and trailing signs (if any) with the
characters for positive values shown in Table 47 on page 273. Otherwise, replace
the leading and trailing signs (if any) with the characters for negative values
shown in that same table.
v Replace all digits before the first non-zero (I) or significant digit (T) with blanks.
v Float the leading sign (if any) to the right, next to the first non-zero (I) or
significant digit (T).
The following additional rule applies to edit masks:
v The specific pattern is determined from the general pattern by including signs,
the rightmost digits needed as determined from the input format and length,
and any characters in between those rightmost digits. This may unintentionally
truncate significant digits (T). As an example, if you specify 5,1,ZD,M4, the
length of the output field will be 2 (1 + 1 + 1/3). The general pattern for M4 is
SI,III,...,IIT.TT, but the specific pattern will be ST (the leading sign and the
rightmost digit).
EDIT
specifies an edit pattern for presenting numeric data. If the pre-defined edit
masks (M0-M26) are not suitable for presenting your numeric data, EDIT gives
you the flexibility to define your own edit patterns. The elements you use to
specify the pattern are the same as those used for the edit masks: I, T, S, and
printable characters. However, S will not be recognized as a sign indicator
unless the SIGNS parameter is also specified.
pattern
specifies the edit pattern to be used. Not enclosing the pattern in single
apostrophes restricts you from specifying the following characters in the
pattern: blank, apostrophe, unbalanced left or right parentheses, and
hexadecimal digits 20, 21, and 22. For example, EDIT=((IIT.TT)) is valid,
whereas EDIT=(C)ITT.TT), EDIT=(I / T) and EDIT=(S'II.T) are not.
The maximum number of digits (I's and T's) you specify in the pattern
must not exceed 31. The maximum length of the pattern must not exceed
44 characters.
'pattern'
specifies the edit pattern to be used. Enclosing the pattern in single
apostrophes allows you to specify any character in the pattern except
hexadecimal digits 20, 21, or 22. If you want to include a single apostrophe
in the pattern, you must specify it as two single apostrophes, which will be
counted as a single character in the pattern. As examples,
EDIT=('C)ITT.TT'), EDIT=('I / T'), and EDIT=('S'II.T') are all valid.
The maximum number of digits (I's and T's) you specify in the pattern
must not exceed 31. The maximum length of the pattern must not exceed
44 characters.
The implied length of the edited output field is the same as the length of the
pattern. The LENGTH parameter can be used to change the implied length of
the edited output field.
276

To illustrate conceptually how DFSORT produces the edited output from the
numeric value, consider the following example:
OUTFIL OUTREC=(1,5,ZD,EDIT=(**I/ITTTCR))
with ZD values of C01230(+1230)
and C0004J (-41)
The digits of C'01230' are mapped to the pattern, resulting in C'**0/1230CR'.

Because the value is positive, the characters (C'CR') to the right of the last digit
are replaced with blanks, resulting in C'**0/1230 '. All digits before the first
non-zero digit (1 in this case) are replaced with blanks, resulting in
C'** /1230 '. Finally, all characters before the first digit in the pattern are
floated to the right, next to the first non-zero digit, resulting in C' **1230 '.
The digits of C'0004J' are mapped to the pattern, resulting in C'**0/0041CR'.
Because the value is negative, the characters (C'CR') to the right of the last
digit are kept. All digits before the first T digit are replaced with blanks,
resulting in C'** / 041CR'. Finally, all characters before the first digit in the
pattern are floated to the right, next to the first non-zero digit, resulting in
C' **041CR'.
In general terms, the steps DFSORT takes conceptually to produce the edited
output from the numeric value are as follows:
v Map the digits of the numeric value to the pattern, padding on the left with
zeros, if necessary.
v If the value is positive, replace the leading and trailing signs (if any) with
the characters for positive values specified by the SIGNS parameter and
replace any characters between the last digit and the trailing sign (if any)
with blanks. Otherwise, replace the leading and trailing signs (if any) with
the characters for negative values specified by the SIGNS parameter and
keep any characters between the last digit and the trailing sign (if any).
v Replace all digits before the first non-zero (I) or significant digit (T) with
blanks.
v Float all characters (if any) before the first digit in the pattern to the right,
next to the first non-zero (I) or significant digit (T).
The following additional rules apply to edit patterns:
v An insignificant digit (I) after a significant digit (T) is treated as a significant
digit.
v If SIGNS is specified, an S in the first or last character of the pattern is
treated as a sign; an S anywhere else in the pattern is treated as the letter S.
If SIGNS is not specified, an S anywhere in the pattern is treated as the letter
S.
v If the pattern contains fewer digits than the value, the leftmost digits of the
value will be lost, intentionally or unintentionally. As an example, if you
specify 5,5,ZD,EDIT=(IIT) for a value of C'12345', the result will be C'345'. As
another example, if you specify 1,6,ZD,EDIT=($IIT.T) for a value of
C'100345', the result will be C' $34.5'.
EDxy
specifies an edit pattern for presenting numeric data. EDxy is a special
variation of EDIT that allows other characters to be substituted for I and T in
the pattern. For example, if you use EDAB instead of EDIT, you must use A in
the pattern instead of I and use B instead of T to represent digits. x and y must
not be the same character. If SIGNS is specified, x and y must not be S. If
SIGNz is specified, x and y must not be the same character as z. You can select
x and y from: A-Z, #, $, @, and 0-9.
277

SIGNS
specifies the sign values to be used when editing numeric values according to
the edit mask (Mn) or pattern (EDIT or EDxy). You can specify any or all of
the four sign values. Any value not specified must be represented by a comma.
Blank will be used for any sign value you do not specify. As examples,
SIGNS=(+,-) specifies + for lp, - for ln, blank for tp, and blank for tn;
SIGNS=(,,+,-) specifies blank for lp, blank for ln, + for tp, and - for tn.
lp specifies the value for the leading positive sign. If an S is specified as the
first character of the edit mask or pattern and the value is positive, the lp
value will be used as the leading sign.
ln specifies the value for the leading negative sign. If an S is specified as the
first character of the edit mask or pattern and the value is negative, the ln
value will be used as the leading sign.
tp specifies the value for the trailing positive sign. If an S is specified as the
last character of the edit mask or pattern and the value is positive, the tp
value will be used as the trailing sign.
tn specifies the value for the trailing negative sign. If an S is specified as the
last character of the edit mask or pattern and the value is negative, the tn
value will be used as the trailing sign.
If you want to use any of the following characters as sign values, you must
enclose them in single apostrophes: comma, blank, or unbalanced left or right
parentheses. A single apostrophe must be specified as four single apostrophes
(that is, two single apostrophes enclosed in single apostrophes).
A semicolon cannot be substituted for a comma as the delimiter between sign
characters.
SIGNz
specifies the sign values to be used when editing numeric values according to
the edit pattern (EDIT or EDxy). SIGNz is a special variation of SIGNS which
allows another character to be substituted for S in the pattern. For example, if
you use SIGNX instead of SIGNS, you must use X in the pattern instead of S
to identify a sign. If EDIT is specified, z must not be I or T. If EDxy is
specified, z must not be the same character as either x or y. You can select z
from: A-Z, #, $, @, and 0-9.
LENGTH
specifies the length of the edited output field. If the implied length of the
edited output field produced using an edit mask or edit pattern is not suitable
for presenting your numeric data, LENGTH can be used to make the edited
output field shorter or longer.
n
specifies the length of the edited output field. The value for n must be
between 1 and 44.
LENGTH does not change the pattern used, only the length of the
resulting edited output field. For example, as discussed previously for Mn,
if you specify:
OUTFIL OUTREC=(5,1,ZD,M4)
the pattern will be C'ST' rather than C'ST.TT' because the digit length is 1.
Specifying:
OUTFIL OUTREC=(5,1,ZD,M4,LENGTH=5)
will change the pattern to C'
278
ST', not to C'ST.TT'.

If you specify a value for n that is shorter than the implied length,
truncation will occur on the left after editing. For example, if you specify:
OUTFIL OUTREC=(1,5,ZD,EDIT=($IIT.TT),LENGTH=5)
with a value of C'12345', editing according to the specified $IIT.TT pattern

will produce C'$123.45', but the specified length of 5 will truncate this to
C'23.45'.
If you specify a value for n that is longer than the implied length, padding
on the left with blanks will occur after editing. For example, if you specify:
OUTFIL OUTREC=(1,5,ZD,EDIT=($IIT.TT),LENGTH=10)
with a value of C'12345', editing according to the specified $IIT.TT pattern

will produce C'$123.45', but the specified length of 10 will pad this to
C' $123.45'.
Sample Syntax:
OUTFIL FNAMES=OUT1,OUTREC=(5:21,19,ZD,M19,35:46,5,ZD,M13)
OUTFIL FILES=1,OUTREC=(4,8,BI,C * ,13,8,BI,80:X),
ENDREC=10,OMIT=(4,8,BI,EQ,13,8,BI)
OUTFIL FILES=(2,3),
OUTREC=(11:55,6,FS,SIGNS=(,,+,-),LENGTH=10,
31:(41,4,PD),EDIT=(**II,IIT.TTXS),SIGNS=(,,+,-))
reformatted OUTFIL output record. See PARSE for details of parsed fields.
See p,m,f,edit or (p,m,f),edit for further details.
specifies that a converted numeric input field is to appear in the
reformatted OUTFIL output record. You can convert BI, FI, PD, PD0, ZD,
FL, CSF, FS, UFF, SFF, DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3,
TC1, TC2, TC3, TC4, TE1, TE2, TE3, TE4, TM1, TM2, TM3, or TM4 fields to
BI, FI, PD, PDC, PDF, ZD, ZDF, ZDC, or CSF/FS fields.
p
See p under p,m,f,edit.
See m under p,m,f,edit.
See f under p,m,f,edit.
to
fo
TO=

fo
(fo)
,LENGTH=
n
(n)
specifies how the numeric field is to be converted for output.

fo specifies the format for the output field, which can be BI, FI, PD, PDC, PDF,
ZD, ZDF, ZDC, CSF or FS. Any one of these output field formats (fo) can be
used with any one of the input field formats (f).
If you do not specify the LENGTH parameter, DFSORT will determine the
implied length of the output field from the length (m) and format (f) of the
input field and the format (fo) of the output field. The implied length of the
output field can be represented as follows for each output format, where d is
the number of digits needed for the input field as shown in Table 48 on page
274, and the result is rounded down to the nearest integer:
279

Table 50. To Output Field Lengths
Output Format
Output Length
Example Input (f,m)
Example Output
Length
BI with d <= 9
FS,9
BI with d > 9
FS,10
FI with d <= 9
ZD,7
FI with d > 9
ZD,12
PD, PDC, or PDF
d/2 + 1
BI,4
ZD, ZDF, or ZDC
PD,9
17
CSF or FS
d+1
FI,4
21
For ZD or ZDF output, F is used as the positive sign and D is used as the
negative sign. For ZDC output, C is used as the positive sign.
For PD or PDC output, C is used as the positive sign and D is used as the
negative sign. For PDF output, F is used as the positive sign.
For CSF or FS output, blank is used as the positive sign, - is used as the
negative sign and leading zeros are suppressed.
For ZD, ZDF, ZDC, PD, PDC, PDF, CSF, or FS output, the maximum output
value is 9999999999999999999999999999999 (31 digits) and the minimum output
value is -9999999999999999999999999999999 (31 digits), which correspond to
the maximum and minimum input values, respectively.
For BI output:
v An input value greater than 18446744073709551615 (X'FFFFFFFFFFFFFFFF')
produces an output value of 18446744073709551615 (X'FFFFFFFFFFFFFFFF').
v An input value less than zero produces an absolute output value. For
example, an input value of P'-5000' produces a BI output value of 5000
(X'1388').
For FI output, an input value greater than 9223372036854775807
(X'7FFFFFFFFFFFFFFF') produces an output value of 9223372036854775807
(X'7FFFFFFFFFFFFFFF'), and an input value less than -9223372036854775808
(X'8000000000000000') produces an output value of -9223372036854775808
(X'8000000000000000').
fo, TO=fo and TO=(fo) are interchangeable except that:
v fo must be specified before the LENGTH parameter whereas TO can be
specified before or after the LENGTH parameter.
v TO=fo or TO=(fo) should be used after a symbol rather than fo to prevent
the misinterpretation of fo as f. See the discussion of OUTFIL OUTREC in
Chapter 7 for details.
LENGTH
specifies the length of the converted output field. If the implied length of the
output field is not suitable, LENGTH can be used to make the output field
shorter or longer.
n
specifies the length of the converted output field. The value for n must be
between 1 and 44.
If you specify a value for n that is shorter than the implied length,
truncation on the left will occur after conversion. For example, if you
specify:
280

OUTFIL OUTREC=(1,8,ZD,TO=PD,LENGTH=3)
with values of ZL8'-12345678' (X'F1F2F3F4F5F6F7D8') and ZL8'58'

(X'F0F0F0F0F0F0F5F8'), conversion with the implied length (5) will produce
PL5'-12345678' (X'012345678D') and PL5'58' (X'0000000058C'). The specified
length of 3 will then result in truncation to PL3'-45678' (X'45678D') and
PL3'58' (X'00058C').
If you specify a value for n that is longer than the implied length, padding
on the left will occur after conversion using:
v Blanks for CSF and FS output values
v Character zeros for ZD output values
v Binary zeros for PD and BI output values
v Binary zeros for positive FI output values
v Binary ones for negative FI output values
For example, if you specify:
OUTFIL OUTREC=(1,4,ZD,TO=FI,LENGTH=6)
with values of ZL4'-1234' (X'F1F2F3D4') and ZL4'58' (X'F0F0F5F8'),

conversion with the implied length (4) will produce FL4'-1234'
(X'FFFFFB2E') and FL4'58' (X'000004D2'). The specified length of 6 will then
result in padding to FL6'-1234' (X'FFFFFFFFFB2E') and FL6'58'
(X'00000000003A').
Sample Syntax:
OUTFIL OUTREC=((21,5,ZD),PD,X,8,4,ZD,TO=FI,LENGTH=2)
reformatted OUTFIL output record. See PARSE for details of parsed fields. See
p,m,f,to or (p,m,f),to for further details.
specifies that an edited decimal constant is to appear in the reformatted
OUTFIL output record. The decimal constant must be in the form +n or -n
where n is 1 to 31 decimal digits. The sign (+ or -) must be specified. A
decimal constant produces a signed, 31-digit zoned decimal (ZD) result to be
edited as specified. If an Mn, EDIT, or EDxy parameter is not specified, the
decimal constant is edited using the M0 edit mask.
The default number of digits (d) used for editing is 15 for a decimal constant
with 1 to 15 significant digits, or 31 for a decimal constant with 16 to 31
significant digits. If EDIT or EDxy is specified, the number of digits in the
pattern (I's and T's) is used.
See edit under p,m,f,edit for further details on the edit fields you can use.
Sample Syntax:
OUTFIL OUTREC=(5,8,+4096,2X,-17,M18,LENGTH=7,2X,
(+2000000),EDIT=(STTTTT.TT),SIGNS=(+))
OUTFIL output record. The decimal constant must be in the form +n or -n
where n is 1 to 31 decimal digits. The sign (+ or -) must be specified. A
decimal constant produces a signed, 31-digit zoned decimal (ZD) result to be
281

The default number of digits (d) used for conversion is 15 for a decimal
constant with 1 to 15 significant digits, or 31 for a decimal constant with 16 to
31 significant digits.
See to under p,m,f,to for further details on the to fields you can use.
Sample Syntax:
OUTFIL FNAMES=OUT1,
OUTREC=(6:+0,TO=PD,LENGTH=6,+0,TO=PD,LENGTH=6,/,
6:(-4096),ZD,LENGTH=12)
Specifies that the edited result of an arithmetic expression is to appear in the
reformatted OUTFIL output record. An arithmetic expression takes the form:
term,operator,term<,operator,...>
where:
v term is a field (p,m,f), a parsed field (%nn,f), or a decimal constant (+n or
-n). See p,m,f under p,m,f,edit for details on the fields you can use. See
deccon under deccon,edit for details on the decimal constants you can use.
v operator is MIN (minimum), MAX (maximum), MUL (multiplication), DIV
(division), MOD (modulus), ADD (addition) or SUB (subtraction).
The order of evaluation precedence for the operators is as follows unless it is
changed by parentheses:
1. MIN and MAX
2. MUL, DIV and MOD
3. ADD and SUB
The intermediate or final result of a DIV operation is rounded down to the
nearest integer. The intermediate or final result of a MOD operation is an
integer remainder with the same sign as the dividend. If an intermediate or
final result of an arithmetic expression overflows 31 digits, the overflowing
intermediate or final result will be truncated to 31 digits, intentionally or
unintentionally. If an intermediate or final result of an arithmetic expression
requires division or modulus by 0, the intermediate or final result will be set to
0, intentionally or unintentionally.
An arithmetic expression produces a signed, 31-digit zoned decimal (ZD) result
to be edited as specified. If an Mn, EDIT, or EDxy parameter is not specified,
the result is edited using the M0 edit mask.
The default number of digits (d) used for editing is 15 if every term in the
expression is one of the following:
v a 1-4 byte BI or FI field
v a 1-8 byte PD field
v a 1-15 byte ZD, FS, CSF, UFF or SFF field
v
a decimal constant with 1-15 significant digits.
The default number of digits (d) used for editing is 31 if any term in the
v a 16-31 byte ZD field
v a 4-byte or 8-byte FL field
282

v a 16-32 byte FS or CSF field
v a 16-44 byte UFF or SFF field
v a decimal constant with 16-31 significant digits.
If EDIT or EDxy is specified, the number of digits in the pattern (I's and T's) is
used.
See edit under p,m,f,edit for further details on the edit fields you can use.
Sample Syntax:
OUTFIL OUTREC=(5:C% REDUCTION FOR ,21,8,C IS ,
((11,6,ZD,SUB,31,6,ZD),MUL,+1000),DIV,11,6,ZD,
EDIT=(SIIT.T),SIGNS=(+,-))
specifies that the converted result of an arithmetic expression is to appear in
the reformatted OUTFIL output record. See arexp under arexp,edit for further
details on arithmetic expressions.
An arithmetic expression produces a signed, 31-digit zoned decimal (ZD) result
to be converted as specified.
The default number of digits (d) used for conversion is 15 if every term in the
v
v
v
v
a
a
a
a
1-4 byte BI or FI field

1-8 byte PD field
1-15 byte ZD, FS, CSF, UFF or SFF field
decimal constant with 1-15 significant digits.
The default number of digits (d) used for conversion is 31 if any term in the
v
v
v
v
v
a
a
a
a
a
16-31 byte ZD field

4-byte or 8-byte FL field
16-32 byte FS or CSF field
16-44 byte UFF or SFF field
decimal constant with 16-31 significant digits.
See to under p,m,f,to for further details on the to fields you can use.
Sample Syntax:
OUTFIL FNAMES=OUT,
OUTREC=(61,3,X,
35,6,FS,ADD,45,6,FS,ADD,55,6,FS,TO=FS,LENGTH=7,X,
(5,11,PD,MIN,112,11,PD),PD,LENGTH=11,X,
64,5,SEQNUM,5,ZD)
p,m,Y2x or p,m,Y4x
using the century window established by the Y2PAST option in effect. Y2x and
Y4x dates with special indicators are expanded appropriately (for example,
p,6,Y2T transforms C'000000' to C'00000000').
Editing involving an input date with an invalid digit (A-F) can result in a data
exception (0C7 ABEND) or an incorrect output value. Editing involving an
invalid input date can result in an invalid output value.
p
See p under p,m,a.

283

Specifies the length in bytes of the Y2x (two-digit year) or Y4x (four-digit
year) date field.
Y2x or Y4x
Specifies the Y2 or Y4 format for the date field. See Appendix C, Data
format descriptions, on page 891 for detailed format descriptions.
Sample Syntax:
OUTFIL BUILD=(21,3,Y2U,X,3,8,Y4W)
Table 51 shows the output produced for each type of edited date.
Table 51. Input and result fields for Yxx date editing
m,Yxx
Input date
Output for p,m,Yxx
3,Y2T
4,Y2T
5,Y2T
6,Y2T
7,Y4T
8,Y4T
2,Y2U
3,Y2V
3,Y2U
4,Y2V
4,Y4U
5,Y4V
3,Y2W
4,Y2W
5,Y2W
6,Y2W
7,Y4W
8,Y4W
2,Y2X
3,Y2Y
3,Y2X
4,Y2Y
4,Y4X
5,Y4Y
2,Y2C
2,Y2Z
2,Y2S
2,Y2P
1,Y2D
1,Y2B
C'yyx' or Z'yyx'
C'yyxx' or Z'yyxx'
C'yyddd' or Z'yyddd'
C'yymmdd' or Z'yymmdd'
C'ccyyddd' or Z'ccyyddd'
C'ccyymmdd' or Z'ccyymmdd'
P'yyx'
P'yyxx'
P'yyddd'
P'yymmdd'
P'ccyyddd'
P'ccyymmdd'
C'xyy' or Z'xyy'
C'xxyy' or Z'xxyy'
C'dddyy' or Z'dddyy'
C'mmddyy' or Z'mmddyy'
C'dddccyy' or Z'dddccyy'
C'mmddccyy' or Z'mmddccyy'
P'xyy'
P'xxyy'
P'dddyy'
P'mmddyy'
P'dddccyy'
P'mmddccyy'
C'yy'
Z'yy'
C'yy'
P'yy'
X'yy'
B'yy'
C'ccyyx'
C'ccyyxx'
C'ccyyddd'
C'ccyymmdd'
C'ccyyddd'
C'ccyymmdd'
C'ccyyx'
C'ccyyxx'
C'ccyyddd'
C'ccyymmdd'
C'ccyyddd'
C'ccyymmdd'
C'xccyy'
C'xxccyy'
C'dddccyy'
C'mmddccyy'
C'dddccyy'
C'mmddccyy'
C'xccyy'
C'xxccyy'
C'dddccyy'
C'mmddccyy'
C'dddccyy'
C'mmddccyy'
C'ccyy'
C'ccyy'
C'ccyy'
C'ccyy'
C'ccyy'
C'ccyy'
%nn,Y2x or %nn,Y4x
Specifies that a parsed input date field is to be edited. See PARSE for details of
parsed fields. See "p,m,Y2x or p,m,Y4x" for further details.
Specifies that the output for a p,m,Yxx input date field as shown in Table 51 is
to be further edited according to the edit parameters you specify. For example,
if you specify:
OUTFIL BUILD=(28,5,Y4V,EDIT=(TTTT-TT-TT)

284

nn,Y2x,edit or %nn,Y4x,edit
Specifies that the output for a parsed Yxx input date field as shown in Table 51
on page 284 is to be further edited according to the edit parameters you
specify. See PARSE for details of parsed fields. See "p,m,Y2x,edit or
p,m,Y4x,edit" for further details.
Specifies that an input date field as shown in Table 51 on page 284 is to be
converted according to the to parameters you specify. For example, if you
specify:
OUTFIL BUILD=(5,6,Y2W,TO=PD,LENGTH=5)
The C'mmddyy' date value will be transformed to a Z'mmddccyy' date value

and then converted to a P'mmddccyy' (X'0mmddccyyC') date value.
Specifies that a parsed input date field as shown in Table 51 on page 284 is to
be converted according to the to parameters you specify. See PARSE for details
of parsed fields. See "p,m,Y2x,to or p,m,Y4x,to" for further details.
|
p,m
,Y2x,
,Y4x,
TOJUL=Yaa
TOJUL=Yaa(s)
TOGREG=Yaa
TOGREG=Yaa(s)
WEEKDAY=CHAR3
WEEKDAY=CHAR9
WEEKDAY=DIGIT1
DT=(abcd)
DTNS=(abc)
WEEKNUM=USA
WEEKNUM=ISO
AGE=YMD
AGE=YM
AGE=YD

corresponding output date field of another type. You can convert date fields
between combinations of 2-digit and 4-digit year dates, CH/ZD and PD dates,
and Julian and Gregorian dates. You can also convert a date field to a
corresponding day of the week in several forms.
Each type of date field you can use as input for date conversions is shown in
Table 52 on page 286.
285

Table 52. Input fields for p,m,Yxx date conversion
m,Yxx
Input date
5,Y2T
6,Y2T
7,Y4T
8,Y4T
3,Y2U
4,Y2V
4,Y4U
5,Y4V
5,Y2W
6,Y2W
7,Y4W
8,Y4W
3,Y2X
4,Y2Y
4,Y4X
5,Y4Y
P'yyddd'
P'yymmdd'
P'ccyyddd'
P'ccyymmdd'
P'dddyy'
P'mmddyy'
P'dddccyy'
P'mmddccyy'
See p under p,m,a.
Specifies the length in bytes of the Y2x (two-digit year) or Y4x

(four-digit year) date field.
Y2x or Y4x
Specifies the Y2 or Y4 format for the date field. See Appendix C, Data
todate Specifies the type of date conversion to be performed.
p,m,Yxx,TOJUL=Yaa
Converts the input date to a Julian output date.
p,m,Yxx,TOJUL=Yaa(s)
Converts the input date to a Julian output date with s
separators. s can be any character except a blank.
p,m,Yxx,TOGREG=Yaa
Converts the input date to a Gregorian output date.
p,m,Yxx,TOGREG=Yaa(s)
Converts the input date to a Gregorian output date with s
separators. s can be any character except a blank.
The output date field created by each valid TOJUL and TOGREG
combination is shown in Table 53 on page 287.
286

Table 53. TOJUL and TOGREG output date fields
Yaa
TOJUL=Yaa
TOJUL=Yaa(s)
TOGREG=Yaa
TOGREG=Yaa(s)
Y2T
Y2W
Y2U
Y2V
Y2X
Y2Y
Y4T
Y4W
Y4U
Y4V
Y4X
Y4Y
C'yyddd'
C'dddyy'
P'yyddd'
n/a
P'dddyy'
n/a
C'ccyyddd'
C'dddccyy'
P'ccyyddd'
n/a
P'dddccyy'
n/a
C'yysddd'
C'dddsyy'
n/a
n/a
n/a
n/a
C'ccyysddd'
C'dddsccyy'
n/a
n/a
n/a
n/a
C'yymmdd'
C'mmddyy'
n/a
P'yymmdd'
n/a
P'mmddyy'
C'ccyymmdd'
C'mmddccyy'
n/a
P'ccyymmdd'
n/a
P'mmddccyy'
C'yysmmsdd'
C'mmsddsyy'
n/a
n/a
n/a
n/a
C'ccyysmmsdd'
C'mmsddsccyy'
n/a
n/a
n/a
n/a
Sample Syntax:
OUTFIL BUILD=(21,3,Y2X,TOGREG=Y4T(/),X,
p,m,Yxx,WEEKDAY={CHAR3|CHAR9|DIGIT1}
Converts an input date field to a corresponding output day of
the week in one of several forms.
Each type of input date field you can use is shown in Table 52
on page 286. The different types of output you can display are
shown in Table 54.
Table 54. Output for weekdays
Day
CHAR3
CHAR9
DIGIT1
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
CSUN
CMON
CTUE
CWED
CTHU
CFRI
CSAT
CSUNDAY
CMONDAY
CTUESDAY
CWEDNESDAY
CTHURSDAY
CFRIDAY
CSATURDAY
C1
C2
C3
C4
C5
C6
C7
Sample Syntax:
* Convert a Pmmddccyy input date to a 3-character weekday
OUTFIL BUILD=(5:15,5,Y4Y,WEEKDAY=CHAR3,
* Convert a Cyyddd input date to a 1-digit weekday
18:27,5,Y2T,WEEKDAY=DIGIT1,
* Convert a Pdddccyy input date to a 9-character weekday
41:121,4,Y4X,WEEKDAY=CHAR9)
p,m,Yxx,DT=(abcd) or p,m,Yxx,DTNS=(abc)
Converts an input date field of one type to a corresponding
Gregorian output date field of another type.
on page 286.
DT=(abcd) creates an output date in the form C'adbdc', where
a, b, and c indicate the order in which the month, day, and
year are to appear and whether the year is to appear as two or
four digits, and d is the character to be used to separate the
287

month, day and year. For a, b, and c, use M to represent the
month (01-12), D to represent the day (01-31), Y to represent
the last two digits of the year (for example, 09), or 4 to
represent the four digits of the year (for example, 2009). M, D,
and Y or 4 can each be specified only once.
DTNS=(abc) creates an output date in the form C'abc', where
a, b and c indicate the order in which the month, day, and year
are to appear and whether the year is to appear as two or four
digits. For a, b and c, use M to represent the month (01-12), D
to represent the day (01-31), Y to represent the last two digits
of the year (for example, 09), or 4 to represent the four digits of
the year (for example, 2009). M, D, and Y or 4 can each be
specified only once.
Sample Syntax:
* Convert a Cyyddd input date to a Cdd/mm/ccyy output date
OUTFIL BUILD=(92,5,Y2T,DT=(DM4/),X,
* Convert a Pccyyddd input date to a Cmmddyy output date
53:32,4,Y4U,DTNS=(MDY))
p,m,Yxx,WEEKNUM={USA|ISO}
Converts an input date field to a corresponding 2 byte output
number of the week in 2 different forms.
|
|
|
|
|
|
USA: The WEEKNUM=USA function returns an integer in the

range of 1 to 54 that represents the week of the year. The week
starts with Sunday, and January 1 is always in the first week.
|
|
|
|
|
|
|
|
ISO: The WEEKNUM=ISO function returns an integer in the

range of 1 to 53 that represents the week of the year. The week
starts with Monday and includes 7 days. Week 1 is the first
week of the year to contain a Thursday, which is equivalent to
the first week containing January 4. Thus, it is possible to have
up to 3 days at the beginning of the year appear as the last
week of the previous year, or to have up to 3 days at the end
of a year appear as the first week of the next year.
|
|
|
on page 286. The different types of output you can display are
shown in Table 55.
Table 55. Output for weeknum
Input - Date
WEEKNUM=USA
WEEKNUM=ISO
2000-01-01
01
52
2000-01-02
02
52
2000-01-03
02
01
2000-12-31
54
52
2014-01-02
01
01
2014-06-09
24
24
|
|
2014-12-31
53
01
Sample Syntax:
288

|
|
|
|
|
|
* Convert a Pmmddccyy input date to a 2-character weeknum in

* ISO format
OUTFIL BUILD=(5:15,5,Y4Y,WEEKNUM=ISO,
* Convert a Cyyddd input date to a 2-digit weeknum in
* USA format
18:27,5,Y2T,WEEKNUM=USA)
|
|
|
|
|
p,m,Yxx,AGE={YMD|YM|YD}
Can be used to calculate the date duration that specifies the
number of years, months, and days between a given date and
the current date. The Input date cannot be greater than the
current date.
|
|
Age=YMD produces a 8 byte result which has duration in

years (0-9999), months (00-12), and days (00-31).
|
|
Age=YM produces a 6 byte result which has duration in years

(0-9999), months (00-12).
|
|
Age=YD produces a 7 byte result which has duration in years

(0-9999), days (00-366).
|
|
|
|
For example, if the input birth date is March 12th 2015 and the
current date is September 30th 2015 then Age=YMD will
produce a duration of 00000618 (0000 years, 06 months, and 18
days).
|
|
|
|
|
|
|
|
|
|
Sample Syntax: If the input birth date is March 12th 2015 and
the current date is September 30th 2015 then Age=YMD will
produce a duration of 00000618 (0000 years, 06 months, and 18
days).
* Calculate the date duration of a Cccyyddd input date to
* current date as years, months and days
OUTFIL BUILD=(15:1,7,Y4T,AGE=YMD,
* Calculate the date duration of a Pyymmdd input date to
* current date as years and days
55:27,4,Y2V,AGE=YD)
Conversion of Real Dates, Special Indicators and Invalid Dates

For CH/ZD dates (Y2T, Y4T, Y2W, Y4W), the zone and sign are
ignored; only the digits are used. For example,
X'F2F0F0F9F1F2F3D0' and X'A2B0C0D9E1F21320' are treated as
20091230. For PD dates (Y2U, Y4U, Y2V, Y4V, Y2X, Y4X, Y2Y,
Y4Y), the sign is ignored; only the relevant digits are used. For
example, X'120091230D' is treated as 20091230.
For CH/ZD dates (Y2T, Y4T, Y2W, Y4W), the special indicators
are X'00...00' (BI zeros), X'40...40' (blanks), C'0...0' (CH zeros),
Z'0...0' (ZD zeros), C'9...9' (CH nines), Z'9...9' (ZD nines) and
X'FF...FF' (BI ones). For PD dates (Y2U, Y4U, Y2V, Y4V, Y2X,
Y4X, Y2Y, Y4Y), the special indicators are P'0...0' (PD zeros) and
P'9...9' (PD nines).
yy for real 2-digit year dates is transformed to ccyy when
appropriate using the century window established by the
Y2PAST option in effect. ccyy for real 4-digit year dates is
transformed to yy when appropriate by removing cc.
Date conversion is not performed for special indicators; the
special indicator is just used appropriately for the output date
field. For example, if p,5,Y2T,TOGREG=Y4T(/) is used, an
input yyddd special indicator of C'99999' results in an output
289

date field of C'9999/99/99'. However, CH/ZD special
indicators of BI zeros, blanks and BI ones cannot be converted
to PD special indicators.
Conversion involving an input date with an invalid digit (A-F)
will result in a data exception (0C7 ABEND) or an incorrect
output value.
Conversion involving an invalid input date or invalid output
date will result in an output value of asterisks and an
informational message (issued once). A date is considered
invalid if any of the following range conditions are not met:
v yy must be between 00 and 99
v ccyy must be between 0001 and 9999
v mm must be between 01 and 12
v dd must be between 01 and 31, and must be valid for the
year and month
v ddd must be 001 to 366 for a leap year, or between 001 and
365 for a non-leap year
A date is also considered invalid if the input field is a CH/ZD
special indicator of binary zeros, blanks or binary ones, and the
output field is PD.
Specifies that an input date field of one type is to be converted
to a corresponding output date field of another type. See
PARSE for details of parsed fields. See "p,m,Y2x,todate or
p,m,Y4x,todate" for further details.
Specifies an arithmetic operation for an input date field. The available
operations are as follows:
v ADDDAYS, ADDMONS or ADDYEARS can be used to add days, months
or years to a date field. As a simple example, you could use the following to
add 10 days to a C'ccyymmdd' date and produce the resulting
C'ccyy/mm/dd' date:
15,8,Y4T,ADDDAYS,+10,TOGREG=Y4T(/)
v SUBDAYS, SUBMONS or SUBYEARS can be used to subtract days,

months or years from a date field. As a simple example, you could use the
following to subtract 3 months from a P'dddyy' date and produce the
resulting P'ccyyddd' date:
21,3,Y2X,SUBMONS,+3,TOJUL=Y4U
v DATEDIFF can be used to calculate the number of days between two date
fields. As a simple example, you could use the following to calculate the
difference in days between two C'ccyymmdd' dates:
41,8,Y4T,DATEDIFF,31,8,Y4T
v NEXTDday can be used to calculate the next specified day for a date field.
As a simple example, you could use the following to calculate the next
Friday for a C'ccyyddd' date as a C'ccyy.ddd' date:
3,7,Y4T,NEXTDFRI,TOJUL=Y4T(.)
v PREVDday can be used to calculate the previous specified day for a date
field. As a simple example, you could use the following to calculate the
previous Wednesday for a P'yyddd' date as a C'ccyymmdd' date:
51,3,Y2U,PREVDWED,TOGREG=Y4T
290

v LASTDAYW, LASTDAYM, LASTDAYQ or LASTDAYY can be used to
calculate the last day of the week, month, quarter or year for a date field. As
a simple example, you could use the following to calculate the last day of
the month for a C'mmddccyy' date as a C'mmddccyy' date:
28,8,Y4W,LASTDAYM,TOGREG=Y4W
Specifies an arithmetic operation for a parsed input date field.
Adding or Subtracting Days, Months or Years
p,m
%nn
,Y2x
,Y4x
ADDDAYS
ADDMONS
ADDYEARS
SUBDAYS
SUBMONS
SUBYEARS
p,m,f
deccon
TOJUL=Yaa
TOJUL=Yaa(s)
TOGREG=Yaa
TOGREG=Yaa(s)
p,m,Yxx,keyword,numeric,todate
Can be used to add n days, months or years to an input date field, or subtract
n days, months or years from an input date field. The valid keywords are
ADDDAYS, ADDMONS, ADDYEARS, SUBDAYS, SUBMONS and SUBYEARS.
The numeric value specifies the number of days, months or years to be added
to or subtracted from the input date field. The resulting output date field is
converted to the form indicated by todate.
p,m,Yxx specifies the starting position (p), length (m) and format (Yxx) of a
2-digit or 4-digit year input date field. The valid length and format for each
type of date field you can use is shown in Table 56.
Table 56. p,m,Yxx fields for date arithmetic
m,Yxx
Type of date
5,Y2T
6,Y2T
7,Y4T
8,Y4T
5,Y2W
6,Y2W
7,Y4W
8,Y4W
3,Y2U
4,Y2V
4,Y4U
5,Y4V
3,Y2X
4,Y2Y
4,Y4X
5,Y4Y
P'yyddd'
P'yymmdd'
P'ccyyddd'
P'ccyymmdd'
P'dddyy'
P'mmddyy'
P'dddccyy'
P'mmddccyy'
ADDDAYS adds n days to the p,m,Yxx date field.

ADDMONS adds n months to the p,m,Yxx date field. If necessary, the
resulting date is adjusted backwards to the last valid day of the month. For
example, if 1 month is added to 20101031, the resulting date will be 20101130
rather than 20101131.
291

ADDYEARS adds n years to the p,m,Yxx date field. If the resulting date is
February 29 of a non-leap year, it will be adjusted backwards to a valid date of
February 28.
SUBDAYS subtracts n days from the p,m,Yxx date field.
SUBMONS subtracts n months from the p,m,Yxx date field. If necessary, the
resulting date is adjusted backwards to the last valid day of the month. For
example, if 1 month is subtracted from 20101031, the resulting date will be
20100930 rather than 20100931.
SUBYEARS subtracts n years from the p,m,Yxx date field. If the resulting date
is February 29 of a non-leap year, it will be adjusted backwards to a valid date
of February 28.
numeric can be a decimal constant (+n or -n), or a valid numeric field (p,m,f)
with BI, FI, ZD, PD, FS, UFF or SFF format and an appropriately
corresponding length.
For ADDDAYS or SUBDAYS, the numeric constant or field value can be from
-3652058 to +3652058.
For ADDMONS or SUBMONS, the numeric constant or field value can be from
-119987 to +119987.
For ADDYEARS or SUBYEARS, the numeric constant or field value can be
from -9998 to +9998.
TOJUL and TOGREG
See TOJUL and TOGREG under "p,m,Y2x,todate or p,m,Y4x,todate" for
details.
%nn,Yxx,keyword,numeric,todate
Can be used to add n days, months or years to a parsed input date field, or
subtract n days, months or years from a parsed input date field. See PARSE for
details of parsed fields. See "p,m,Yxx,keyword,numeric,todate" for further
details.
%nn cannot be used for the numeric field.
Calculating Difference in Days
p,m
%nn
,Y2x
,Y4x
DATEDIFF p,m
,Y2x
,Y4x
p,m,Yxx,DATEDIFF,p,m,Yxx
Can be used to calculate the number of days difference between two input
date fields. The result is an 8-byte value consisting of a sign and 7 digits
(sddddddd). If the first date is greater than or equal to the second date, the
sign is + (plus). If the first date is less than the second date, the sign is (minus).
type of date field you can use is shown in Table 56 on page 291.
DATEDIFF calculates the number of days that result when the second date is
subtracted from the first date. The result is in the form sddddddd with + or for s as appropriate and leading zeros. The result can be between -3652058 and
+3652058 days.
292

%nn,Yxx,DATEDIFF,p,m,Yxx
Can be used to calculate the number of days difference between a parsed input
date field and another input date field. See PARSE for details of parsed fields.
See "p,m,Yxx,DATEDIFF,p,m,Yxx" for further details.
%nn cannot be used for the second date field.
Next, Previous or Last Day
p,m
%nn
,Y2x
,Y4x
NEXTDday
PREVDday
LASTDAYx
TOJUL=Yaa
TOJUL=Yaa(s)
TOGREG=Yaa
TOGREG=Yaa(s)
p,m,Yxx,keyword,todate
Can be used to specify the next specified day, previous specified day, or last
day of the week, month, quarter or year for an input date. The valid keywords
are NEXTDday, PREVDday or LASTDAYx (where day can be SUN, MON,
TUE, WED, THU, FRI or SAT and x can be W, M, Q or Y).
The result is an output date field which is converted to the form indicated by
todate.
type of date field you can use is shown in Table 56 on page 291.
NEXTDSUN calculates the next Sunday for a date field.
NEXTDMON calculates the next Monday for a date field.
NEXTDTUE calculates the next Tuesday for a date field.
NEXTDWED calculates the next Wednesday for a date field.
NEXTDTHU calculates the next Thursday for a date field.
NEXTDFRI calculates the next Friday for a date field.
NEXTDSAT calculates the next Saturday for a date field.
PREVDSUN calculates the previous Sunday for a date field.
PREVDMON calculates the previous Monday for a date field.
PREVDTUE calculates the previous Tuesday for a date field.
PREVDWED calculates the previous Wednesday for a date field
PREVDTHU calculates the previous Thursday for a date field.
PREVDFRI calculates the previous Friday for a date field.
PREVDSAT calculates the previous Saturday for a date
LASTDAYW calculates the last day of the week (Friday) for a date field.
LASTDAYM calculates the last day of the month for a date field.
LASTDAYQ calculates the last day of the quarter for a date field.
LASTDAYY calculates the last day of the year for a date field.
TOJUL and TOGREG
See TOJUL and TOGREG under "p,m,Y2x,todate or p,m,Y4x,todate" for
details.
293

%nn,Yxx,keyword,todate
Can be used to specify the next specified day, previous specified day, or last
day of the week, month, quarter or year for a parsed input date. See PARSE
for details of parsed fields. See p,m,Yxx,keyword,todate for further details.
Arithmetic with Real Dates, Special Indicators and Invalid Dates
For CH/ZD dates (Y2T, Y4T, Y2W, Y4W), the zone and sign are ignored; only
the digits are used. For example, X'F2F0F0F9F1F2F3D0' and
X'A2B0C0D9E1F21320' are treated as 20091230. For PD dates (Y2U, Y4U, Y2V,
Y4V, Y2X, Y4X, Y2Y, Y4Y), the sign is ignored; only the relevant digits are used.
For example, X'120091230D' is treated as 20091230.
For CH/ZD dates (Y2T, Y4T, Y2W, Y4W), the special indicators are X'00...00' (BI
zeros), X'40...40' (blanks), C'0...0' (CH zeros), Z'0...0' (ZD zeros), C'9...9' (CH
nines), Z'9...9' (ZD nines) and X'FF...FF' (BI ones). For PD dates (Y2U, Y4U,
Y2V, Y4V, Y2X, Y4X, Y2Y, Y4Y), the special indicators are P'0...0' (PD zeros) and
P'9...9' (PD nines).
yy for real 2-digit year dates is transformed to ccyy when appropriate using
the century window established by the Y2PAST option in effect. ccyy for real
4-digit year dates is transformed to yy when appropriate by removing cc.
If a special indicator is used for DATEDIFF, the result will be an output value
of asterisks and an informational message (issued once).
Date arithmetic is not performed for special indicators; the special indicator is
just used appropriately for the output date field. For example, if
p,5,Y2T,ADDDAYS,+10,TOGREG=Y4T(/) is used, an input yyddd special
indicator of C'99999' results in an output date field of C'9999/99/99'. However,
CH/ZD special indicators of BI zeros, blanks and BI ones cannot be converted
to PD special indicators.
Arithmetic involving an input date with an invalid digit (A-F) will result in a
data exception (0C7 ABEND) or an incorrect output value.
Arithmetic involving an invalid input date or invalid output date will result in
an output value of asterisks and an informational message (issued once). A
date is considered invalid if any of the following range conditions are not met:
v yy must be between 00 and 99
v ccyy must be between 0001 and 9999
v mm must be between 01 and 12
v dd must be between 01 and 31, and must be valid for the year and month
v ddd must be between 001 and 366 for a leap year, or between 001 and 365
for a non-leap year.
A date is also considered invalid if the input field is a CH/ZD special indicator
of binary zeros, blanks or binary ones, and the output field is PD.
Specifies that an input date field is to be edited with separators. s can be any
character except a blank. Real Y2x dates are edited using the century window
indicators are expanded appropriately (for example, p,8,Y4T(s) transforms
C'00000000' to C'0000/00/00').
p
294
See p under p,m,a.

m

Y2x(s) or Y4x(s)
Specifies the Y2 or Y4 format for the date field and the separator
character(s). See Appendix C, Data format descriptions, on page 891
for detailed format descriptions.
Sample Syntax:
OUTFIL BUILD=(19,7,Y4W(/),X,
43,5,Y4V(-))
Table 57 shows the output produced for each type of edited date.
Table 57. Input and result fields for Yxx(s) date editing
m,Yxx
Input date
Output for p,m,Yxx(s)
3,Y2T(s)
4,Y2T(s)
5,Y2T(s)
6,Y2T(s)
7,Y4T(s)
8,Y4T(s)
2,Y2U(s)
3,Y2V(s)
3,Y2U(s)
4,Y2V(s)
4,Y4U(s)
5,Y4V(s)
3,Y2W(s)
4,Y2W(s)
5,Y2W(s)
6,Y2W(s)
7,Y4W(s)
8,Y4W(s)
2,Y2X(s)
3,Y2Y(s)
3,Y2X(s)
4,Y2Y(s)
4,Y4X(s)
5,Y4Y(s)
C'yyx' or Z'yyx'
C'yyxx' or Z'yyxx'
P'yyx'
P'yyxx'
P'yyddd'
P'yymmdd'
P'ccyyddd'
P'ccyymmdd'
C'xyy' or Z'xyy'
C'xxyy' or Z'xxyy'
P'xyy'
P'xxyy'
P'dddyy'
P'mmddyy'
P'dddccyy'
P'mmddccyy'
C'ccyysx'
C'ccyysxx'
C'ccyysddd'
C'ccyysmmsdd'
C'ccyysddd'
C'ccyysmmsdd'
C'ccyysx'
C'ccyysxx'
C'ccyysddd'
C'ccyysmmsdd'
C'ccyysddd'
C'ccyysmmsdd'
C'xsccyy'
C'xxsccyy'
C'dddsccyy'
C'mmsddsccyy'
C'dddsccyy'
C'mmsddsccyy'
C'xsccyy'
C'xxsccyy'
C'dddsccyy'
C'mmsddsccyy'
C'dddsccyy'
C'mmsddsccyy'
PARSE for details of parsed fields. See "p,m,Y2x(s) or p,m,Y4x(s)" for further
details.
p,m,Y2xP
Specifies that an input date field is to be converted to a packed decimal output
date field. Real Y2x dates are edited using the century window established by
the Y2PAST option in effect. Y2x and Y4x dates with special indicators are
expanded appropriately (for example, p,6,Y2TP transforms C'000000' to
P'00000000').
p
See p under p,m,a.

295

m

Y2xP
Specifies the Y2 format for the date field. See Appendix C, Data
Sample Syntax:
OUTFIL BUILD=(11,3,Y2XP,X,21,4,Y2WP)
Table 58 shows the output produced for each type of date.

Table 58. Input and result fields for Yxx(s) date editing
m,Y2xP
Input date
Output for p,m,Y2xP
3,Y2TP
4,Y2TP
5,Y2TP
6,Y2TP
2,Y2UP
3,Y2VP
3,Y2UP
4,Y2VP
3,Y2WP
4,Y2WP
5,Y2WP
6,Y2WP
2,Y2XP
3,Y2YP
3,Y2XP
4,Y2YP
2,Y2PP
1,Y2DP
C'yyx' or Z'yyx'
C'yyxx' or Z'yyxx'
P'yyx'
P'yyxx'
P'yyddd'
P'yymmdd'
C'xyy' or Z'xyy'
C'xxyy' or Z'xxyy'
P'xyy'
P'xxyy'
P'dddyy'
P'mmddyy'
P'yy'
X'yy'
P'ccyyx'
P'ccyyxx'
P'ccyyddd'
P'ccyymmdd'
P'ccyyx'
P'ccyyxx'
C'ccyyddd'
C'ccyymmdd'
C'xccyy'
C'xxccyy'
C'dddccyy'
C'mmddccyy'
C'xccyy'
C'xxccyy'
C'dddccyy'
C'mmddccyy'
C'ccyy'
C'ccyy'
%nn,Y2xP
Specifies that a parsed input date field is to be converted to a packed decimal
output date field. See PARSE for details of parsed fields. See "p,m,Y2xP" for
further details
specifies that a character constant, hexadecimal constant, input field (p,m), or
parsed input field (%nn) from a lookup table is to appear in the reformatted
OUTFIL output record. You can use p,m,lookup or %nn,lookup to select a
specified set constant (that is, a character or hexadecimal string) or set field
(that is, an input field or parsed input field) based on matching an input value
against find constants (that is, character, hexadecimal, or bit constants).
p
See p under p,m,a.
specifies the length in bytes of the input field to be compared to the find
constants. The value for m must be 1 to 64 if character or hexadecimal find
constants are used, or 1 if bit find constants are used.
%nn
specifies a parsed input field to be compared to the find constants. See
PARSE for details of parsed fields.
lookup
Specifies how the input field or parsed input field is to be changed to the
output field, using a lookup table.
CHANGE
296

.
CHANGE=(-v ,find,set

,NOMATCH=(set)
specifies a list of change pairs, each consisting of a find constant to be

compared to the input field value or parsed input field value and a set
constant or set field to use as the output field when a match occurs.
v
specifies the length in bytes of the output field to be inserted in the

reformatted OUTFIL output record. The value for v must be between 1 and
64.
find
specifies a find constant to be compared to the input field value or parsed
input field value. If the input field value or parsed input field value
matches the find constant, the corresponding set constant or set field is
used for the output field.
The find constants can be either character string constants, hexadecimal
string constants, or bit constants:
v Character string constants (C'xx...x') and hexadecimal string constants
(X'yy...yy') can be 1 to m bytes and can be intermixed with each other,
but not with bit constants. See INCLUDE control statement on page 96
for details of coding character and hexadecimal string constants.
If the string is less than m bytes, it will be padded on the right to a
length of m bytes, using blanks (X'40') for a character string constant or
zeros (X'00') for a hexadecimal string constant.
v Bit constants (B'bbbbbbbb') must be 1 byte and cannot be intermixed
with character or string constants. See INCLUDE control statement on
page 96 for details of coding bit constants.
For bit constants, because of the specification of bits to be ignored, more
than one find constant can match an input field value; the set constant
for the first match found will be used as the output field. For example, if
you specify:
OUTFIL OUTREC=(5,1,
CHANGE=(1,B11......,CA,B1.......,CB))
input field value X'C0' (B'11000000') matches both bit constants, but C'A'
will be used for the set constant, because its find constant is the first
match.
set
specifies a set constant or set field to be used as the output field if the
corresponding find constant matches the input field value or parsed input
field value. Set constants and set fields can be intermixed.
Set constants can be either character string constants (C'xx...x') or
hexadecimal string constants (X'yy...yy') of 1 to v bytes. See INCLUDE
If the string is less than v bytes, it will be padded on the right to a length
of v bytes, using blanks (X'40') for a character string constant or zeros
(X'00') for a hexadecimal string constant.
Set fields are specified as q,n or %nn. q specifies the input position. See p
under p,m,a for details of coding q. n specifies the input length of 1 to v
297

bytes. %nn specifies a parsed input field. See PARSE for details of parsed
fields. If the set field length is less than v, the input field or parsed input
field will be padded on the right to a length of v bytes, using blanks
(X'40').
NOMATCH
specifies the action to be taken if an input field value or parsed input field
value does not match any of the find constants. If you do not specify
NOMATCH, and a match is not found for any input value, DFSORT
terminates processing.
If you specify NOMATCH, it must follow CHANGE.
set
specifies a set constant or set field to be used as the output field if a
match is not found.
Set constants can be either character string constants (C'xx...x') or
hexadecimal string constants (X'yy...yy') of 1 to v bytes and can be
intermixed. See INCLUDE control statement on page 96 for details of
coding character and hexadecimal string constants.
If the string is less than v bytes, it will be padded on the right to a
length of v bytes, using blanks (X'40') for a character string constant or
zeros (X'00') for a hexadecimal string constant.
Set fields are specified as q,n or %nn. q specifies the input position. See
p under p,m,a for details of coding q. n specifies the input length of 1
to v bytes. %nn specifies a parsed input field. See PARSE for details of
parsed fields. If the set field length is less than v, the input field or
parsed input field will be padded on the right to a length of v bytes,
using blanks (X'40').
Sample Syntax:
OUTFIL FILES=1,
OUTREC=(11,1,
CHANGE=(6,
CR,CREAD,
CU,CUPDATE,
XFF,CEMPTY,
CX,35,6,
CA,CALTER),
NOMATCH=(11,6),
4X,
21,1,
CHANGE=(10,
B.1......,CVSAM,
p,m,justify
reformatted OUTFIL output record. For a left-justified field, leading blanks are
removed and the characters from the first nonblank to the last nonblank are
shifted left, with blanks inserted on the right if needed. For a right-justified
field, trailing blanks are removed and the characters from the last nonblank to
the first nonblank are shifted right, with blanks inserted on the left if needed.
Optionally:
298

v the output length can be changed (it's equal to the input length by default)
p
See p under p,m,a.
See m under p,m,a.
justify
specifies how the input field is to be justified for output.
JFY
,
JFY=
(
SHIFT=LEFT
SHIFT=RIGHT
LENGTH=n
PREBLANK=list
LEAD=string
TRAIL=string
Note: For clarity, in the examples in this section, b is used to represent

a blank in the input fields.
SHIFT=LEFT
specifies that a left-justified input field is to appear in the reformatted
OUTFIL output record. Leading blanks are removed and the characters
from the first nonblank to the last nonblank are shifted left, with
blanks inserted on the right if needed. For example, with:
1,15,JFY=(SHIFT=LEFT)
an input field of:

bbbbABbCDbbEFbb
results in an output field of:

ABbCDbbEFbbbbbb
m is used for the output field length unless LENGTH=n is specified. If

the left-justified input field is shorter than the output field length,
blanks are inserted on the right. If the left-justified input field is longer
than the output field length, characters are truncated on the right. You
can use LENGTH=n to prevent padding or truncation.
SHIFT=RIGHT
specifies that a right-justified input field is to appear in the reformatted
OUTFIL output record. Trailing blanks are removed and the characters
from the last nonblank to the first nonblank are shifted right, with
blanks inserted on the left if needed. For example, with:
1,15,JFY=(SHIFT=RIGHT)
an input field of:

bbbbABbCDbbEFbb

bbbbbbABbCDbbEF

the right-justified input field is shorter than the output field length,
299

blanks are inserted on the left. If the right-justified input field is longer
than the output field length, characters are truncated on the left. You
can use LENGTH=n to prevent padding or truncation.
LENGTH=n
specifies the length of the justified output field. The value for n must
be between 1 and 32752. If LENGTH=n is not specified, m (the input
field length) is used for the length of the justified output field. You can
use LENGTH=n to prevent padding or truncation.
PREBLANK=list
specifies a list of one or more characters to be changed to blanks
before justify processing begins. Only leading and trailing characters
are changed to blanks. Scanning from left to right, each nonblank
character at the start of the input field that matches a character in the
list is replaced by a blank character, until a nonblank character not in
the list is found. Scanning from right to left, each nonblank character at
the end of the input field that matches a character in the list will be
replaced by a blank character, until a nonblank character not in the list
is found.
list can be 1 to 10 characters specified using a character string constant
Each character in the list is treated as one character to be preblanked.
Thus PREBLANK=C'*' specifies an asterisk is to be preblanked, and
PREBLANK=C',;$' specifies a comma, a semicolon and a dollar sign are
each to be preblanked.
For example, let's say we have an input field of:
**b<*ABbCD*bEF>*b**
If we specify:
1,19,JFY=(SHIFT=LEFT,PREBLANK=C*)
each leading or trailing asterisk is changed to a blank before leftjustify processing begins. So the output field is:
<*ABbCD*bEF>bbbbbbb
If we specify:
1,19,JFY=(SHIFT=RIGHT,PREBLANK=C*<>)
each leading or trailing asterisk, less than sign and greater than sign is
changed to a blank before right-justify processing begins. So the output
field is:
bbbbbbbbbbABbCD*bEF
LEAD=string
specifies a string to be inserted in the output field before the first
nonblank character in the input field.
string can be 1 to 50 characters specified using a character string
constant (C'xx...x') or a hexadecimal string constant (X'yy...yy'). See
INCLUDE control statement on page 96 for details of coding
character and hexadecimal string constants.
300

bABCbEbbbbb
If we specify:
1,11,JFY=(SHIFT=RIGHT,LEAD=CXYZ)
the output field is:

bbbXYZABCbE
When we add characters with LEAD=string, it's often necessary to

specify LENGTH=n to avoid truncation. For example, let's say we have
an input field of:
AbBbC
If we specify:
1,5,JFY=(SHIFT=LEFT,LEAD=CXYZ)

XYZAb
Since the output field length is defaulted to the input field length of 5,
the resulting 8 characters (XYZAbBbC) are truncated on the right to 5
characters (XYZAb) for output. If we instead specify:
1,5,JFY=(SHIFT=LEFT,LEAD=CXYZ,LENGTH=8)

XYZAbBbC
LENGTH=8 increases the output field by the 3 LEAD characters and

truncation is prevented.
TRAIL=string
specifies a string to be inserted in the output field after the last
bABCbEbbbbb
If we specify:
1,11,JFY=(SHIFT=LEFT,TRAIL=CXYZ)

ABCbEXYZbbb
When we add characters with TRAIL=string, it's often necessary to

an input field of:
AbBbC
If we specify:
1,5,JFY=(SHIFT=RIGHT,TRAIL=CXYZ)
301

bCXYZ
the resulting 8 characters (AbBbCXYZ) are truncated on the left to 5
characters (bCXYZ) for output. If we instead specify:
1,5,JFY=(SHIFT=RIGHT,TRAIL=CXYZ,LENGTH=8)

AbBbCXYZ
LENGTH=8 increases the output field by the 3 TRAIL characters and

truncation is prevented.
Sample Syntax:
OUTFIL BUILD=(5:16,20,JFY=(SHIFT=LEFT,PREBLANK=C*,
LEAD=C<A>,TRAIL=C</A>,LENGTH=22))
%nn,justify
specifies that a left-justified or right-justified parsed input field is to appear in
the reformatted OUTFIL output record. See PARSE for details of parsed fields.
See p,m,justify for further details.
p,m,squeeze
specifies that a left-squeezed or right-squeezed input field is to appear in the
reformatted OUTFIL output record. For a left-squeezed field, all blanks are
removed and the characters from the first nonblank to the last nonblank are
shifted left, with blanks inserted on the right if needed. For a right-squeezed
field, all blanks are removed and the characters from the last nonblank to the
first nonblank are shifted right, with blanks inserted on the left if needed.
Optionally:
v specific characters can be changed to blanks before squeezing begins
v a string (for example, a comma delimiter) can be inserted wherever a group
of blanks is removed between the first nonblank and the last nonblank
v blanks can be kept as is between paired apostrophes ('AB CD EF') or paired
quotes ("AB CD EF")
v the output length can be changed (it is equal to the input length by default)
p
See p under p,m,a.
See m under p,m,a.
squeeze
Specifies how the input field is to be squeezed for output.
SQZ
302

,
SQZ=
(
SHIFT=LEFT
SHIFT=RIGHT
LENGTH=n
PREBLANK=list
LEAD=string
MID=string
TRAIL=string
PAIR=APOST
PAIR=QUOTE
Note: For clarity, in the examples in this section, b is used to represent

a blank in the input fields.
SHIFT=LEFT
specifies that a left-squeezed input field is to appear in the reformatted
OUTFIL output record. All blanks are removed and the characters from
the first nonblank to the last nonblank are shifted left, with blanks
inserted on the right if needed. For example, with:
1,15,SQZ=(SHIFT=LEFT)
an input field of:

bbbbABbCDbbEFbb

ABCDEFbbbbbbbbb

the left-squeezed input field is shorter than the output field length,
blanks are inserted on the right. If the left-squeezed input field is
longer than the output field length, characters are truncated on the
right.
SHIFT=RIGHT
specifies that a right-squeezed input field is to appear in the
reformatted OUTFIL output record. All blanks are removed and the
characters from the last nonblank to the first nonblank are shifted
right, with blanks inserted on the left if needed. For example, with:
1,15,SQZ=(SHIFT=RIGHT)
an input field of:

bbbbABbCDbbEFbb

bbbbbbbbbABCDEF

the right-squeezed input field is shorter than the output field length,
blanks are inserted on the left. If the right-squeezed input field is
longer than the output field length, characters are truncated on the left.
LENGTH=n
specifies the length of the squeezed output field. The value for n must
be between 1 and 32752. If LENGTH=n is not specified, m (the input
field length) is used for the length of the squeezed output field.
303

PREBLANK=list
specifies a list of one or more characters to be changed to blanks
before squeeze processing begins. Each nonblank character in the input
field that matches a character in the list is replaced by a blank
character.
list can be 1 to 10 characters specified using a character string constant
Each character in the list is treated as one character to be preblanked.
Thus PREBLANK=C'*' specifies an asterisk is to be preblanked, and
PREBLANK=C',;$' specifies a comma, a semicolon and a dollar sign are
each to be preblanked.
**b<*ABbC<>D*bEF>*b**
If we specify:
1,21,SQZ=(SHIFT=LEFT,PREBLANK=C*,LENGTH=12)
each asterisk is changed to a blank before left-squeeze processing

begins. The output field is:
<ABC<>DEF>bb
If we specify:
1,21,SQZ=(SHIFT=RIGHT,PREBLANK=C*<>)
each asterisk, less than sign and greater than sign is changed to a
blank before right-squeeze processing begins. The output field is:
bbbbbbbbbbbbbbbABCDEF
LEAD=string
specifies a string to be inserted in the output field before the first
bABCbEbbbbb
If we specify:
1,11,SQZ=(SHIFT=RIGHT,LEAD=CXYZ)

bbbbXYZABCE
When we add characters with LEAD=string, it's often necessary to

specify LENGTH=n to avoid truncation. For additional information on
this, see "LEAD=string" for JFY shown previously in this section.
304

MID=string
specifies a string to be inserted in the output field wherever one or
more blanks is removed between the first nonblank and the last
nonblank.
bABbbCbbbDbE
If we specify:
1,12,SQZ=(SHIFT=LEFT,MID=C,)

AB,C,D,Ebbbb
When we add characters with MID=string, it's often necessary to

an input field of:
AbBbC
If we specify:
1,5,SQZ=(SHIFT=LEFT,MID=CXY)

AXYBX
the resulting 7 characters (AXYBXYC) are truncated on the right to 5
characters (AXYBX) for output. If we instead specify:
1,5,SQZ=(SHIFT=LEFT,MID=CXY,LENGTH=7)

AXYBXYC
LENGTH=7 increases the output field to accommodate the MID strings

and truncation is prevented.
TRAIL=string
specifies a string to be inserted in the output field after the last
bABCbEbbbbb
If we specify:
1,11,SQZ=(SHIFT=LEFT,LEAD=X7D,TRAIL=X7D)

305

ABCEbbbbb
When we add characters with TRAIL=string, it's often necessary to

specify LENGTH=n to avoid truncation. For additional information on
this, see "TRAIL=string" for JFY as shown previously in this section.
PAIR=APOST
specifies that blanks and PREBLANK characters between apostrophe (')
pairs are to be kept as is. Use PAIR=APOST when you have literals
that should not be squeezed.
b*bABbb**CbbbD*Ebb*
If we specify:
1,25,SQZ=(SHIFT=LEFT,PREBLANK=C*,MID=C,)
all of the blanks and asterisks, including those inside the apostrophe
pairs, are squeezed out. The output field is:
AB,,C,D,E,bbbbbbbbb
However, if we specify:
1,25,SQZ=(SHIFT=LEFT,PREBLANK=C*,MID=C,,
PAIR=APOST)
only the blanks and asterisks outside of the apostrophe pairs are
squeezed out. The output field is:
ABbb**C,D*Ebbbbbbbb
If an apostrophe is specified in the PREBLANK list (for example,

PREBLANK=C'''' or PREBLANK=X'7D'), it is ignored, that is, an
apostrophe in the input field is not replaced by a blank. So if you want
each apostrophe to be replaced by a blank, do not specify
PAIR=APOST.
For SHIFT=LEFT, scanning is left to right. If an apostrophe is found
with no subsequent apostrophe, all of the characters from the
apostrophe to the end of the input field are ignored. For example, let's
say we have an input field of:
bbXbYABCbbbDbb
If we specify:
1,15,SQZ=(SHIFT=LEFT,PAIR=APOST)
there's an unpaired apostrophe, so all of the characters from the

apostrophe to the end of the input field are ignored. The output field
is:
XYABCbbbDbbbbb
For SHIFT=RIGHT, scanning is right to left. If an apostrophe is found

with no previous apostrophe, all of the characters from the apostrophe
to the beginning of the input field are ignored. For example, let's say
we have an input field of:
bbXbYABCbbbDbb
If we specify:
306

1,15,SQZ=(SHIFT=RIGHT,PAIR=APOST)
there's an unpaired apostrophe, so all of the characters from the

apostrophe to the beginning of the input field are ignored. The output
field is:
bbbbbbbXbYABCD
PAIR=QUOTE
specifies that blanks and PREBLANK characters between quote (")
pairs are to be kept as is. Use PAIR=QUOTE when you have literals
that should not be squeezed.
b*b"ABbb*""*C"bbb"D*Ebb"*
If we specify:
1,25,SQZ=(SHIFT=LEFT,PREBLANK=C*,MID=C,)
all of the blanks and asterisks, including those inside the quote pairs,
are squeezed out. The output field is:
"AB,"",C","D,E,"bbbbbbbbb
However, if we specify:
1,25,SQZ=(SHIFT=LEFT,PREBLANK=C*,MID=C,,
PAIR=QUOTE)
only the blanks and asterisks outside of the quote pairs are squeezed
out. The output field is:
"ABbb*""*C","D*Ebbbbbbbb
If a quote is specified in the PREBLANK list (for example,

PREBLANK=C'"' or PREBLANK=X'7F'), it is ignored, that is, a quote in
the input field is not replaced by a blank. So if you want each quote to
be replaced by a blank, do not specify PAIR=QUOTE.
For SHIFT=LEFT, scanning is left to right. If a quote is found with no
subsequent quote, all of the characters from the quote to the end of the
input field are ignored. For example, let's say we have an input field
of:
bbXbY"ABCbbbDbb
If we specify:
1,15,SQZ=(SHIFT=LEFT,PAIR=QUOTE)
there's an unpaired quote, so all of the characters from the quote to the
end of the input field are ignored. The output field is:
XY"ABCbbbDbbbbb
For SHIFT=RIGHT, scanning is right to left. If a quote is found with no

previous quote, all of the characters from the quote to the beginning of
the input field are ignored. For example, let's say we have an input
field of:
bbXbY"ABCbbbDbb
If we specify:
1,15,SQZ=(SHIFT=RIGHT,PAIR=QUOTE)
307

there's an unpaired quote, so all of the characters from the quote to the
beginning of the input field are ignored. The output field is:
bbbbbbbXbY"ABCD
Sample Syntax:
OUTFIL BUILD=(5:16,20,SQZ=(SHIFT=LEFT,PAIR=QUOTE,PREBLANK=C<>))
%nn,squeeze
specifies that a left-squeezed or right-squeezed parsed input field is to appear
in the reformatted OUTFIL output record. See PARSE for details of parsed
fields. See p,m,squeeze for further details.
seqnum
SEQNUM,n,fs

,START=j
,INCR=i
,RESTART=(p,m)
,RESTART=(%nn)
specifies that a sequence number is to appear in the reformatted OUTFIL

output record. The sequence numbers are assigned in the order in which the
records are received for OUTFIL OUTREC processing. You can create BI, PD,
ZD, CSF, or FS sequence numbers and control their lengths, starting values and
increment values. You can restart the sequence number at the start value each
time a specified OUTFIL input field (p,m) or parsed input field (%nn) changes.
n
specifies the length of the sequence number. The value for n must be
between 1 and 16.
fs specifies the format for the sequence number, which can be BI, PD, ZD,
CSF, or FS.
For a ZD format sequence number, F is used as the sign.
For a PD format sequence number, C is used as the sign.
For a CSF or FS format sequence number, blank is used as the sign and
leading zeros are suppressed.
For a PD, ZD, CSF, or FS format sequence number, the maximum value
DFSORT can create is limited to the lesser of 15 decimal digits or the
output field length (n). If a sequence number overflows this limit, it will be
truncated to the lesser of 15 decimal digits or the output field length, and
then subsequently incremented as usual.
For a BI format sequence number, the maximum value DFSORT can create
is limited to the lesser of 8 bytes of ones (X'FFFFFFFFFFFFFFFF') or the
number of ones that will fit in the specified output field length (n). If a
sequence number overflows this limit, it will be truncated to the lesser of 8
bytes or the output field length, and then subsequently incremented as
usual.
START
specifies the starting value for the sequence number.
j
specifies the starting value. The value for j must be between 0 and
100000000000. The default for j is 1.
INCR
specifies the increment value for the sequence number.
i
308
specifies the increment value. The value for i must be between 1 and
10000000. The default for i is 1.

RESTART
specifies that DFSORT is to restart the sequence number at the starting
value (START=j) when the binary value for the specified OUTFIL input
field or parsed input field changes. This allows you to sequence each
set of records with the same value (that is, duplicate records)
separately. For example: Without RESTART, if you had six OUTFIL
input records with 'A', 'A', 'A', 'B', 'B' and 'C', respectively, in position
1, the output records would be given the sequence numbers 1, 2, 3, 4, 5
and 6. But with RESTART=(1,1), the output records are given the
sequence numbers 1, 2, 3, 1, 2 and 1; DFSORT restarts at the starting
value (1, by default) when the input value at position 1 changes from
'A' to 'B' and again when it changes from 'B' to 'C'.
p,m or %nn
specifies the OUTFIL input field or parsed input field to be used to
determine when the sequence number is to be restarted at the
starting value. If a variable-length OUTFIL input record is too short
to contain a specified restart field, binary zeros (or blanks for a
parsed field) will be used for the missing bytes, intentionally or
unintentionally.
p
See p under p,m,a
specifies the length in bytes of the input field. The value for m
must be between 1 and 256.
%nn
specifies the parsed input field. See PARSE for details of
parsed fields.
Sample Syntax:
OUTFIL FNAMES=O1,OUTREC=(SEQNUM,6,ZD,START=1000,INCR=50,
X,22,8,X,13,5)
OUTFIL FNAMES=O2,OUTREC=(1,12,SEQNUM,4,BI)
OUTFIL FNAMES=O3,OUTREC=(1,80,81:SEQNUM,8,ZD,START=21,INCR=20,
RESTART=(35,8))
Default for BUILD or OUTREC: None; must be specified.

OVERLAY
309

,
OVERLAY= (
c:
s
p,m
,a
%nn
p,m
%nn
,TRAN
LTOU
UTOL
ALTSEQ
ATOE
ETOA
HEX
UNHEX
BIT
UNBIT
p,m
,HEX
%nn
p,m,f
,edit
%nn,f
,to
(p,m,f)
(%nn,f)
deccon
,edit
(deccon)
,to
arexp
,edit
(arexp)
,to
p,m
,Y2x
%nn
,Y4x
,edit
,to
,todate
,dateop
,Y2x(s)
,Y4x(s)
,Y2xP
p,m
,lookup
%nn
p,m
,justify
%nn
p,m
,squeeze
%nn
seqnum

rearrange, or delete fields, use BUILD or OUTREC rather than OVERLAY. Use
record. OVERLAY can be easier to use than BUILD or OUTREC when you just
OUTFIL OVERLAY=(25,2,11:CA,15,3,C**)
310

and C'**' is placed at output positions 15-16. The rest of the record remains
unchanged.
OUTFIL OVERLAY=(21:8,4,ZD,MUL,+10,TO=ZD,LENGTH=6,
5:5,1,TRAN=UTOL,
variable-length records, the RDW length is increased to correspond to the
For example, if your OUTFIL input record has a length of 40 and you specify:
OUTFIL OVERLAY=(16:CABC,51:5C*,35:15,2)
the OUTFIL output record is given a length of 55. Blanks are filled in from
columns 41-50. For variable-length records, the length in the RDW is changed
from 40 to 55 after all of the OVERLAY items are processed.
You can use the OVERLAY parameter with the FTOV parameter to convert
fixed-length record data sets to variable-length record data sets.
You can use the VLTRIM parameter with the OVERLAY parameter to remove
specified trailing bytes from the end of variable-length records.
You can use the VLTRAIL parameter with the OVERLAY parameter to insert a
string at the end of variable-length records.
You can use the OVERLAY parameter with any or all of the report parameters
in the same way as for the BUILD or OUTREC parameter.
The OVERLAY parameter of the OUTREC statement applies to all input
records whereas the OVERLAY parameter of the OUTFIL statement only
applies to the OUTFIL input records for its OUTFIL group.
See OUTFIL OUTREC for details of the items listed in the OVERLAY syntax
diagram shown previously in this section. You can specify all of the items for
OVERLAY in the same way that you can specify them for BUILD or OUTREC
with the following exceptions:
v You cannot specify / for OVERLAY.
v For p,m,H or p,m,F or or p,m,D fields specified for OVERLAY, fields are
aligned as necessary without changing the preceding bytes.
for OVERLAY, so be sure to specify the first column (c:) as 5 or greater. Do
not specify 1:, 2:, 3: or 4: anywhere in your OVERLAY parameter. If you do
not specify the first column, it will default to 1: which is invalid for
variable-length records with OVERLAY. Whereas OUTREC=(1,m,...) is
required, OVERLAY=(1,m) is not allowed since it would overlay the RDW.
311

Sample Syntax:
OUTFIL FNAMES=FLR,
OVERLAY=(21:21,4,ZD,TO=PD,LENGTH=4,
2:5,8,HEX,45:C*,32,4,C*,81:SEQNUM,5,ZD)

OUTFIL FNAMES=VLR,
OVERLAY=(5:X0001,28:CDate is ,YDDDNS=(4D),
17:5C*)
Default for OVERLAY: None; must be specified.

FINDREP
FINDREP
IN=incon,OUT=outcon
,
IN=( incon
,
,OUT=outcon
)

,
STARTPOS=p
ENDPOS=q
DO=n
MAXLEN=n
OVERRUN=ERROR
OVERRUN=TRUNC
SHIFT=YES
SHIFT=NO
You can use the FINDREP parameter with the FTOV parameter to convert
fixed-length record data sets to variable-length record data sets.
You can use the VLTRIM parameter with the FINDREP parameter to remove
specified trailing bytes from the end of variable-length records.
You can use the VLTRAIL parameter with the FINDREP parameter to insert a
string at the end of variable-length records.
You can use the FINDREP parameter with any or all of the report parameters
in the same way as for the BUILD or OUTREC parameter.
312

The FINDREP parameter of the OUTREC statement applies to all input records
whereas the FINDREP parameter of the OUTFIL statement only applies to the
OUTFIL input records for its OUTFIL group.
Input and output constants for find and replace processing are defined as
follows:
input constant
An input constant can be specified as a single character string, a
repeated character string, a single hexadecimal string, or a repeated
hexadecimal string. The syntax is: C'string', nC'string', X'string' or
nX'string'. n can be 1 to 256. The string can be 1 to 256 characters, or 1
to 256 pairs of hexadecimal digits. The total length of the constant
must not exceed 256 bytes. Use two apostrophes for a single
apostrophe.
output constant
An output constant can be specified as a null string, a single character
string, a repeated character string, a single hexadecimal string, or a
repeated hexadecimal string. The syntax is: C'' (null), C'string',
nC'string', X'string' or nX'string'. n can be 1 to 256. The string can be a
null, or 1 to 256 characters, or 1 to 256 pairs of hexadecimal digits. The
total length of the constant must not exceed 256 bytes. Use two
apostrophes for a single apostrophe. A null string can be used to
remove input constants.
You must specify the input and output constants to be used for find and
replace processing in one of the following ways. The default processing for
each method described later in this section can be changed with various
options described later.
IN=incon,OUT=outcon
Specifies one input constant and one output constant. Position 1 (for
fixed-length records) or 5 (for variable-length records) will be set as the
current position. The current position of the input record will be
checked for the input constant. If a match is not found at the current
position, the current position will be incremented by 1, and the process
will be repeated. If a match is found at the current position, the output
constant will replace the input constant, the current position will be
incremented past the input constant, and the process will be repeated.
Bytes after the replaced constants up to the end of the record will be
shifted left or right as needed. Processing will stop when the current
position is beyond the end of the input record.
Example:
OUTFIL IFTHEN=(WHEN=(11,1,CH,EQ,C3),
FINDREP=(IN=CYES,OUT=CNO))
Replaces every C'YES' input constant in records with a C'3' in position

11 with a C'NO' output constant, and shifts the bytes after each
replaced constant to the left.
IN=(incon1,incon2,...,inconx),OUT=outcon
Specifies multiple input constants and one output constant. Position 1
(for fixed-length records) or 5 (for variable-length records) will be set
as the current position. The current position of the input record will be
checked for each input constant in turn until a match is found or all of
the input constants have been checked. If a match is not found at the
current position for any input constant, the current position will be
313

incremented by 1, and the process will be repeated. If a match is found
at the current position, the output constant will replace the input
constant, the current position will be incremented past the input
constant, and the process will be repeated. Bytes after the replaced
constants up to the end of the record will be shifted left or right as
needed. Processing will stop when the current position is beyond the
end of the input record.
Example:
OUTFIL FINDREP=(IN=(XFF,3X00),OUT=C)
Removes every X'FF' and X'000000' input constant, and shifts the bytes
after each removed constant to the left.
INOUT=(incon1,outcon1,incon2,outcon2,...,inconx,outconx)
Specifies one or more pairs of input and output constants. Position 1
(for fixed-length records) or 5 (for variable-length records) will be set
as the current position. The current position of the input record will be
checked for each input constant in turn until a match is found or all of
the input constants have been checked. If a match is not found at the
current position for any input constant, the current position will be
incremented by 1, and the process will be repeated. If a match is found
at the current position for an input constant, the corresponding output
constant will replace the input constant, the current position will be
incremented past the input constant, and the process will be repeated.
Bytes after the replaced constants up to the end of the record will be
shifted left or right as needed. Processing will stop when the current
position is beyond the end of the input record.
Example:
OUTFIL FINDREP=(INOUT=(CSAT,CSATURDAY,CSUN,CSUNDAY))
Replaces every C'SAT' input constant with a C'SATURDAY' output

constant, and every C'SUN' input constant with a C'SUNDAY' output
constant, and shifts the bytes after each replaced constant to the right.
You can specify any or all of the following options to change the default
processing described earlier:
STARTPOS=p
Specifies the starting position in the input record for the find scan,
overriding the default of position 1 for a fixed-length record or
position 5 for a variable-length record. Use STARTPOS=p if you want
to start your find scan at a particular position. p can be 1 to 32752. If p
is less than 5 for a variable-length record, 5 will be used for p. If p is
beyond the end of the input record, find and replace processing will
not be performed for the record.
Example:
OUTFIL FINDREP=(IN=CYes,OUT=CYES,STARTPOS=11)
Replaces every C'Yes' input constant found starting at or after position

11 with a C'YES' output constant.
ENDPOS=q
Specifies the ending position in the input record for the find scan,
overriding the default of the end of the record. Use ENDPOS=q if you
want to end your find scan at a particular position. ENDPOS=q only
applies to the end of the find scan; bytes will still be shifted up to the
314

end of the record as needed. q can be 1 to 32752. If q is less than 5 for
a variable-length record, 5 will be used for q. If q is beyond the end of
the input record, the end of the record will be used for q. If
STARTPOS=p and ENDPOS=q are both specified, and p is greater than
q, find and replace processing will not be performed for the record.
Example:
OUTFIL FINDREP=(IN=(CD27,CA52,CX31),OUT=CINVALID,
ENDPOS=2015)
Replaces every C'D27', C'A52' and C'X31' input constant found before
or at position 2015 with a C'INVALID' output constant, and shifts the
bytes after each replaced constant to the right (past 2015 up to the end
of the record).
DO=n Specifies the maximum number of times find and replace is to be
performed for a record, overriding the default of every time. Scanning
for the input constant stops when n input constants have been found
and replaced. Use DO=n if you want to stop after a particular number
of constants have been replaced. n can be 1 to 1000.
Example:
OUTFIL FNAMES=OUT1,FINDREP=(IN=X015C,OUT=X015D,DO=3)
Replaces the first three X'015C' input constants found with X'015D'
output constants.
MAXLEN=n
Specifies the maximum length to be used for the output record created
by find and replace processing, overriding the default of using the
maximum length of the input record. Use MAXLEN=n if you want to
increase or decrease the output record length. (For IFTHEN FINDREP,
MAXLEN=n can only be used to increase the output record length, not
decrease it. See "Notes" later in this section for more information.)
If an output constant is larger than a corresponding input constant,
MAXLEN=n can be used to increase the size of the output record to
allow for shifting characters to the right. n can be 1 to 32752.
MAXLEN=n will be used to set the LRECL of the output data set,
when appropriate.
Example:
OUTFIL FINDREP=(INOUT=(C01,CJanuary,C02,CFebruary,
C03,CMarch),MAXLEN=150)
Replaces every C'01' input constant with a C'January' output constant,

every C'02' input constant with a C'February' output constant, and
every C'03' input constant with a C'March' output constant, and shifts
the bytes after each replaced constant to the right, allowing the record
to expand to 150 bytes.
OVERRUN=ERROR or OVERRUN=TRUNC
Specifies the action DFSORT is to take if an overrun occurs, that is:
v if nonblank bytes are shifted to the right past the end of the output
record as specified by MAXLEN=n, or as defaulted to the input
record length, or
v if MAXLEN=n is used to make the output record smaller than the
input record, and nonblank bytes are found in the input record past
the end of the output record.
315

OVERRUN=ERROR is the default; it tells DFSORT to issue an error
message and terminate if an overrun occurs.
Use OVERRUN=TRUNC if you want DFSORT to truncate the output
record to the MAXLEN=n or input record length if an overrun occurs,
rather than terminating. Bytes beyond the end of the output record are
lost.
Example:
OUTFIL FINDREP=(IN=XFFFF,OUT=CINVALID,MAXLEN=50,OVERRUN=TRUNC)
Replaces every X'FFFF' input constant with a C'INVALID' output

constant, and shifts the bytes after the replaced constants to the right.
50 byte output records are created. Bytes shifted past position 50 are
lost. Without OVERRUN=TRUNC, if nonblank characters were shifted
past position 50, an overrun error message would be issued and the
job would terminate.
SHIFT=YES or SHIFT=NO
Specifies the action DFSORT is to take if an input constant is to be
replaced by an output constant of a different length.
SHIFT=YES is the default; it tells DFSORT to shift the bytes after each
replaced input constant to the left or right as needed.
Use SHIFT=NO if you want DFSORT to overlay the input constant
with the output constant instead of shifting bytes. If a matching input
constant is found and the output constant is smaller than the input
constant, the current position will be incremented past the output
constant rather than past the input constant before processing
continues.
Example:
OUTFIL FINDREP=(IN=(CKW1=YES,,CKW1=NO,),OUT=CKW2,SHIFT=NO)
Replaces every C'KW1=YES,' input constant with a C'KW2=YES,'

output constant. Replaces every C'KW1=NO,' input constant with a
C'KW2=NO,' output constant. SHIFT=NO ensures that C'KW1' is
replaced with C'KW2' and the '=YES,' and '=NO,' bytes are kept.
Without SHIFT=NO, the '=YES,' and '=NO,' bytes would be removed.
Default for FINDREP: None; must be specified.
FINDREP notes:
v For a single FINDREP operand, when a constant is replaced at the current
position, no further checks are performed at that position. So a single
FINDREP cannot be used to change a constant and then change it again. For
example, if you had an input record with:
ABC
and used this OUTFIL statement:

OUTFIL FINDREP=(INOUT=(CAB,CXY,CXYC,CRST))
the output record would be:

XYC
After C'AB' is changed to C'XY', the current pointer is advanced to C'C' so

C'XYZ' is not found. However, since each IFTHEN clause starts at the
beginning of the record, multiple IFTHEN clauses with FINDREP can be
used to change a constant and then change it again. If you used:
316

OUTFIL IFTHEN=(WHEN=INIT,FINDREP=(INOUT=(CAB,CXY))),
IFTHEN=(WHEN=INIT,FINDREP=(INOUT=(CXYC,CRST)))
for the ABC record, the output would be:

RST
The first IFTHEN clause would change C'AB' to C'XY'. The second IFTHEN
clause would start over from the first position and change C'XYC' to C'RST'.
v Duplicates and supersets of the same input constant after the first are
effectively ignored, whereas subsets of the same input constant after the first
are processed. For example, if you had this input record:
ABCD ABQ
and you used this OUTFIL statement:

OUTFIL FINDREP=(INOUT=(CAB,CXY,CABCD,CRSTU))

XYCD XYQ
Since C'AB' is changed to C'XY', C'ABCD' is not found. If you wanted to

change C'ABCD' to C'RSTU' and other instances of C'AB' to C'XY', you
would need to specify C'ABCD' first. If you used this OUTFIL statement:
OUTFIL FINDREP=(INOUT=(CABCD,CRSTU,CAB,CXY))

RSTU XYQ
If C'ABCD' is found, it is changed to C'RSTU'. Otherwise, if C'AB' is found,

it is changed to C'XY'.
v For fixed-length records, FINDREP in an IFTHEN clause operates against the
maximum record padded with blanks on the right as needed. If a blank
constant is being replaced, the blanks on the right will be replaced. For
example, if we have an 80 byte FB input record and use:
OUTFIL IFTHEN=(WHEN=INIT,BUILD=(1,20)),
IFTHEN=(WHEN=(2,1,CH,EQ,CR),
FINDREP=(IN=C ,OUT=CABC))
Although in this case BUILD=(1,20) will result in a 20-byte output record,

IFTHEN FINDREP will operate against the maximum 80-byte record padded
with 60 blanks on the right. For a record with 'R' in position 2, each of those
60 blanks on the right will be replaced with C'ABC'. This will cause an
overrun of the 80 byte record, thus resulting in termination
(OVERRUN=ERROR) or truncation (OVERRUN=TRUNC). ENDPOS=20
could be used to stop FINDREP from replacing the 60 blanks on the right.
Note that if we have an 80-byte input record with 'ABC' in positions 1-3 and
we use:
IFTHEN=(WHEN=INIT,FINDREP=(IN=CABC,OUT=C12345))
an overrun will not occur because IFTHEN FINDREP will operate against an
80-byte record padded with 77 blanks on the right, rather than against a
3-byte record even though the output record will be 3 bytes.
v For FINDREP in an IFTHEN clause, MAXLEN=n can be used to increase the
maximum output length, but cannot be used to decrease the maximum
output length. For example, with:
IFTHEN=(WHEN=INIT,
FINDREP=(IN=CA,OUT=CB,MAXLEN=150))
the output length will be set to 150, whereas with:
317

IFTHEN=(WHEN=INIT,
FINDREP=(IN=CA,OUT=CB,MAXLEN=70))
the output length will be set to 100.

IFTHEN
,
IFTHEN=(
WHEN=INIT
,BUILD=(items)
,OVERLAY=(items)
,FINDREP=(items)
,KEYBEGIN=(p,m)
,END=(logexp)
,RECORDS=n
,BUILD=(items)
,HIT=NEXT
,OVERLAY=(items)
,FINDREP=(items)
,BUILD=(items)
,HIT=NEXT
,OVERLAY=(items)
,FINDREP=(items)
,BUILD=(items)
,OVERLAY=(items)
,FINDREP=(items)

reformatted.
record, use BUILD or OUTREC rather than IFTHEN. If you want to do find
and replace operations in the same way for every record, use FINDREP rather
than IFTHEN. Use IFTHEN clauses if you want to insert, rearrange, delete, or
overlay fields, or perform find/replace operations in different ways for
different records, or if you want to perform operations on groups of records.
v WHEN=INIT: Use one or more WHEN=INIT clauses to apply build, overlay,
clauses.
evaluates as true.
318

v WHEN=INIT clauses and WHEN=GROUP clauses
v WHEN=NONE clauses
not specified.
v A WHEN=(logexp), WHEN=ANY or WHEN=NONE clause is satisfied, and
BUILD with / is specified (multiple output records).
Example:
OUTFIL IFTHEN=(WHEN=(12,1,BI,ALL,X3F),OVERLAY=(18:CYes)),
IFTHEN=(WHEN=(35,2,PD,EQ,+8),BUILD=(1,40,2/,45,10)),

If IFTHEN clause 1 is satisfied, its overlay item is applied and IFTHEN
processing stops.
If IFTHEN clause 1 is not satisfied, its overlay item is not applied and
If IFTHEN clause 2 is satisfied, its build items are applied and IFTHEN
If IFTHEN clause 2 is not satisfied, its build items are not applied and
processing stops.
processing stops.
If IFTHEN clause 6 is satisfied, its overlay item is applied and IFTHEN
processing stops.
If IFTHEN clause 6 is not satisfied, its overlay item is not applied and
319

processing stops.
OUTFIL IFTHEN=(WHEN=INIT,OVERLAY=(8:8,4,ZD,ADD,+1,TO=ZD,LENGTH=4)),
DFSORT sets an appropriate LRECL for the OUTFIL output records based on
the build, overlay, find/replace and group operation items specified by the
IFTHEN clauses. However, DFSORT does not analyze the possible results of
you use OUTFIL IFTHEN clauses, you can override the OUTFIL LRECL
determined by DFSORT with the OUTFIL IFOUTLEN parameter.
input is:
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
A
B
B
C
A
C
B
D
1
1
2
1
2
2
3
1
and you specify:

OUTFIL IFTHEN=(WHEN=(8,1,CH,EQ,CA),OVERLAY=(15:SEQNUM,4,ZD)),

RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
320
A
B
B
C
A
C
B
D
1
1
2
1
2
2
3
1
0001
0001
0002
0001
0002
0002
0003
0003

You can use IFTHEN clauses with the FTOV parameter to convert fixed-length
record data sets to variable-length record data sets.
You can use the VLTRIM parameter with IFTHEN clauses to remove specified
trailing bytes from the end of variable-length records.
You can use the VLTRAIL parameter with IFTHEN clauses to insert a string at
the end of variable-length records.
You can use IFTHEN clauses with any or all of the report parameters (LINES,
HEADER1, TRAILER1, HEADER2, TRAILER2, SECTIONS, BLKCCH1,
BLKCCH2, BLKCCT1, and NODETAIL) in the same way as for the BUILD or
OUTREC parameter.
The IFTHEN clauses of the OUTREC statement apply to all input records
whereas the IFTHEN clauses of the OUTFIL statement only apply to the
WHEN=INIT clause
Specifies build, overlay or find/replace items to be applied to all records.
Multiple WHEN=INIT clauses and WHEN=GROUP clauses may be
intermixed. They are processed before any other type of IFTHEN clauses
and in the order in which they are specified.
WHEN=INIT
Identifies a WHEN=INIT clause.
PARSE=(definitions)
Defines %nn fixed parsed fields into which variable position/length
fields are extracted for all records. See OUTFIL PARSE for details. You
can use the %nn parsed fields defined in a WHEN=INIT clause in the
BUILD or OVERLAY operand of that clause and all subsequent
IFTHEN clauses.
Sample Syntax:
OUTFIL IFOUTLEN=50,
IFTHEN=(WHEN=INIT,
PARSE=(%01=(ABSPOS=11,STARTAT=NONBLANK,
ENDBEFR=C,,FIXLEN=12),
%03=(FIXLEN=12))),
IFTHEN=(WHEN=(5,2,CH,EQ,CUS),BUILD=(%02,%03)),
IFTHEN=(WHEN=(5,2,CH,EQ,CUK),BUILD=(%01,%02)),
IFTHEN=(WHEN=NONE,BUILD=(%03,%01))
BUILD=(items)
Specifies the build items to be applied to each record. See OUTFIL
BUILD for details. You can specify all of the items in the same way
that you can specify them for OUTFIL BUILD, except that you cannot
specify / to create blank records or new records.
OVERLAY=(items)
Specifies the overlay items to be applied to each record. See OUTFIL
OVERLAY for details. You can specify all of the items in the same way
that you can specify them for OUTFIL OVERLAY.
Sample Syntax:
321

OUTFIL FNAMES=OUT1,
IFTHEN=(WHEN=INIT,BUILD=(1,20,21:CDepartment,31:3X,21,60)),
FINDREP=(items)
Specifies find and replace operations to be applied to each record. See
OUTFIL FINDREP for details. You can specify all of the items in the
same way that you can specify them for OUTFIL FINDREP.
Sample Syntax
OUTFIL FNAMES=OUT2,
IFTHEN=(WHEN=INIT,FINDREP=(IN=X00,OUT=X40)),
IFTHEN=(WHEN=INIT,OVERLAY=(81:SEQNUM,8,ZD))
WHEN=GROUP clause
Identifies groups of records in various ways and propagates fields,
identifiers and sequence numbers to the records of each group. Fields,
identifiers and sequence numbers are not propagated to records before,
between or after identified groups. Multiple WHEN=INIT clauses and
WHEN=GROUP clauses may be intermixed. They are processed before any
other type of IFTHEN clauses and in the order in which they are specified.
You can specify the BEGIN, KEYBEGIN, END and RECORDS operands in
any combination to define the groups, but you must specify at least one of
these operands.
BEGIN=(logexp)
Specifies the criteria to be tested to determine if a record starts a
group. See the INCLUDE statement for details of the logical
expressions you can use. You can specify all of the logical expressions
in the same way that you can specify them for the INCLUDE
statement except that:
v You cannot specify FORMAT=f with BEGIN=(logexp).
v You cannot specify D2 format in BEGIN=(logexp).
v Locale processing is not used for BEGIN=(logexp).
v VLSCMP and VLSHRT are not used with BEGIN=(logexp). Instead,
missing bytes in specified input fields are replaced with blanks so
the padded fields can be processed.
A new group starts with a record that satisfies the BEGIN condition,
that is, when the specified logical expression is true for that record. If
BEGIN is specified without END or RECORDS, all of the records from
the begin record up to but not including the next begin record belong
to a group. Here's an example of groups with
BEGIN=(1,1,CH,EQ,CA):
H
R
A
B
C
A
A
B
group
group
group
group
group
group
1
1
1
2
3
3
Example:
OUTFIL IFTHEN=(WHEN=GROUP,
PUSH=(41:ID=5))
322

Starts a new group each time C'J82' or 'M72' is found anywhere in
positions 1-40 of a record. Overlays positions 41-45 of each record of a
group with a 5-byte ZD identifier. The records before the first group
are not changed.
KEYBEGIN=(p,m)
Specifies a field to be used to determine if a record starts a group. A
new group starts when the binary value in the specified key field (p,m)
changes. The first input record will start a group. KEYBEGIN=(p,m)
operates in a similar way to BEGIN=(logexp) but uses a key change as
a trigger rather than a condition to be satisfied. Use KEYBEGIN=(p,m)
if you want to start a new group for each consecutive set of key values
in your records.
p is the starting position of the key and can be from 1 to 32752.
m is the length of the key and can be from 1 to 256.
If KEYBEGIN=(p,m) is specified, the first input record will start a
group.
Here is an example of groups with
KEYBEGIN=(1,1)
A
A
A
B
B
C
group
group
group
group
group
group
1
1
1
2
2
3
Example:
KEYBEGIN=(11,5),PUSH=(81:21,8))
Starts a new group each time the value in positions 11-15 changes.
Overlays positions 81-88 of each record of a group with positions 21-28
from the first record of that group.
END=(logexp)
Specifies the criteria to be tested to determine if a record ends a group.
See the INCLUDE statement for details of the logical expressions you
can use. You can specify all of the logical expressions in the same way
that you can specify them for the INCLUDE statement except that:
v You cannot specify FORMAT=f with END=(logexp).
v You cannot specify D2 format in END=(logexp).
v Locale processing is not used for END=(logexp).
v VLSCMP and VLSHRT are not used with END=(logexp). Instead,
A group ends whenever a record satisfies the END condition, that is,
whenever the specified logical expression is true for that record. If
END is specified without BEGIN or KEYBEGIN, all of the records up
to the end record (or the last record of the data set) belong to a group.
Here's an example of groups with
END=(1,1,CH,EQ,CT)
A group 1
B group 1
T group 1
323

T
A
T
M
group
group
group
group
2
3
3
4
If END is specified with BEGIN or KEYBEGIN, all of the records from

the begin record up to the end record (or the last record of the data
set) belong to a group. Here's an example of groups with
BEGIN=(1,1,CH,EQ,CH),END=(1,1,CH,EQ,CT)
H
B
T
T
H
T
M
N
H
A
group 1
group 1
group 1
group 2
group 2
group 3
group 3
Example:
BEGIN=(1,2,CH,EQ,C02,AND,8,3,CH,EQ,CYES),
END=(11,5,CH,EQ,CPAGE:),PUSH=(61:SEQ=3))
Starts a new group each time C'02' is found in positions 1-2 and C'YES'
is found in positions 8-10 of a record. Ends the group when C'PAGE:'
is found in positions 11-15 of a record. Overlays positions 61-63 of each
record of a group with a 3-byte ZD sequence number. The records
between the groups are not changed.
RECORDS=n
Specifies the maximum number of records in a group. n can be 1 to
2000000000. If RECORDS is specified without BEGIN, KEYBEGIN or
END, a new group starts after n records. Here's an example of groups
with
RECORDS=3
H
B
T
H
B
T
M
N
group
group
group
group
group
group
group
group
1
1
1
2
2
2
3
3
If RECORDS is specified with BEGIN or KEYBEGIN, up to n records

starting with the begin record belong to a group. Here's an example of
groups with
BEGIN=(1,1,CH,EQ,CH),RECORDS=3
H
B
T
A
H
B
H
M
N
P
324
group 1
group 1
group 1
group
group
group
group
group
2
2
3
3
3

The records between and after the groups are not changed.
If RECORDS is specified with END, the group ends after up to n
records or with the end record (or the last record of the data set),
whichever comes first. Here's an example of groups with
BEGIN=(1,1,CH,EQ,CH),END=(1,1,CH,EQ,CT),RECORDS=4
H
B
T
A
H
B
C
D
E
H
M
group 1
group 1
group 1
group
group
group
group
2
2
2
2
group 3
group 3
The records between the groups are not changed.

You must specify the PUSH operand to define at least one field, identifier
or sequence number to be added to the records of each group. You can
specify any combination or number of fields, identifiers and sequence
numbers.
PUSH=(c:item,...)
Specifies the position where each field, identifier or sequence number
is to be overlaid in the records of each group.
For fixed-length records, the first input and output data byte starts at
position 1. For variable-length records, the first input and output data
byte starts at position 5, after the RDW in positions 1-4.
You can use the following in PUSH:
c: Specifies the output position (column) to be overlaid. If you do not
specify c: for the first item, it defaults to 1:. If you do not specify c:
for any other item, it starts after the previous item. You can specify
items in any order and overlap output columns. c can be 1 to
32752.
If you specify an item that extends the output record beyond the
end of the input record, the record length is automatically
increased to that length, and blanks are filled in on the left as
needed. For variable-length records, the RDW length is increased
to correspond to the larger record length after all of the items are
processed. Missing bytes in specified input fields are replaced with
blanks so the padded fields can be processed.
p,m
Specifies a field in the first input record of each group to be
propagated to every record of the group. p specifies the starting
position of the field in the input record and m specifies its length.
A field must not extend beyond position 32752.
ID=n
Specifies a ZD identifier of length n is to be added to every record
of each group. The identifier starts at 1 for the first group and is
incremented by 1 for each subsequent group. n can be 1 to 15.
SEQ=n
Specifies a ZD sequence number of length n is to be added to
325

every record of each group. The sequence number starts at 1 for
the first record of each group and is incremented by 1 for each
subsequent record of the group. n can be 1 to 15.
Sample Syntax
BEGIN=(1,5,CH,EQ,CDATE:),
RECORDS=3,
PUSH=(15:ID=3,31:21,8))
Specifies build, overlay or find/replace items to be applied to records for
which the specified logical expression is true. If multiple WHEN=(logexp)
clauses are specified, they are processed in the order in which they are
specified.
WHEN=(logexp)
Identifies a WHEN=(logexp) clause and specifies the criteria to be
tested to determine if the build, overlay or find/replace items are to be
applied to the record. See the INCLUDE statement for details of the
logical expressions you can use. You can specify all of the logical
expressions in the same way that you can specify them for the
INCLUDE statement except that:
v You cannot specify FORMAT=f with WHEN=(logexp).
v You cannot specify D2 format in WHEN=(logexp).
v Locale processing is not used for WHEN=(logexp).
v VLSCMP and VLSHRT are not used with WHEN=(logexp). Instead,
PARSE=(definitions)
fields are extracted for each record for which the logical expression is
true. See OUTFIL PARSE for details. You can only use the %nn parsed
fields defined in a WHEN=(logexp) clause in the BUILD or OVERLAY
operand of that WHEN=(logexp) clause.
Sample Syntax:
OUTFIL IFTHEN=(WHEN=(45,2,CH,EQ,CAL),
PARSE=(%01=(ENDBEFR=C,,FIXLEN=12),
%02=(FIXLEN=10)),
OVERLAY=(21:%02,51:%01))
BUILD=(items)
Specifies the build items to be applied to each record for which the
logical expression is true. See OUTFIL BUILD for details. You can
specify all of the items in the same way that you can specify them for
OUTFIL BUILD.
OVERLAY=(items)
Specifies the overlay items to be applied to each record for which the
logical expression is true. See OUTFIL OVERLAY for details. You can
OUTFIL OVERLAY.
FINDREP=(items)
Specifies find and replace operations to be applied to each record for
326

which the logical expression is true. See OUTFIL FINDREP for details.
You can specify all of the items in the same way that you can specify
them for OUTFIL FINDREP.
Sample Syntax
OUTFIL IFTHEN=(WHEN=(11,2,CH,EQ,C**),
FINDREP=(IN=C**,OUT=C))
HIT=NEXT
Specifies that IFTHEN processing should continue even if the logical
expression is true. By default (if HIT=NEXT is not specified), IFTHEN
processing stops if the logical expression is true.
Sample Syntax:
OUTFIL FNAMES=OUT2,
BUILD=(1,21,42,13)),
BUILD=(1,25,42,13))
WHEN=ANY clause
which the logical expression in any "preceding" WHEN=(logexp) clause is
true. For the first WHEN=ANY clause, the "preceding" WHEN=(logexp)
clauses are those before this WHEN=ANY clause. For the second or
subsequent WHEN=ANY clause, the "preceding" WHEN=(logexp) clauses
are those between the previous WHEN=ANY clause and this WHEN=ANY
clause. At least one WHEN=(logexp) clause must be specified before a
WHEN=ANY clause. A WHEN=ANY clause can be used without any
build, overlay or find/replace items to just stop IFTHEN processing if any
preceding WHEN=(logexp) clause is satisfied.
PARSE=(definitions)
fields are extracted for each record for which the logical expression in
any preceding WHEN=(logexp) clause is true. See OUTFIL PARSE for
details. You can only use the %nn parsed fields defined in a
WHEN=ANY clause in the BUILD or OVERLAY operand of that
WHEN=ANY clause.
BUILD=(items)
Specifies the build items to be applied to each record for which the
logical expression in any preceding WHEN=(logexp) clause is true. See
OUTFIL BUILD for details. You can specify all of the items in the same
way that you can specify them for OUTFIL BUILD.
OVERLAY=(items)
Specifies the overlay items to be applied to each record for which the
logical expression in any preceding WHEN=(logexp) clause is true. See
OUTFIL OVERLAY for details. You can specify all of the items in the
same way that you can specify them for OUTFIL OVERLAY.
FINDREP=(items)
which the logical expression in any preceding WHEN=(logexp) clause
327

is true. See OUTFIL FINDREP for details. You can specify all of the
items in the same way that you can specify them for OUTFIL
FINDREP.
HIT=NEXT
Specifies that IFTHEN processing should continue even if the logical
expression in a preceding WHEN=(logexp) clause is true. By default (if
HIT=NEXT is not specified), IFTHEN processing stops if the logical
expression in a preceding WHEN=(logexp) clause is true.
Sample Syntax:
OUTFIL FNAMES=OUT3,
IFTHEN=(WHEN=ANY,OVERLAY=(16:CGroup Found))
WHEN=NONE clause
which none of the logical expressions in any WHEN=(logexp) clause is
true. If there are no WHEN=(logexp) clauses, the build, overlay or
find/replace items are applied to all of the records. If multiple
WHEN=NONE clauses are specified, they are processed in the order in
which they are specified. WHEN=NONE clauses are processed after any
other type of IFTHEN clauses.
PARSE=(definitions)
fields are extracted for each record for which no logical expression was
true. See OUTFIL PARSE for details. You can only use the %nn parsed
fields defined in a WHEN=NONE clause in the BUILD or OVERLAY
operand of that WHEN=NONE clause.
Sample Syntax:
OUTFIL IFTHEN=(WHEN=(5,2,ZD,GT,+5),
PARSE=(%01=(STARTAT=C<,ENDAT=C>,FIXLEN=12),
BUILD=(5,2,21:%01,X,%02,HEX)),
IFTHEN=(WHEN=NONE,
PARSE=(%03=(STARTAFT=C(,ENDBEFR=C),FIXLEN=8)),
BUILD=(5,2,%03,SFF,M12))
BUILD=(items)
Specifies the build items to be applied to each record for which no
logical expression was true. See OUTFIL BUILD for details. You can
OUTFIL BUILD.
OVERLAY=(items)
Specifies the overlay items to be applied to each record for which no
logical expression was true. See OUTFIL OVERLAY for details. You can
OUTFIL OVERLAY.
Sample Syntax:
328

OUTFIL FNAMES=OUT4,
IFTHEN=(WHEN=INIT,BUILD=(1,20,21:CDepartment,31:3X,21,60)),
FINDREP=(items)
which no logical expression was true. See OUTFIL FINDREP for
details. You can specify all of the items in the same way that you can
specify them for OUTFIL FINDREP.
Default for IFTHEN clauses: None; must be specified.
IFOUTLEN
IFOUTLEN=n
Overrides the OUTFIL LRECL determined by DFSORT from your OUTFIL

IFTHEN clauses. DFSORT sets an appropriate LRECL for the output records
based on the build, overlay, find/replace and group operation items specified
by the IFTHEN clauses. However, DFSORT does not analyze the possible
results of WHEN=(logexp) conditions when determining an appropriate
OUTFIL LRECL. When you use OUTFIL IFTHEN clauses, you can override the
OUTFIL LRECL determined by DFSORT with the OUTFIL IFOUTLEN
parameter.
specifies the length to use for the OUTFIL LRECL. The value for n must be
between 1 and 32767, but must not be larger than the maximum LRECL
allowed for the RECFM, and must not conflict with the specified or
retrieved LRECL for the fixed-length OUTFIL data set.
Sample Syntax:
OUTFIL FNAMES=OUT5,IFOUTLEN=70,
VTOF or CONVERT
VTOF
CONVERT
Specifies that variable-length OUTFIL input records are to be converted to

fixed-length OUTFIL output records for this OUTFIL group.
You must specify a BUILD or OUTREC parameter. The items you specify
produce a reformatted fixed-length OUTFIL output record without an RDW
(the data starts at position 1). Any BUILD or OUTREC fields you specify apply
to the variable-length OUTFIL input records (the data starts at position 5 after
the 4-byte RDW). However, you cannot specify the variable-part of the OUTFIL
329

input records (for example, p or p,HEX). Any BUILD or OUTREC columns you
specify apply to the reformatted fixed-length OUTFIL output records.
By default, VTOF or CONVERT automatically uses VLFILL=X'40' (blank fill
byte) to allow processing of variable-length input records which are too short
to contain all specified BUILD or OUTREC fields. You can specify
VLFILL=byte to change the fill byte.
If you do not specify a RECFM for the OUTFIL data set, it will be given a
record format of FB.
If you specify a RECFM for the OUTFIL data set, it must have a fixed-length
record format (for example, FB).
If VTOF or CONVERT is specified for fixed-length input records, it will not be
used.
If VTOF or CONVERT is specified with FTOV, IFTRAIL, IFTHEN, FINDREP or
OVERLAY, DFSORT will terminate.
Sample Syntax:
OUTFIL FNAMES=FIXOUT,VTOF,
OUTREC=(1:5,14,35:32,8,50:22,6,7c*)
Default for VTOF or CONVERT: None; must be specified.

VLFILL
VLFILL=byte
Allows DFSORT to continue processing if a variable-length OUTFIL input

record is found to be too short to contain all specified OUTFIL BUILD or
OUTREC fields for this OUTFIL group. Without VLFILL=byte, a short record
causes DFSORT to issue message ICE218A and terminate. With VLFILL=byte,
missing bytes in OUTFIL BUILD or OUTREC fields are replaced with fill bytes
so the filled fields can be processed.
If VLFILL=byte is specified for fixed-length input records, it will not be used.
If VLFILL=byte is specified with FTOV, IFTRAIL, IFTHEN, FINDREP or
OVERLAY, DFSORT will terminate.
byte
specifies the fill byte. Permissible values are C'x' and X'yy'.
C'x'
Character byte: The value x must be one EBCDIC character. If you want to
use an apostrophe as the fill byte, you must specify it as C''''.
X'yy'
Hexadecimal byte: The value yy must be one pair of hexadecimal digits
(00-FF).
Sample Syntax:
OUTFIL FNAMES=FIXOUT,VTOF,OUTREC=(5,20,2X,35,10),VLFILL=C*
OUTFIL FNAMES=OUT1,VLFILL=XFF,OUTREC=(1,4,15,5,52)
Default for VLFILL: VLFILL=X'40' (blank fill byte) if VTOF or CONVERT is

specified. Otherwise, none; must be specified.
FTOV
FTOV
Specifies that fixed-length OUTFIL input records are to be converted to
330

variable-length OUTFIL output records for this OUTFIL group.
If you do not specify an OUTREC, BUILD, OVERLAY, FINDREP or IFTHEN
parameter, the fixed-length OUTFIL input record is converted to a
variable-length OUTFIL output record. A 4-byte RDW is prepended to the
fixed-length record before it is written.
If you specify an OUTREC, BUILD, OVERLAY, FINDREP or IFTHEN
parameter, the items you specify produce a reformatted fixed-length record
that is converted to a variable-length OUTFIL output record. Any OUTREC,
BUILD, OVERLAY, FINDREP or IFTHEN fields you specify apply to the
fixed-length OUTFIL input records (the data starts at position 1). A 4-byte
RDW is prepended to the reformatted fixed-length record before it is written.
If you do not specify a RECFM for the OUTFIL data set, it will be given a
record format of VB.
If you specify a RECFM for the OUTFIL data set, it must have a
variable-length record format (for example, VB or VBS).
If you do not specify an LRECL for the OUTFIL data set, it will be given an
LRECL that can contain the largest variable-length output record to be
produced, up to a maximum of 32756 for an unspanned record format (for
example, VB) or up to 32767 for a spanned record format (for example, VBS).
If you specify an LRECL for the OUTFIL data set, it must be big enough to
contain the largest variable-length output record to be produced.
If your largest variable-length output record is between 32757 and 32767 bytes,
you'll need to specify a spanned record format (for example, VBS) for the
output data set.
If FTOV is specified for variable-length input records, it will not be used.
If FTOV is specified with VTOF, IFTRAIL, CONVERT or VLFILL=byte,
DFSORT will terminate.
Sample Syntax:
OUTFIL FNAMES=VAROUT,FTOV
OUTFIL FNAMES=V1,FTOV,OUTREC=(1,20,26:21,10,6C*)
OUTFIL FNAMES=V2,FTOV,
IFTHEN=(WHEN=(12,5,ZD,GT,+20000),OVERLAY=(25:CYes)),
Default for FTOV: None; must be specified.
VLTRIM
VLTRIM=byte
VLTRIM=byte specifies that the trailing bytes are to be removed from the end
of variable-length OUTFIL output records for this OUTFIL group before the
records are written.
The trim byte can be any value, such as blank, binary zero, or asterisk. If
DFSORT finds one or more trim bytes at the end of a variable-length OUTFIL
data record or report record, it will decrease the length of the record
accordingly, effectively removing the trailing trim bytes. However,
VLTRIM=byte will not remove the RDW, the ANSI carriage control character (if
produced), or the first data byte.
For example, say that you have the following 17-byte fixed-length data records
that you want to convert to variable-length data records:
123456***********
0003*************
ABCDEFGHIJ*****22
*****************
If you use:
331

OUTFIL FTOV
the following variable-length output records will be written (4-byte RDW

followed by data):
Length
21
21
21
21
Data
123456***********
0003*************
ABCDEFGHIJ*****22
*****************
but if you use:

OUTFIL FTOV,VLTRIM=C*
the following variable-length output records will be written (4-byte RDW

followed by data):
Length
10
8
21
5
Data
123456
0003
ABCDEFGHIJ*****22
*
VLTRIM=C'*' removed the trailing asterisks from the first and second records.
The third record did not have any trailing asterisks to remove. The fourth
record had all asterisks, so one asterisk was kept.
If VLTRIM=byte is specified for fixed-length output records, it will not be
used.
If VLTRIM is specified with IFTRAIL, DFSORT will terminate.
byte
specifies the trim byte. Permissible values are C'x' and X'yy'.
C'x'
Character byte: The value x must be one EBCDIC character. If

you want to use an apostrophe as the trim byte, you must
specify it as C''''.
X'yy'
Hexadecimal byte: The value yy must be one pair of

hexadecimal digits (00-FF)
Sample Syntax:
Fixed input:
OUTFIL FNAMES=TRIM1,FTOV,VLTRIM=C
Variable input:
OUTFIL FNAMES=TRIM2,VLTRIM=X00
OUTFIL FNAMES=TRIM3,VLTRIM=C*,
OUTREC=(1,15,5X,16,8,5X,28)
Default for VLTRIM: None; must be specified.
VLTRAIL
VLTRAIL=string
VLTRAIL=string specifies that the indicated string is to be inserted at the end

of variable-length OUTFIL output records for this OUTFIL group before the
records are written.
(C'xx...x') or a hexadecimal string constant (X'yy...yy'). See INCLUDE control
statement on page 96 for details of coding character and hexadecimal string
constants.
332

DFSORT will insert the specified string at the end of each variable-length
OUTFIL data record or report record, and increase the length of the record
accordingly. You must ensure that the LRECL of the OUTFIL data set is large
enough to contain the added bytes. If the added bytes increase the length of a
record beyond the LRECL for the data set, DFSORT will terminate. To avoid
this, you can specify a larger LRECL on the DD statement, when necessary.
For example, say that you have the following variable-length input records:
Length | Data
28
Professional Development
18
Psychoanalysis
25
Theory of Computation
and you want to add X'0D0A' (CR,LF) to the end of each record. If you use the
following OUTFIL statement:
OUTFIL FNAMES=OUT1,VLTRAIL=X0D0A
X'0D0A' will be inserted at the end of each record and the length of each
record will be increased by 2 bytes. If we represent X'0D0A' by CL, the OUT1
records will look like this:
Length | Data
30
Professional DevelopmentCL
20
PsychoanalysisCL
27
Theory of ComputationCL
Note that if the LRECL of the output data set is less than the maximum output
record (30 bytes for this example), we must increase it to at least the maximum
(LRECL=30 or greater for this example).
If VLTRIM=byte is used with VLTRAIL=string, the indicated bytes will be
trimmed from the end of the record before the indicated string is added. For
example, say that you have the following variable-length input records:
Length | Data
30
Abby is a sweet rat*******
31
Samantha is a curious rat**
43
Rachel is our oldest rat***************
29
Cathy is a mischief maker
and you want to remove any trailing asterisks, and add '|March 21|' at the
end of each record. If you use the following OUTFIL statement:
OUTFIL FNAMES=OUT2,VLTRIM=C*,VLTRAIL=C|March 21|
The trailing asterisks will be removed, 'March 21' will be inserted, and the
length of each record will be updated appropriately. The OUT2 records will
look like this:
Length | Data
33
Abby is a sweet rat|March 21|
39
Samantha is a curious rat|March 21|
38
Rachel is our oldest rat|March 21|
39
Cathy is a mischief maker|March 21|
If VLTRAIL=string is specified for fixed-length output records, it will not be

used.
If VLTRAIL is specified with IFTRAIL, DFSORT will terminate.
Sample Syntax:
Variable input:
OUTFIL FNAMES=ADD1,VLTRIM=X00,VLTRAIL=C$$ADD$$
OUTFIL FNAMES=(ADD2A,ADD2B),VLTRAIL=XFFFF,SPLIT
333
Fixed input:
OUTFIL FNAMES=ADD3,FTOV,VLTRIM=C ,VLTRAIL=Cend
Default for VLTRAIL: None; must be specified.

REPEAT
REPEAT=n
Specifies the number of times each OUTFIL output record is to be repeated for
this OUTFIL group. Each OUTFIL output record is written n times.
If SEQNUM is used in the OUTREC, BUILD, OVERLAY, or IFTHEN parameter
for this OUTFIL group, the sequence number will be incremented for each
repeated record. For example, if your input is:
RECORD A
RECORD B
and you specify:

OUTFIL OUTREC=(1,8,X,SEQNUM,5,ZD),REPEAT=2

RECORD
RECORD
RECORD
RECORD
A
A
B
B
00001
00002
00003
00004
If SEQNUM is used in multiple IFTHEN clauses for this OUTFIL group, the
sequence number will be incremented for each repeated record that satisfies
the IFTHEN clause. For example, if your input is:
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
A
B
C
A
C
B
B
1
1
1
2
2
2
3
and you specify:

OUTFIL REPEAT=2,
IFTHEN=(WHEN=(8,1,CH,EQ,CA),OVERLAY=(15:SEQNUM,4,ZD)),

RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
334
A
A
B
B
C
C
A
A
C
C
B
B
B
B
1
1
1
1
1
1
2
2
2
2
2
2
3
3
0001
0002
0001
0002
0001
0002
0003
0004
0003
0004
0003
0004
0005
0006

If you specify REPEAT=n with / in the OUTREC, BUILD, or IFTHEN BUILD
parameter for this OUTFIL group, the first line is written n times, then the
second line is written n times, and so on. (For IFTHEN, this means each record
that has the SEQNUM item applied to it.) If SEQNUM is used, all lines for the
same record are given the same sequence number. For example, if your input
is:
RECORD A
RECORD B
and you specify:

OUTFIL OUTREC=(CP1>,X,1,6,X,SEQNUM,4,ZD,/,
CP2>,X,8,1,X,SEQNUM,4,ZD),REPEAT=2

P1>
P1>
P2>
P2>
P1>
P1>
P2>
P2>
RECORD
RECORD
A 0001
A 0002
RECORD
RECORD
B 0003
B 0004
0001
0002
0003
0004
The REPEAT parameter cannot be used with any of the following parameters:
IFTRAIL, LINES, HEADER1, TRAILER1, HEADER2, TRAILER2, SECTIONS,
and NODETAIL.
n
specifies the number of times each OUTFIL output record is to be repeated.

The value for n starts at 2 (write record twice) and is limited to 28 digits
(15 significant digits).
Sample Syntax:
* WRITE EACH OUTPUT RECORD 12 TIMES.
OUTFIL FNAMES=OUT1,REPEAT=12
* WRITE EACH INCLUDED AND REFORMATTED OUTPUT RECORD 50 TIMES.
* (THE SEQUENCE NUMBER WILL BE INCREMENTED FOR EACH REPETITION.)
OUTFIL FNAMES=OUT2,INCLUDE=(5,2,SS,EQ,CB2,C5,M3),
Default for REPEAT: None; must be specified.
SPLIT
SPLIT
Splits the output records one record at a time in rotation among the data sets
of this OUTFIL group until all of the output records have been written. As a
result, the records will be split as evenly as possible among all of the data sets
in the group.
As an example, for an OUTFIL group with three data sets:
v the first OUTFIL data set in the group will receive records 1, 4, 7, and so on.
v the second OUTFIL data set in the group will receive records 2, 5, 8, and so
on.
v the third OUTFIL data set in the group will receive records 3, 6, 9, and so
on.
335

The records are not contiguous in each OUTFIL data set.
SPLIT is equivalent to SPLITBY=1.
The SPLIT parameter cannot be used with any of the following parameters:
and NODETAIL.
Sample Syntax:
* WRITE RECORD 1 TO PIPE1, RECORD 2 TO PIPE2, RECORD 3 TO PIPE3,
* RECORD 4 TO PIPE4, RECORD 5 TO PIPE1, RECORD 6 TO PIPE2, AND SO ON.
OUTFIL FNAMES=(PIPE1,PIPE2,PIPE3,PIPE4),SPLIT
* SPLIT THE INCLUDED AND REFORMATTED OUTPUT RECORDS EVENLY BETWEEN
* TAPE1 AND TAPE2.
OUTFIL FNAMES=(TAPE1,TAPE2),SPLIT,
INCLUDE=(8,2,ZD,EQ,27),OUTREC=(5X,1,75)
Default for SPLIT: None; must be specified.
SPLITBY
SPLITBY=n
Splits the output records n records at a time in rotation among the data sets of
this OUTFIL group until all of the output records have been written.
As an example, if SPLITBY=10 is specified for an OUTFIL group with three
data sets:
v the first OUTFIL data set in the group will receive records 1-10, 31-40, and
so on.
v the second OUTFIL data set in the group will receive records 11-20, 41-50,
and so on.
v the third OUTFIL data set in the group will receive records 21-30, 51-60, and
so on.
The records are not contiguous in each OUTFIL data set.
SPLITBY=1 is equivalent to SPLIT.
The SPLITBY parameter cannot be used with any of the following parameters:
and NODETAIL.
n
specifies the number of records to split by. The value for n starts at 1 and
is limited to 28 digits (15 significant digits).
Sample Syntax:
* WRITE RECORDS 1-5 TO PIPE1, RECORDS 6-10 TO PIPE2, RECORDS 11-15 TO
* PIPE3, RECORDS 16-20 TO PIPE4, RECORDS 21-25 TO PIPE1, RECORDS 26-30
* TO PIPE2, AND SO ON.
OUTFIL FNAMES=(PIPE1,PIPE2,PIPE3,PIPE4),SPLITBY=5
* SPLIT THE INCLUDED AND REFORMATTED OUTPUT RECORDS 100 AT A TIME
* BETWEEN TAPE1 AND TAPE2.
OUTFIL FNAMES=(TAPE1,TAPE2),SPLITBY=100,
Default for SPLITBY: None; must be specified.

SPLIT1R
336

SPLIT1R=n
Splits the output records n records at a time for one rotation among the data
sets of this OUTFIL group until all of the output records have been written.
As an example, if SPLIT1R=10 is specified for an input data set with 35 records
and an OUTFIL group with three data sets:
v the first OUTFIL data set in the group will receive records 1-10.
v the second OUTFIL data set in the group will receive records 11-20.
v the third OUTFIL data set in the group will receive records 21-35.
The records are contiguous in each OUTFIL data set.
The SPLIT1R parameter cannot be used with any of the following parameters:
and NODETAIL.
n
specifies the number of records to split by. The value for n starts at 1 and
is limited to 28 digits (15 significant digits).
Sample Syntax:
* WRITE RECORDS 1-20 TO PIPE1, RECORDS 21-40 TO PIPE2,
* RECORDS 41-60 TO PIPE3 AND RECORDS 61-85 TO PIPE4.
OUTFIL FNAMES=(PIPE1,PIPE2,PIPE3,PIPE4),SPLIT1R=20
* SPLIT THE INCLUDED AND REFORMATTED OUTPUT RECORDS ONCE
* CONTIGUOUSLY BETWEEN TAPE1 AND TAPE2.
OUTFIL FNAMES=(TAPE1,TAPE2),SPLIT1R=100,
Default for SPLIT1R: None; must be specified.
NULLOFL
NULLOFL=
RC0
RC4
RC16
specifies the action to be taken by DFSORT when there are no data records for
a data set associated with this OUTFIL statement, as indicated by a DATA
count of 0 in message ICE227I. OUTFIL report records have no affect on the
action taken as a result of this option.
RC0
0, and continue processing when there are no data records for a data set
associated with this OUTFIL statement.
RC4
4, and continue processing when there are no data records for the a data
set associated with this OUTFIL statement.
RC16
a return code of 16 when there are no data records for a data set associated
with this OUTFIL statement.
Default for NULLOFL: The NULLOFL installation default.
337

Note:
1. The NULLOFL value specified for each OUTFIL statement applies to the
data sets associated with that OUTFIL statement. If a NULLOFL value is
not specified for an OUTFIL statement, the NULLOFL installation default
value applies to the data sets associated with that OUTFIL statement. For
example, if the installation default is NULLOFL=RC0 (IBM's shipped
default) and the following is specified:
OUTFIL FNAMES=OUT1,NULLOFL=RC16,INCLUDE=(5,3,CH,EQ,CD01)
OUTFIL FNAMES=(OUT2,OUT3),INCLUDE=(5,3,CH,EQ,CD02)
OUTFIL FNAMES=OUT4,NULLOFL=RC4,SAVE
then NULLOFL=RC16 applies to the data set for OUT1, NULLOFL=RC0

(the installation default) applies to the data sets for OUT2 and OUT3, and
NULLOFL=RC4 applies to the data set for OUT4.
2. If you receive message ICE174I or ICE209A, a DATA count of 0 in message
ICE227I identifies an OUTFIL data set for which there are no data records.
3. The return code of 0 or 4 resulting from NULLOFL=RC0 or
NULLOFL=RC4, respectively, can be overridden by a higher return code set
for some other reason, including a return code of 16 resulting from
NULLOFL=RC16.
4. NULLOUT applies to the SORTOUT data set. NULLOFL applies to OUTFIL
data sets
LINES
LINES=n
Specifies the number of lines per page to be used for the reports produced for
this OUTFIL group. DFSORT uses ANSI carriage control characters to control
page ejects and the placement of the lines in your report, according to your
specifications.
The LINES parameter cannot be used with the IFTRAIL parameter.
n
specifies the number of lines per page. The value for n must be between 1
and 255. However, n--or the default for n if LINES is not specified--must
be greater than or equal to the number of lines needed for each of the
following:
v The HEADER1 lines
v The TRAILER1 lines
v The sum of all lines for HEADER2, TRAILER2, HEADER3s, TRAILER3s,
and the data lines and blank lines produced from an input record.
Sample Syntax:
OUTFIL FNAMES=RPT1,LINES=50
Default for LINES: None; must be specified, unless HEADER1, TRAILER1,

HEADER2, TRAILER2, SECTIONS, or NODETAIL is specified, in which case
the default for LINES is 60.
HEADER1
338

,
HEADER1=(
c:
r
p,m
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
&PAGE
PAGE=( edit
)
to
&PAGE=(
edit
to
Specifies the report header to be used for the reports produced for this
OUTFIL group. The report header appears by itself as the first page of the
report. DFSORT uses ANSI carriage control characters to control page ejects
and the placement of the lines in your report, according to your specifications.
You can use BLKCCH1 to replace the '1' (page eject) for the ANSI carriage
control character in the first line of the report header with a blank, thus
avoiding forcing a page eject. You can use REMOVECC to remove the ANSI
carriage control characters from a report.
You can choose to include any or all of the following report elements in your
report header:
v Blanks, character strings, and hexadecimal strings
v Unedited input fields from the first OUTFIL input record
v Current date in a variety of different forms
v Current time in a variety of different forms
v Page number, converted to different numeric formats, or edited to contain
signs, decimal points, leading zeros or no leading zeros, and so on.
The HEADER1 parameter cannot be used with the IFTRAIL parameter.
The report header consists of the elements you select, in the order in which
you specify them, and in the columns or lines you specify.
c: specifies the column in which the first position of the associated report
element is to appear, relative to the start of the data in the report record.
Ignore the RDW (variable-length report records only) and carriage control
character when specifying c:. That is, 1: indicates the first byte of the data
in the report record for both fixed-length and variable-length report
records.
Unused space preceding the specified column is padded with EBCDIC
blanks. The following rules apply:
v c: must be followed by a report element, but must not precede / or n/.
339

v c must not overlap the previous report element in the report record.
r
specifies that blanks, a character string, or a hexadecimal string are to

appear in the report record, or that a new report record is to be started in
the header, with or without intervening blank lines. These report elements
can be specified before or after any other report elements. Consecutive
report elements can be specified. Permissible values are nX, n'xx...x',
nC'xx...x', nX'yy...yy', /.../ and n/.
nX
Blanks. n bytes of EBCDIC blanks (X'40') are to appear in the

report record. n can range from 1 to 4095. If n is omitted, 1 is used.
n'xx...x'
Character string. n repetitions of the character string constant
('xx...x') are to appear in the report record. n can range from 1 to
4095. If n is omitted, 1 is used. x can be any EBCDIC character. You
can specify 1 to 256 characters.
nC'xx...x' can be used instead of n'xx...x'.
If you want to include a single apostrophe in the character string,
you must specify it as two single apostrophes:
Required:
ONEILL
Specify:
ONEILL or CONEILL
nX'yy...yy'
Hexadecimal string. n repetitions of the hexadecimal string
constant (X'yy...yy') are to appear in the report record. n can range
from 1 to 4095. If n is omitted, 1 is used.
specify from 1 to 256 pairs of hexadecimal digits
/.../ or n/
Blank lines or a new line. A new report record is to be started in
the header with or without intervening blank lines. If /.../ or n/ is
specified at the beginning or end of the header, n blank lines are to
appear in the header. If /.../ or n/ is specified in the middle of the
header, n-1 blank lines are to appear in the header (thus, / or 1/
indicates a new line with no intervening blank lines).
Either n/ (for example, 5/) or multiple /'s (for example, /////)
can be used. n can range from 1 to 255. If n is omitted, 1 is used.
As an example, if you specify:
OUTFIL HEADER1=(2/,First line of text,/,
Second line of text,2/,
Third line of text,2/)
the report header appears as follows when printed:

blank line
blank line
First line of text
Second line of text
blank line
Third line of text
blank line
blank line
340

p,m
specifies that an unedited input field, from the first OUTFIL input record
for which a data record appears in the report, is to appear in the report
record.
p
relative position 5, because the first four bytes are occupied by the
RDW. All fields must start on a byte boundary, and no field can extend
beyond byte 32752. See OUTFIL statements notes on page 374 for
special rules concerning variable-length records.
specifies the length in bytes of the input field. The value for m must be
DATE
specifies that the current date is to appear in the report record in the form
'mm/dd/yy', where mm represents the month (01-12), dd represents the
day (01-31), and yy represents the last two digits of the year (for example,
95).
&DATE
DATE=(abcd)
'adbdc', where a, b, and c indicate the order in which the month, day, and
year are to appear and whether the year is to appear as two or four digits,
and d is the character to be used to separate the month, day and year.
For a, b, and c, use M to represent the month (01-12), D to represent the
day (01-31), Y to represent the last two digits of the year (for example, 05),
or 4 to represent the four digits of the year (for example, 2005). M, D, and
Y or 4 can each be specified only once. Examples: DATE=(DMY.) would
produce a date of the form 'dd.mm.yy', which on March 29, 2005, would
appear as '29.03.05'. DATE=(4MD-) would produce a date of the form
'yyyy-mm-dd', which on March 29, 2005, would appear as '2005-03-29'.
a, b, c, and d must be specified.
&DATE=(abcd)
DATENS=(abc)
'abc', where a, b and c indicate the order in which the month, day, and
year are to appear and whether the year is to appear as two or four digits.
For a, b and c, use M to represent the month (01-12), D to represent the
day (01-31), Y to represent the last two digits of the year (for example, 02),
or 4 to represent the four digits of the year (for example, 2002). M, D, and
Y or 4 can each be specified only once. Examples: DATENS=(DMY) would
produce a date of the form 'ddmmyy', which on March 29, 2002, would
appear as '290302'. DATENS=(4MD) would produce a date of the form
'yyyymmdd', which on March 29, 2002, would appear as '20020329'.
&DATENS=(abc)
341

YDDD=(abc)
'acb', where a and b indicate the order in which the year and day of the
year are to appear and whether the year is to appear as two or four digits,
and c is the character to be used to separate the year and day of the year.
For a and b, use D to represent the day of the year (001-D366), Y to
represent the last two digits of the year (for example, 04), or 4 to represent
the four digits of the year (for example, 2004). D, and Y or 4 can each be
specified only once. Examples: YDDD=(DY-) would produce a date of the
form 'ddd-yy' which on April 7, 2004, would appear as '098-04'.
YDDD=(4D/) would produce a date of the form 'yyyy/ddd' which on
April 7, 2004, would appear as '2004/098'.
&YDDD=(abc)
YDDDNS=(ab)
'ab', where a and b indicate the order in which the year and day of the
year are to appear and whether the year is to appear as two or four digits.
represent the last two digits of the year (for example, 04), or 4 to represent
the four digits of the year (for example, 2004). D, and Y or 4 can each be
specified only once. Examples: YDDDNS=(DY) would produce a date of
the form 'dddyy' which on April 7, 2004, would appear as '09804'.
YDDDNS=(4D) would produce a date of the form 'yyyyddd' which on
April 7, 2004, would appear as '2004098'.
a and b must be specified.
&YDDDNS=(ab)
TIME
specifies that the current time is to appear in the report record in the form
'hh:mm:ss', where hh represents the hour (00-23), mm represents the
minutes (00-59), and ss represents the seconds (00-59).
&TIME
TIME=(abc)
'hhcmmcss' (24-hour time) or 'hhcmmcss xx' (12-hour time).
If ab is 24, the time is to appear in the form 'hhcmmcss' (24-hour time)
where hh represents the hour (00-23), mm represents the minutes (00-59),
ss represents the seconds (00-59), and c is the character used to separate
the hours, minutes, and seconds. Example: TIME=(24.) would produce a
time of the form 'hh.mm.ss' which at 08:25:13 pm would appear as
'20.25.13'.
If ab is 12, the time is to appear in the form 'hhcmmcss xx' (12-hour time)
ss represents the seconds (00-59), xx is am or pm, and c is the character
used to separate the hours, minutes, and seconds. Example: TIME=(12.)
would produce a time of the form 'hh.mm.ss xx' which at 08:25:13 pm
would appear as '08.25.13 pm'.
342

ab and c must be specified.
&TIME=(abc)
TIMENS=(ab)
'hhmmss' (24-hour time) or 'hhmmss xx' (12-hour time).
If ab is 24, the time is to appear in the form 'hhmmss' (24-hour time) where
hh represents the hour (00-23), mm represents the minutes (00-59), and ss
represents the seconds (00-59). Example: TIMENS=(24) would produce a
time of the form 'hhmmss' which at 08:25:13 pm would appear as '202513'.
If ab is 12, the time is to appear in the form 'hhmmss xx' (12-hour time)
and ss represents the seconds (00-59). Example: TIMENS=(12) would
produce a time of the form 'hhmmss xx' which at 08:25:13 pm would
appear as '082513 pm'.
ab must be specified.
&TIMENS=(ab)
PAGE
specifies that the page number is to appear in the report record. The page
number for the report header appears as '
1'.
If HEADER1 is specified with PAGE, PAGE for the report header (first
page) will be '
1' and PAGE for the next page (second page) will be
'
2'. If HEADER1 is specified without PAGE, PAGE for the page after
the report header (second page) will be '
1' (typical of a report with a
cover sheet).
&PAGE
can be used instead of PAGE.
PAGE=(edit) or PAGE=(to)
same as PAGE except that the 15-digit page number appears edited or
converted as specified. See p,m,f,edit under OUTREC for further details on
the edit fields you can use. See p,m,f,to under OUTREC for further details
on the to fields you can use.
&PAGE=(edit) or &PAGE=(to)
can be used instead of PAGE=(edit) or PAGE=(to).
Sample Syntax:
OUTFIL FNAMES=(RPT1,RPT2),
HEADER1=(30:January Report,4/,
28:Prepared on ,DATE,//,
32:at ,TIME,//,
28:using DFSORTS OUTFIL,5/,
10:Department: ,12,8,50:Page:,PAGE=(EDIT=(TTT)))
Default for HEADER1: None; must be specified.

TRAILER1
343

,
TRAILER1= (
c:
r
p,m
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
&PAGE
PAGE=( edit )
to
&PAGE=(
edit
to
COUNT
COUNT15
COUNT=( edit
)
to
COUNT+n=( edit
to
COUNT-n=( edit
to
SUBCOUNT
SUBCOUNT15
SUBCOUNT=( edit
to
TOTAL=
(p,m,f
TOT=
MIN=
(p,m,f
)
)
)
)
,edit
,to
)
,edit
,to
MAX=
(p,m,f
)
,edit
,to
AVG=
(p,m,f
)
,edit
,to
Additional Parameters for TRAILER1
Additional Parameters for TRAILER1
344

SUBTOTAL=
(p,m,f
SUBTOT=
,edit
SUB=
,to
SUBMIN= (p,m,f
)
,edit
,to
SUBMAX= (p,m,f
)
,edit
,to
SUBAVG= (p,m,f
)
,edit
,to
Specifies the report trailer to be used for the reports produced for this OUTFIL
group. The report trailer appears by itself as the last page of the report.
placement of the lines in your report, according to your specifications. You can
use BLKCCT1 to replace the '1' (page eject) for the ANSI carriage control
character in the first line of the report trailer with a blank, thus avoiding
forcing a page eject. You can use REMOVECC to remove the ANSI carriage
control characters from a report.
report trailer:
v Unedited input fields from the last OUTFIL input record
signs, decimal points, leading zeros or no leading zeros, and so on
v Any or all of the following statistics:
Count of data records in the report, converted to different numeric
formats, or edited to contain signs, decimal points, leading zeros or no
leading zeros, and so on. You can add a decimal number to the count
before converting or editing it (for example, add 1 to account for writing
a trailer record, or add 2 to account for writing a header and trailer
record).
Total, minimum, maximum, or average for each specified ZD, PD, BI, FI,
FL, CSF, FS, UFF, or SFF numeric input field in the data records of the
report, converted to different numeric formats, or edited to contain signs,
The TRAILER1 parameter cannot be used with the IFTRAIL parameter.
The report trailer consists of the elements you select, in the order in which you
specify them, and in the columns or lines you specify.
c: See c: under HEADER1.
r
specifies that blanks, a character string, or a hexadecimal string are to

appear in the report record, or that a new report record is to be started in
the trailer, with or without intervening blank lines. These report elements
can be specified before or after any other report elements. Consecutive
report elements can be specified. Permissible values are nX, n'xx...x',
nC'xx...x', nX'yy...yy', /.../, and n/.
nX
Blanks. See nX under r for HEADER1.
n'xx...x'
345

Character string. See n'xx...x' under r for HEADER1. nC'xx...x' can
be used instead of n'xx...x'
nX'xx...x'
Hexadecimal string. See nX'xx...x' under r for HEADER1.
/.../ or n/
Blank lines or a new line. A new report record is to be started in
the trailer, with or without intervening blank lines. If /.../ or n/ is
specified at the beginning or end of the trailer, n blank lines are to
appear in the trailer. If /.../ or n/ is specified in the middle of the
trailer, n-1 blank lines are to appear in the trailer (thus, / or 1/
indicates a new line with no intervening blank lines).
Either n/ (for example, 5/) or multiple /'s (for example, /////)
can be used. n can range from 1 to 255. If n is omitted, 1 is used.
p,m
specifies that an unedited input field, from the last OUTFIL input record
for which a data record appears in the report, is to appear in the report
record.
p
See p under HEADER1.
See m under HEADER1.
DATE
See DATE under HEADER1.
&DATE
DATE=(abcd)
See DATE=(abcd) under HEADER1.
&DATE=(abcd)
DATENS=(abc)
See DATENS=(abc) under HEADER1.
&DATENS=(abc)
YDDD=(abc)
See YDDD=(abc) under HEADER1.
&YDDD=(abc)
YDDDNS=(ab)
See YDDDNS=(ab) under HEADER1.
&YDDDNS=(ab)
TIME
See TIME under HEADER1.
&TIME
TIME=(abc)
See TIME=(abc) under HEADER1.
346

&TIME=(abc)
TIMENS=(ab)
See TIMENS=(ab) under HEADER1.
&TIMENS=(ab)
PAGE
specifies that the current page number is to appear in the report record.
The page number for the trailer appears as 6 digits, right-justified, with
leading zeros suppressed. For example, if the page is numbered 12, it
appears as '
12'.
&PAGE
COUNT
specifies that the count of data records in the report is to appear in the
report record as 8 digits, right-justified, with leading zeros suppressed. For
example, if there are 6810 input records in the report, the count appears as
'
6810'.
If slash (/) is used in OUTREC or BUILD to produce multiple data records,
COUNT only counts the number of data records processed as input to
OUTREC or BUILD. For example, if OUTREC processes 3 input records
and creates 2 output records for each input record, the count is 3, not 6
COUNT15
same as COUNT except that the count appears as 15 digits.
COUNT=(edit) or COUNT=(to)
same as COUNT except that the 15digit count appears edited or
on the to fields you can use. For example, if there are 6810 input records,
COUNT=(M11,LENGTH=6) produces a count of '006810'.
COUNT+n=(edit) or COUNT+n=(to)
same as COUNT=(edit) or COUNT=(to) except that n is added to the
15-digit count before it is edited or converted. n can be 1 to 3 decimal
digits. For example, if there are 6810 input records,
COUNT+2=(M11,LENGTH=6) produces a count of '006812'. One important
use for this parameter is to add 1 for the TRAILER1 record to the count of
data records.
COUNT-n=(edit) or COUNT-n=(to)
same as COUNT=(edit) or COUNT=(to) except that n is subtracted from
the 15-digit count before it is edited or converted. n can be 1 to 3 decimal
digits. For example, if there are 6810 input records, COUNT-
347

1=(M11,LENGTH=6) produces a count of '006809'. One important use for
this parameter is to subtract 1 for a header record from the count of all
records.
SUBCOUNT
specifies that the running count of input records in the report is to appear
in the report record as 8 digits, right-justified, with leading zeros
suppressed.
For TRAILER1, the running count is the same as the count, so SUBCOUNT
produces the same value as COUNT.
SUBCOUNT counts only the number of data records processed as input to
and creates 2 output records for each input record, the running count is 3,
not 6.
SUBCOUNT15
same as SUBCOUNT except that the running count appears as 15 digits.
SUBCOUNT=(edit) or SUBCOUNT=(to)
same as SUBCOUNT except that the 15digit running count appears edited
or converted as specified. See p,m,f,edit under OUTREC for further details
on the edit fields you can use. See p,m,f,to under OUTREC for further
details on the to fields you can use.
TOTAL
specifies that an edited or converted total, for the values of a numeric
input field in all data records of the report, is to appear in the report
record.
TOT can be used instead of TOTAL.
p,m,f,edit or p,m,f,to
specifies the numeric input field for which the total is to be produced
and how the output field (that is, the total) is to be edited or
converted.
See p,m,f,edit under OUTREC for further details. However, note that
PD0, DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3,
TC4, TE1, TE2, TE3, TE4, TM1, TM2, TM3, and TM4 are not allowed
for TOTAL and that for TOTAL, the default number of digits (d) used
for editing or conversion is as follows:
Table 59. Digits for TOTAL Fields
348
Format (f)
Length (m)
Digits (d)
ZD
1-15
15
ZD
16-31
31
PD
1-8
15
PD
9-16
31
BI
1-4
10
BI
5-8
20
FI
1-4
10
FI
5-8
20
FL
4 or 8
20
CSF or FS
1-15
15

Table 59. Digits for TOTAL Fields (continued)
Format (f)
Length (m)
Digits (d)
CSF or FS
16-32
31
UFF
1-15
15
UFF
16-44
31
SFF
1-15
15
SFF
16-44
31
If EDIT or EDxy is specified, the number of digits in the pattern (I's

and T's) is used.
MIN
specifies that an edited or converted minimum, for the values of a numeric
record.
specifies the numeric input field for which the minimum is to be
produced and how the output field (that is, the minimum) is to be
edited or converted.
See p,m,f,edit or p,m,f,to under OUTREC for further details. However,
note that PD0, DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1,
TC2, TC3, TC4, TE1, TE2, TE3, TE4, TM1, TM2, TM3 and TM4 are not
allowed for MIN.
MAX
specifies that an edited or converted maximum, for the values of a numeric
record.
specifies the numeric input field for which the maximum is to be
produced and how the output field (that is, the maximum) is to be
allowed for MAX.
AVG
specifies that an edited or converted average, for the values of a numeric
record. The average (or mean) is calculated by dividing the total by the
count and rounding down to the nearest integer. For example:
+2305 / 152 = +15
-2305 / 152 = -15
specifies the numeric input field for which the average is to be
produced and how the output field (that is, the average) is to be edited
or converted.
allowed for AVG.
349

SUBTOTAL
specifies that an edited or converted running total, for the values of a
numeric input field in all data records of the report, is to appear in the
report record.
SUBTOT or SUB can be used instead of SUBTOTAL.
For TRAILER1, the running total is the same as the total, so SUBTOTAL
produces the same value as TOTAL.
specifies the numeric input field for which the running total is to be
produced and how the output field (that is, the running total) is to be
See p,m,f,edit or p,m,f,to under TOTAL for further details.
SUBMIN
specifies that an edited or converted running minimum, for the values of a
report record.
For TRAILER1, the running minimum is the same as the minimum, so
SUBMIN produces the same value as MIN.
specifies the numeric input field for which the running minimum is to
be produced and how the output field (that is, the running minimum)
is to be edited or converted.
allowed for SUBMIN.
SUBMAX
specifies that an edited or converted running maximum, for the values of a
report record.
For TRAILER1, the running maximum is the same as the maximum, so
SUBMAX produces the same value as MAX.
specifies the numeric input field for which the running maximum is to
be produced and how the output field (that is, the running maximum)
is to be edited or converted.
allowed for SUBMAX.
SUBAVG
specifies that an edited or converted running average, for the values of a
report record.
For TRAILER1, the running average is the same as the average, so
SUBAVG produces the same value as AVG.
350

specifies the numeric input field for which the running average is to be
produced and how the output field (that is, the running average) is to
be edited or converted.
allowed for SUBAVG.
Sample Syntax:
OUTFIL FNAMES=RPT1,
TRAILER1=(5/,
10:Summary of Report for Division Revenues,3/,
10:Number of divisions reporting: ,COUNT,2/,
10:Total revenue: ,TOTAL=(25,5,PD,M5),2/,
10:Lowest revenue: ,MIN=(25,5,PD,M5),2/,
10:Highest revenue: ,MAX=(25,5,PD,M5),2/,
10:Average revenue: ,AVG=(25,5,PD,M5))
OUTFIL FNAMES=RPT2,BLKCCT1,
TRAILER1=(/,10:Grand Total: ,TOTAL=(25,5,PD,M5))
Default for TRAILER1: None; must be specified.

HEADER2
,
HEADER2= (
c:
r
p,m
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
&PAGE
PAGE=( edit
)
to
&PAGE=(
edit
to
Specifies the page header to be used for the reports produced for this OUTFIL
group. The page header appears at the top of each page of the report, except
for the report header page (if any) and report trailer page (if any). DFSORT
uses ANSI carriage control characters to control page ejects and the placement
of the lines in your report, according to your specifications. You can use
BLKCCH2 to replace the '1' (page eject) for the ANSI carriage control character
in the first line of the first page header with a blank, thus avoiding forcing a
351

page eject. You can use REMOVECC to remove the ANSI carriage control
characters from a report.
page header:
v Unedited input fields from the first OUTFIL input record for which a data
record appears on the page
The HEADER2 parameter cannot be used with the IFTRAIL parameter.
The page header consists of the elements you select, in the order in which you
r
See r under HEADER1.
p,m
for which a data record appears on the page, is to appear in the report
record. See p,m under HEADER1 for further details.
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
352

&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
specifies that the current page number is to appear in the OUTFIL report
record. The page number for the header appears as 6 digits, right-justified,
with leading zeros suppressed. For example, if the page is numbered 3, it
appears as '
3'.
If HEADER1 is specified with PAGE and HEADER2 is specified with
PAGE, the page number for the first page header will be '
2'. If
HEADER1 is not specified or is specified without PAGE and HEADER2 is
specified with PAGE, the page number for the first page header will be
'
1'.
&PAGE
Sample Syntax:
OUTFIL FNAMES=STATUS,
HEADER2=(5:Page ,PAGE, of Status Report for ,DATE=(MD4/),
at ,TIME=(12:),2/,
10:Item ,20:Status
,35:Count,/,
10:-----,20:------------,35:-----),
OUTREC=(10:6,5,
20:14,1,CHANGE=(12,
CS,CShip,
CH,CHold,
CT,CTransfer),
NOMATCH=(C*Check Code*),
36:39,4,ZD,M10,
132:X)
Default for HEADER2: None; must be specified.

TRAILER2
353

,
TRAILER2= (
c:
r
p,m
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
&PAGE
PAGE=( edit )
to
&PAGE=(
edit
to
COUNT
COUNT15
COUNT=( edit )
to
COUNT+n=( edit
to
COUNT-n=( edit
to
SUBCOUNT
SUBCOUNT15
SUBCOUNT=( edit
to
TOTAL=
(p,m,f
TOT=
MIN=
(p,m,f
)
)
)
)
,edit
,to
)
,edit
,to
MAX=
(p,m,f
)
,edit
,to
AVG=
(p,m,f
)
,edit
,to
(p,m,f
SUBTOTAL=
SUBTOT=
SUB=
SUBMIN= (p,m,f
)
,edit
,to
)
,edit
,to
SUBMAX=
(p,m,f
)
,edit
,to
SUBAVG=
(p,m,f
)
,edit
,to
Specifies the page trailer to be used for the reports produced for this OUTFIL
group. The page trailer appears at the very bottom of each page of the report
(as specified or defaulted by the LINES value), except for the report header
354

page (if any) and report trailer page (if any). DFSORT uses ANSI carriage
control characters to control page ejects and the placement of the lines in your
report, according to your specifications.
page trailer:
v Unedited input fields from the last OUTFIL input record for which a data
record appears on the page
Count of data records on the page, converted to different numeric
record).
FL, CSF, FS, UFF, or SFF numeric input field in the data records on the
page, converted to different numeric formats, or edited to contain signs,
Running total, minimum, maximum, or average for each specified ZD,
PD, BI, FI, FL, CSF, FS, UFF, or SFF numeric input field in the data
records up to this point, converted to different numeric formats, or edited
to contain signs, decimal points, leading zeros or no leading zeros, and so
on.
The TRAILER2 parameter cannot be used with the IFTRAIL parameter.
The page trailer consists of the elements you select, in the order in which you
r
See r under TRAILER1.
p,m
for which a data record appears on the page, is to appear in the report
record. See p,m under TRAILER1 for further details.
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
355

&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
See PAGE under TRAILER1.
&PAGE
See PAGE=(edit) or PAGE=(to) under TRAILER1.
COUNT
specifies that the count of data records on the page is to appear in the
example, if page 1 has 40 input records, page 2 has 40 input records, and
page 3 has 26 input records, COUNT will show '
40' for page 1,
'
40' for page 2, and '
26' for page 3.
COUNT counts only the number of data records processed as input to
and creates 2 output records for each input record, the count is 3, not 6.
COUNT15
356

digits.
digits.
SUBCOUNT
specifies that the count of input records up to this point in the report is to
appear in the report record as 8 digits, right-justified, with leading zeros
suppressed. The running count accumulates the count for all pages up to
and including the current page. For example, if page 1 has 40 input
records, page 2 has 40 input records, and page 3 has 26 input records,
SUBCOUNT will show '
40' for page 1, '
80' for page 2, and
'
106' for page 3.
and creates 2 output records for each input record, the running count is 3,
not 6.
SUBCOUNT15
TOTAL
input field in the data records on the page, is to appear in the report
record.
See p,m,f,edit or p,m,f,to under TOTAL for TRAILER1.
MIN
record.
See p,m,f,edit or p,m,f,to under MIN for TRAILER1.
MAX
record.
See p,m,f,edit or p,m,f,to under MAX for TRAILER1.
357

AVG
record.
See p,m,f,edit or p,m,f,to under AVG for TRAILER1.
SUBTOTAL
numeric input field in the data records up to this point in the report, is to
appear in the report record. The running total accumulates the total for all
pages up to and including the current page. For example, if the total for a
selected numeric field is +200 for page 1, -250 for page 2, and +90 for page
3, SUBTOTAL will be +200 for page 1, -50 for page 2, and +40 for page 3.
See p,m,f,edit or p,m,f,to under SUBTOTAL for TRAILER1.
SUBMIN
appear in the report record. The running minimum selects the minimum
from all pages up to and including the current page. For example, if the
minimum for a selected numeric field is +200 for page 1, -250 for page 2,
and +90 for page 3, SUBMIN will be +200 for page 1, -250 for page 2, and
-250 for page 3.
See p,m,f,edit or p,m,f,to under SUBMIN for TRAILER1.
SUBMAX
appear in the report record. The running maximum selects the maximum
from all pages up to and including the current page. For example, if the
maximum for a selected numeric field is -100 for page 1, +250 for page 2,
and +90 for page 3, SUBMAX will be -100 for page 1, +250 for page 2, and
+250 for page 3.
See p,m,f,edit or p,m,f,to under SUBMAX for TRAILER1.
SUBAVG
appear in the report record. The running average computes the average for
all pages up to and including the current page. For example, if the count of
data records and total for a selected numeric field are 60 and +2205 for
page 1, respectively, 60 and -6252 for page 2, respectively, and 23 and -320
for page 3, respectively, SUBAVG will be +36 for page 1, -33 for page 2,
and -30 for page 3.
See p,m,f,edit or p,m,f,to under SUBAVG for TRAILER1.
Sample Syntax:
OUTFIL FNAMES=STATS,
STARTREC=3,
OUTREC=(20:23,3,PD,M16,
358

30:40,3,PD,M16,
80:X),
TRAILER2=(/,2:Average on page:,
20:AVG=(23,3,PD,M16),
30:AVG=(40,3,PD,M16),/,
2:Average so far:,
20:SUBAVG=(23,3,PD,M16),
30:SUBAVG=(40,3,PD,M16))
Default for TRAILER2: None; must be specified.

SECTIONS
,
,
SECTIONS= ( p,m
)
SKIP=
P
L
nL
,
HEADER3= ( field
)
,PAGEHEAD
,
TRAILER3= ( field
Specifies the section break processing to be used for the reports produced for
this OUTFIL group. A section break field divides the report into sets of
sequential OUTFIL input records with the same binary value for that field,
which result in corresponding sets of data records (that is, sections) in the
report. A break is said to occur when the binary value changes. Of course,
because a break can occur in any record, the data records of a section can be
split across pages in your report.
For each section break field you specify, you can choose to include any or all of
the following:
v A page eject between sections.
v Zero, one or more blank lines to appear between sections on the same page.
v A section header to appear before the first data record of each section and
optionally, at the top of each page. When a page header and section header
are both to appear at the top of a page, the section header will follow the
page header.
v A section trailer to appear after the last data record of each section. When a
page trailer and section trailer are both to appear at the bottom of a page,
the page trailer will follow the section trailer.
The SECTIONS parameter cannot be used with the IFTRAIL parameter.
placement of the lines in your report, according to your specifications.
If multiple section break fields are used, they are processed in first-to-last
order, in the same way they would be sorted by these fields. In fact, the input
data set is generally sorted by the section break fields, to group the records
with the same section break values together for the report. This sorting can be
done by the same application that produces the report or by a previous
application.
A break in section break field 1 results in a break in section break fields 2
through n. A break in section break 2 results in a break in section break fields 3
through n, and so on. The section headers appear before each section in
359

first-to-last order, whereas the section trailers appear in last-to-first order. For
example, if section break fields represented by B1 with header H3A and trailer
T3A, B2 with header H3B and trailer T3B, and B3 with header H3C and trailer
T3C are specified in order, the following can appear:
H3A (header for B1=1 section)
H3B (header for B2=1 section)
H3C (header for B3=1 section)
data records for B1=1, B2=1,
T3C (trailer for B3=1 section)
T3B (trailer for B2=1 section)
T3A (trailer for B1=1 section)
H3A (header for B1=2 section)
T3A (trailer for B1=2 section)
B3=1 (new B1, B2, and B3 section)

B3=2 (new B3 section)
B3=1 (new B2 and B3 section)
B3=0 (new B1, B2, and B3 section)

B3=1 (new B3 section)
p,m
specifies a section break field in the OUTFIL input records to be used to divide
the report into sections. Each set of sequential OUTFIL input records, with the
same binary value for the section break field, results in a corresponding set of
data records. Each such set of data records is treated as a section in the report.
A break is said to occur when the binary value changes.
p
relative position 5, because the first four bytes are occupied by the RDW.
All fields must start on a byte boundary, and no field can extend beyond
byte 32752. See OUTFIL statements notes on page 374 for special rules
specifies the length in bytes of the input field. The value for m must be
between 1 and 256.
You can specify any, all or none of SKIP, HEADER3 and TRAILER3 after p,m.
If you do not specify SKIP, HEADER3 or TRAILER3 after p,m, SKIP=0L is used
as the default.
SKIP
SKIP=
P
L
nL
Specifies, for reports produced for this OUTFIL group, that either:
v Each section for the associated section break field is to appear on a new
page, or
360

v Zero, one or more blank lines to appear after each section associated with
this section break field, when it is followed by another section on the same
page.
Thus, you can use SKIP to specify how sections will be separated from each
other.
P
specifies that each section is to appear on a new page.
specifies that one blank line is to appear between sections on the same
page. L is the same as 1L.
nL specifies that n blank lines are to appear between sections on the same
page. You can specify from 0 to 255 for n.
Sample Syntax:
OUTFIL FNAMES=(PRINT,TAPE),
SECTIONS=(10,20,SKIP=P,
42,10,SKIP=3L)
HEADER3
,
HEADER3= (
c:
r
p,m
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
&PAGE
PAGE=( edit
)
to
&PAGE=(
edit
to
Specifies the section header to be used with the associated section break field
for the reports produced for this OUTFIL group. The section header appears
before the first data record of each section. DFSORT uses ANSI carriage control
characters to control page ejects and the placement of the lines in your report,
according to your specifications.
section header:
v Unedited input fields from the first OUTFIL input record for which a data
record appears in the section
361

The section header consists of the elements you select, in the order in which
r
See r under HEADER1.
p,m
for which a data record appears in the section, is to appear in the report
record. See p,m under HEADER1 for further details.
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
can be used instead of YDDD=(ab).
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
specifies that the current page number is to appear in the OUTFIL report
362

record. The page number for the header appears as 6 digits, right-justified,
with leading zeros suppressed. For example, if the page is numbered 3, it
appears as '
3'.
&PAGE
Sample Syntax:
OUTFIL FNAMES=STATUS1,
HEADER2=(10:Status Report for all departments,5X,
- ,&PAGE, -),
SECTIONS=(10,8,
HEADER3=(2/,10:Report for department ,10,8, on ,&DATE,2/,
10:
Number,25:Average Time,/,
10:Completed,25:
(in days),/,
10:---------,25:------------)),
OUTREC=(10:21,5,ZD,M10,LENGTH=9,
25:38,4,ZD,EDIT=(III.T),LENGTH=12,
132:X)
PAGEHEAD
PAGEHEAD
Specifies that the section header to be used with the associated section break
field is to appear at the top of each page of the report, except for the report
header page (if any) and report trailer page (if any), as well as before each
section. If you do not specify PAGEHEAD, the section header appears only
before each section; so if a section is split between pages, the section header
appears only in the middle of the page. PAGEHEAD can be used when you
want HEADER3 to be used as a page header as well as a section header.
If PAGEHEAD is specified for a section break field for which HEADER3 is not
also specified, PAGEHEAD will not be used.
Sample Syntax:
OUTFIL FNAMES=STATUS2,
HEADER2=(10:Status Report for all departments,5X,
- ,&PAGE, -),
SECTIONS=(10,8,
HEADER3=(2/,10:Report for department ,10,8, on ,&DATE,2/,
10:
Number,25:Average Time,/,
10:Completed,25:
(in days),/,
10:---------,25:------------),
PAGEHEAD,SKIP=P),
OUTREC=(10:21,5,ZD,M10,LENGTH=9,
25:38,4,ZD,EDIT=(III.T),LENGTH=12,
132:X)
TRAILER3
363

,
TRAILER3= (
c:
r
p,m
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
&PAGE
PAGE=( edit )
to
&PAGE=(
edit
to
COUNT
COUNT15
COUNT=( edit )
to
COUNT+n=( edit
to
COUNT-n=( edit
to
SUBCOUNT
SUBCOUNT15
SUBCOUNT=( edit
to
TOTAL=
(p,m,f
TOT=
MIN=
(p,m,f
)
)
)
)
,edit
,to
)
,edit
,to
MAX=
(p,m,f
)
,edit
,to
AVG=
(p,m,f
)
,edit
,to
(p,m,f
SUBTOTAL=
SUBTOT=
SUB=
SUBMIN= (p,m,f
)
,edit
,to
)
,edit
,to
SUBMAX=
(p,m,f
)
,edit
,to
SUBAVG=
(p,m,f
)
,edit
,to
Specifies the section trailer to be used with the associated section break field
for the reports produced for this OUTFIL group. The section trailer appears
after the last data record of each section. DFSORT uses ANSI carriage control
characters to control page ejects and the placement of the lines in your report,
364

according to your specifications.
section trailer:
v Unedited input fields from the last OUTFIL input record for which a data
record appears in the section
Count of data records in the section, converted to different numeric
record).
FL, CSF, FS, UFF, or SFF numeric input field in the data records in the
section, converted to different numeric formats, or edited to contain signs,
decimal points, leading zeros or no leading zeros, and so on
Running total, minimum, maximum, or average for each specified ZD,
PD, BI, FI, FL, CSF, FS, UFF, or SFF numeric input field in the data
records up to this point, converted to different numeric formats, or edited
to contain signs, decimal points, leading zeros or no leading zeros, and so
on.
The section trailer consists of the elements you select, in the order in which
r
See r under TRAILER1.
p,m
for which a data record appears in the section, is to appear in the report
record. See p,m under TRAILER1 for further details.
DATE
&DATE
DATE=(abcd)
&DATE=(abcd)
DATENS=(abc)
&DATENS=(abc)
YDDD=(abc)
365

&YDDD=(abc)
YDDDNS=(ab)
&YDDDNS=(ab)
TIME
&TIME
TIME=(abc)
&TIME=(abc)
TIMENS=(ab)
&TIMENS=(ab)
PAGE
See PAGE under TRAILER1.
&PAGE
See PAGE=(edit) or PAGE=(to) under TRAILER1.
COUNT
specifies that the count of data records in the section is to appear in the
example, if the first section has 40 input records, the second section has 40
input records, and the third section has 26 input records, COUNT will
show '
40' for the first section, '
40' for the second section, and
'
26' for the third section.
COUNT counts only the number of data records processed as input to
OUTREC or BUILD. For example, if OUTREC or BUILD processes 3 input
records and creates 2 output records for each input record, the count is 3,
not 6.
COUNT15
366

digits.
digits.
SUBCOUNT
specifies that the running count of input records up to this point in the
report is to appear in the report record 8 digits, right-justified, with leading
zeros suppressed. The running count accumulates the count for all sections
up to and including the current section. For example, if the first section has
40 input records, the second section has 40 input records, and the third
section has 26 input records, SUBCOUNT will show '
40' for the first
section, '
80' for the second section, and '
106' for the third
section.
OUTREC or BUILD. For example, if OUTREC or BUILD processes 3 input
records and creates 2 output records for each input record, the running
count is 3, not 6
SUBCOUNT15
TOTAL
input field in the data records in the section, is to appear in the report
record.
See p,m,f,edit or p,m,f,to under TOTAL for TRAILER1.
MIN
record.
See p,m,f,edit or p,m,f,tounder MIN for TRAILER1.
MAX
record.
See p,m,f,edit or p,m,f,to under MAX for TRAILER1.
367

AVG
record.
See p,m,f,edit or p,m,f,to under AVG for TRAILER1.
SUBTOTAL
appear in the report record. The running total accumulates the total for all
sections up to and including the current section. For example, if the total
for a selected numeric field is +200 for the first section, -250 for the second
section and +90 for the third section, SUBTOTAL will be +200 for the first
section, -50 for the second section and +40 for the third section.
See p,m,f,edit or p,m,f,to under SUBTOTAL for TRAILER1.
SUBMIN
appear in the report record. The running minimum selects the minimum
from all sections up to and including the current section. For example, if
the minimum for a selected numeric field is +200 for the first section, -250
for the second section and +90 for the third section, SUBMIN will be +200
for the first section, -250 for the second section and -250 for the third
section.
See p,m,f,edit or p,m,f,to under SUBMIN for TRAILER1.
SUBMAX
appear in the report record. The running maximum selects the maximum
from all sections up to and including the current section. For example, if
the maximum for a selected numeric field is -100 for the first section, +250
for the second section and +90 for the third section, SUBMAX will be -100
for the first section, +250 for the second section and +250 for the third
section.
See p,m,f,edit or p,m,f,to under SUBMAX for TRAILER1.
SUBAVG
appear in the report record. The running average computes the average for
all sections up to and including the current section. For example, if the
count of data records and total for a selected numeric field are 60 and
+2205 for the first section, respectively, 60 and -6252 for the second section,
respectively, and 23 and -320 for the third section, respectively, SUBAVG
will be +36 for the first section, -33 for the second section and -30 for the
third section.
See p,m,f,edit or p,m,f,to under SUBAVG for TRAILER1.
368

Sample Syntax:
OUTFIL FNAMES=SECRPT,
INCLUDE=(11,4,CH,EQ,CSSD),
HEADER3=(2:Department: ,3,5,4X,Date: ,&DATE,2/),
TRAILER3=(2/,2:The average for ,3,5, is ,
AVG=(40,3,PD,M12),/,
2:The overall average so far is ,
SUBAVG=(40,3,PD,M12)),
45,8,SKIP=3L,
HEADER3=(4:Week Ending ,45,8,2/,
4:Item Number,20:Completed,/,
4:-,20:-),
TRAILER3=(4:The item count for week ending ,45,8,
is ,COUNT=(EDIT=(II,IIT)))),
OUTREC=(11:16,4,22:40,3,PD,M12,120:X)
Default for SECTIONS: None; must be specified.

NODETAIL
NODETAIL
Specifies that data records are not to be output for the reports produced for
this OUTFIL group. With NODETAIL, the data records are completely
processed with respect to input fields, statistics, counts, sections breaks, and so
on, but are not written to the OUTFIL data set and are not included in line
counts for determining the end of a page. You can use NODETAIL to
summarize the data records without actually showing them.
The NODETAIL parameter cannot be used with the IFTRAIL parameter.
Sample Syntax:
OUTFIL FNAMES=SUMMARY,NODETAIL,
HEADER2=( Date: ,DATENS=(DMY.),4X,Page: ,PAGE,2/,
10:Division,25:
Total Revenue,/,
10:--------,25:-----------------),
SECTIONS=(3,5,
TRAILER3=(10:3,5,
25:TOTAL=(25,4,FI,M19,
LENGTH=17))),
TRAILER1=(5/,10:Summary of Revenue ,4/,
12:Number of divisions reporting is ,
COUNT,/,
12:Total revenue is ,
TOTAL=(25,4,FI,M19))
Default for NODETAIL: None; must be specified.

Default for OUTFIL Statements: None; must be specified. Multiple OUTFIL
statements can be specified in the same and different sources; override is at the
ddname level.
Applicable Functions for OUTFIL Statements: Sort, merge, and copy.
BLKCCH1
specifies that the ANSI carriage control character of '1' (page eject) in the first
line of the report header (HEADER1) is to be replaced with a blank. You can
use BLKCCH1 to avoid forcing a page eject at the start of the report header. If
BLKCCH1 is specified without HEADER1, it will not be used.
Sample Syntax:
OUTFIL FNAMES=RPT1,BLKCCH1,
HEADER1=(30:January Report,4/)
Default for BLKCCH1: None; must be specified.

369

BLKCCH2
line of the first page header (HEADER2) is to replaced with a blank. You can
use BLKCCH2 to avoid forcing a page eject at the start of the first page header.
BLKCHH2 does not prevent a page eject for the second and subsequent page
headers. If BLKCCH2 is specified without HEADER2, it will not be used.
Sample Syntax:
OUTFIL FNAMES=RPT2,
* Do page eject for HEADER1, but not for first HEADER2.
HEADER1=(30:January Report,2/),
BLKCCH2,HEADER2=(5:Account Number,25:Name)
Default for BLKCCH2: None; must be specified.

BLKCCT1
line of the report trailer (TRAILER1) is to be replaced with a blank. You can
use BLKCCT1 to avoid forcing a page eject at the start of the report trailer. If
BLKCCT1 is specified without TRAILER1, it will not be used.
Sample Syntax:
,
OUTFIL FNAMES=RPT3,BLKCCT1
TRAILER1=(5:Grand Total is ,TOT=(21,10,ZD,M5))
Default for BLKCCT1: None; must be specified.

REMOVECC
REMOVECC
Specifies that the ANSI carriage control character is to be removed from

OUTFIL output records for this OUTFIL group before the records are written.
In addition, blank lines are not used to position the page trailer (TRAILER2) at
the bottom of the page.
If REMOVECC is specified without any of the following report parameters, it
will not be used: LINES, HEADER1, TRAILER1, HEADER2, TRAILER2,
SECTIONS, or NODETAIL.
Sample Syntax:
OUTFIL FNAMES=RPTWOCC,
TRAILER1=(3/,Number of records is ,
COUNT=(TO=ZD,LENGTH=6)),
REMOVECC
Default for REMOVECC: None; must be specified.

IFTRAIL
IFTRAIL=(TRLID=(logexp),TRLUPD=(c:item,...)

,HD=YES
IFTRAIL allows you to update count and total values in an existing trailer
(last) record to reflect the actual data records in an OUTFIL data set. Whereas
TRAILER1 lets you build a new trailer record, IFTRAIL lets you update an
existing trailer record using the COUNT and TOTAL features of TRAILER1.
When you add, delete or modify input data records to create an OUTFIL data
set, you can use IFTRAIL to ensure that the trailer record has accurate updated
count and total values.
The count and total values are determined using the data records processed as
input to OUTFIL. If slash (/) is used in OUTFIL BUILD to produce multiple
output data records, the count and total values are determined using the data
370

records processed as input to OUTFIL. For example, if you use OUTFIL
BUILD=(1,80,/,1,80) to create 6 OUTFIL output records from 3 OUTFIL input
records, the count is 3, not 6.
You must identify the trailer record using a unique condition (for example,
'TRL' in positions 1-3). You then specify how count and total values in the
identified trailer record are to be updated. The trailer record will NOT be
treated as a data record, that is, no other OUTFIL processing will be performed
against the trailer record. You can optionally specify that the header (first)
record will NOT be treated as a data record.
As a simple example, you could use the following OUTFIL statement to write
the header (first) record, data records with 'D1', and trailer record (identified
by '9' in position 1) in the OUTFIL data set, and set an accurate count and total
in the trailer record for the included input data records.
OUTFIL FNAMES=OUT1,INCLUDE=(10,2,CH,EQ,CD1),
IFTRAIL=(HD=YES,TRLID=(1,1,CH,EQ,C9),
TRLUPD=(11:COUNT=(M11,LENGTH=8),
25:TOTAL=(15,6,ZD,EDIT=(IIIIIT))))
The trailer record will not be treated as a data record; it will not be subject to
INCLUDE processing, and will not be used for the updated count or total
values.
Since HD=YES is specified, the header record will not be treated as a data
record; it will not be subject to INCLUDE processing, and will not be used for
the updated count or total values.
IFTRAIL
Allows you to update count and total fields in the trailer record of the
data sets for an OUTFIL group. You can use IFTRAIL to identify a
trailer record and indicate count and total fields to be updated in that
record. The identified trailer record will not be used to determine the
count or totals. If you specify HD=YES, the header (first) record will
not be used to determine the count or totals.
You cannot specify IFTRAIL with any of the following operands in an
OUTFIL statement: HEADER1, TRAILER1, HEADER2, TRAILER2,
NODETAIL, SECTIONS, LINES, SPLIT, SPLITBY, SPLIT1R, REPEAT,
FTOV, VTOF, VLTRIM, VLTRAIL or VLFILL.
You must specify the TRLID and TRLUPD operands. The HD=YES
operand is optional.
TRLID=(logexp)
Specifies the criteria to be tested to determine if a record is the trailer
record. When a record is found that meets the criteria, it will be treated
as the last record, and subsequent records will not be processed for the
OUTFIL group. Thus if you use TRLID=(1,1,CH,EQ,C'9') to identify the
trailer record and you have three records with '9' in position 1, the first
record with '9' in position 1 will be treated as the trailer record, and
records after that will not appear in the OUTFIL data sets.
See the discussion of INCLUDE statement in INCLUDE control
statement on page 96 for details of the logical expressions that you
can use. You can specify all of the logical expressions for TRLID in the
same way that you can specify them for the INCLUDE statement,
except that:
v You cannot specify FORMAT=f
v You cannot specify D2 format
v Locale processing is not used
371

v VLSCMP and VLSHRT are not used. The trailer record must be long
enough to contain all fields in the logical expression. Trailer record
checking will be skipped for any variable-length record that is too
short to contain a field in the TRLID logical expression.
The trailer record will not be treated as a data record; all other OUTFIL
processing (for example, INCLUDE, OMIT, BUILD, OVERLAY, and so
on) will be bypassed for the trailer record. The updated count and total
values will not include the trailer record.
TRLUPD=(c:item,...)
Specifies the position where each count or total is to be updated in the
trailer record.
For fixed-length records, the first input and output data byte starts at
position 1. The trailer record will be truncated or padded with blanks
to the length of the data records, if appropriate. You cannot change the
length of the trailer record using TRLUPD.
For variable-length records, the first input and output data byte starts
at position 5, after the RDW in positions 1-4. The original length of the
trailer record will not be changed for output. You cannot change the
length of the trailer record using TRLUPD. If you specify an item
beyond the end of the trailer record, it will not be updated.
c: specifies the output position (column) in the trailer record to be
updated. If you do not specify c: for the first item, it defaults to 1:.
If you do not specify c: for any other item, it starts after the
previous item. You must specify items in ascending output column
order, and an item must not overlap a previous item in the trailer
record. For a variable-length trailer record, you must not specify
columns 1-4.
item
specifies the count or total you want to update. Any of the
following count and total items can be used as described for
TRAILER1:
v COUNT=(edit)
v COUNT=(to)
v
v
v
v
v
v
COUNT+n=(edit)
COUNT+n=(to)
COUNT-n=(edit)
COUNT-n=(to)
TOTAL=(p,m,f,edit)
TOTAL=(p,m,f,to)
v TOT=(p,m,f,edit)
v TOT=(p,m,f,to)
The count and total values are determined using the data records
processed as input to OUTFIL.
If TOT or TOTAL is used, the specified p,m,f field from each data
record will be totalled, even if the trailer record is not found.
OUTFIL BUILD, OVERLAY or IFTHEN processing does NOT affect
the values that are totalled; the values in the original OUTFIL
input records are used for the totals.
372

An invalid field amount can result in a data exception (0C7
ABEND) or incorrect numeric output. Be careful to use HD=YES if
the header record does not have valid total fields. For example, if
the input file has these records:
H***
0001
0002
T***
and this IFTRAIL operand is specified:

IFTRAIL=(TRLID=(1,1,CH,EQ,CT),
TRLUPD=(5:TOT=(1,4,ZD,TO=ZD,LENGTH=5))
an 0C7 ABEND will result when DFSORT attempts to total the

invalid 1,4,ZD field in the header (H***) record. The use of
HD=YES would prevent this.
An invalid total field can also be encountered if the trailer record is
not found due to erroneous TRLID conditions. For example, if the
input file has these records:
H***
0001
0002
T***
and this IFTRAIL operand is specified:

IFTRAIL=(TRLID=(1,2,CH,EQ,CTR),
TRLUPD=(5:TOT=(1,4,ZD,TO=ZD,LENGTH=5))
the last (T***) record will not be identified as the trailer record
since it does not have 'TR' in positions 1-2. Thus, an 0C7 ABEND
will result when DFSORT attempts to total the invalid 1,4,ZD field
in the last (T***) record. The use of the correct condition for
identifying the trailer (T***) record would prevent this.
If slash (/) is used in OUTFIL BUILD to produce multiple output
data records, the count and total values are determined using the
data records processed as input to OUTFIL. For example, if you
use OUTFIL BUILD=(1,80,/,1,80) to create 6 OUTFIL output
records from 3 OUTFIL input records, the count is 3, not 6.
HD=YES
Specify HD=YES if you want the first record treated as a header record
rather than as a data record. With HD=YES, OUTFIL processing (for
example, INCLUDE, OMIT, BUILD, OVERLAY, and so on) will be
bypassed for the header record. The updated count and total values
will not include the header record.
If HD=YES is specified, the TRLID test will not be applied to the first
record.
Sample Syntax:
OUTFIL IFTRAIL=(HD=YES,
TRLID=(1,3,CH,EQ,CTRL),
21:TOT=(M11,LENGTH=6)))
Default for IFTRAIL: None; must be specified.
373
OUTFIL statements notes

v OUTFIL processing is supported for sort, merge, and copy applications, but only
by the Blockset technique.
v
v The ODMAXBF value in effect specifies the maximum buffer space to be used
for each OUTFIL data set. ODMAXBF can be specified as an installation or
run-time option, or in an ICEIEXIT routine. The default value of 2M is
recommended for the ODMAXBF option in effect. Lowering ODMAXBF can
cause performance degradation for the application, but might be necessary if
you consider the amount of storage used for OUTFIL processing to be a
problem. Raising ODMAXBF can improve EXCPs for the application, but can
also increase the amount of storage needed.
v
v The storage used for OUTFIL processing will be adjusted automatically
according to the total storage available, the storage needed for non-OUTFIL
processing, and the number of OUTFIL data sets and their attributes (for
example, block size). OUTFIL processing will be subject to the ODMAXBF limit
in effect and the system storage limits (for example, IEFUSI), but not to the
DFSORT storage limits (that is, SIZE, MAXLIM, and TMAXLIM). DFSORT
attempts to use storage above 16MB virtual for OUTFIL processing whenever
possible.
v The VSAMBSP option applies to SORTOUT data sets, but not to OUTFIL data
sets. The NOBLKSET option will be ignored if OUTFIL data sets are being
processed. An E39 exit routine is entered for the SORTOUT data set, but not for
OUTFIL data sets.
v For fixed-format OUTFIL data sets: DFSORT will determine each OUTFIL data
set LRECL based on the length of the reformatted records for the group, or the
length of the OUTFIL input records (if OUTREC, BUILD, OVERLAY, FINDREP
and IFTHEN are not specified for the group). For VSAM data sets, the
maximum record size defined in the cluster is equivalent to the LRECL.
If an OUTFIL data set LRECL is not specified or available, DFSORT will set it to
the determined LRECL. If an OUTFIL data set LRECL is specified or available, it
must not be less than the determined LRECL, or more than the determined
LRECL if the OUTREC, BUILD, OVERLAY, FINDREP or IFTHEN parameter is
specified. In other words, the LRECL value cannot be used to pad the output
records, or to truncate the records produced by OUTREC, BUILD, OVERLAY,
FINDREP or IFTHEN parameter processing.
In general, OUTREC, BUILD, OVERLAY, or IFTHEN processing should be used
to pad the records and OUTREC, BUILD, or IFTHEN processing should be used
to truncate the records. In either case, the LRECL should either not be specified
or set to the length of the reformatted records.
v For variable-format OUTFIL data sets: DFSORT will determine each OUTFIL
data set maximum LRECL based on the length of the reformatted records for the
group, or the length of the OUTFIL input records (if OUTREC, BUILD,
OVERLAY, FINDREP and IFTHEN are not specified for the group). If an
OUTFIL data set maximum LRECL is not specified or available, DFSORT will set
it to the determined maximum LRECL. For VSAM data sets, the maximum
record size defined in the cluster is four bytes more than the maximum LRECL.
v When you create an OUTFIL report, the length for the longest or only data
record must be equal to or greater than the maximum report record length. You
can use the OUTREC, BUILD, OVERLAY, FINDREP or IFTHEN parameter to
force a length for the data records that is longer than any report record; you can
then either let DFSORT compute and set the LRECL, or ensure that the
374

computed LRECL is equal to the existing or specified LRECL. Remember to
allow an extra byte in the LRECL for the ANSI carriage control character.
For example, if your data records are 40 bytes, but your longest report record is
60 bytes, you can use an OUTREC parameter such as:
OUTREC=(1,40,80:X)
DFSORT will then set the LRECL to 81 (1 byte for the ANSI carriage control
character plus 80 bytes for the length of the data records), and pad the data
records with blanks on the right.
data set, use the REMOVECC parameter to remove them. For example, if you
specify:
OUTREC=(1,40,80:X),REMOVECC
DFSORT will set the LRECL to 80 instead of 81 and remove the ANSI carriage
control character from each record before it is written.
System errors can result if you print an OUTFIL report containing records longer
than your printer can handle.
v DFSORT uses appropriate ANSI carriage controls (for example, C'-' for triple
space) in header and trailer records when possible to reduce the number of
report records written. DFSORT always uses the single space carriage control (C'
') in data records. Although these carriage control characters may not be shown
when you view an OUTFIL data set (depending on how you view it), they will
be used if you print the report. If you are creating a report for viewing and want
blank lines to appear in headers and trailers, specify a line of blanks instead of
using n/. For example, instead of specifying:
OUTFIL FNAMES=RPT,
HEADER2=(2/,start of header,2/,next line)
which will result in blank lines for the printer, but not for viewing, specify:
OUTFIL FNAMES=RPT,
HEADER2=(X,/,X,/,start of header,/,X,/,next line)
data set, use the REMOVECC parameter to remove them.
v For variable-length records, the first entry in the OUTREC, BUILD or IFTHEN
BUILD parameter must specify or include the unedited 4-byte record descriptor
word (RDW), that is, the first field must be 1,4 or 1,m with m greater than 4.
DFSORT sets the length of the reformatted record in the RDW.
the reformatted record immediately following the RDW, the entry in the
OUTREC, BUILD, or IFTHEN BUILD parameter can specify both RDW and data
field in one (1,m,...). Otherwise, the RDW must be specifically included in the
reformatted record (for example, 1,4,1,4,HEX).
For variable-length OUTFIL header or trailer records, you must not specify the
4-byte RDW at the beginning of the record.
v With FIELDS, BUILD or IFTHEN BUILD, the variable part of the input record
reformatted record and, if included, must be the last part. For example:
OUTFIL BUILD=(1,8,20C*,9)
375
v
v
For variable-length records, the FIELDS or BUILD parameter of the INREC and
OUTREC statement and the OUTREC or FIELDS parameter of the OUTFIL
statement must all specify position-only for the last part, or all not specify
position-only for the last part. OVERLAY or IFTHEN, and FIELDS, OUTREC or
BUILD, can differ with respect to position-only. See INREC statement notes on
page 148 for more details.
If there are no OUTFIL input records for an OUTFIL group, the headers and
trailers appear without any data records. Blanks will be used for any specified
unedited input fields, and zero values will be used for any specified statistics
fields.
If a variable-length OUTFIL input record is too short to contain a specified
unedited input field for a report header or trailer, blanks will be used for the
missing bytes. If a variable-length OUTFIL input record is too short to contain a
specified section break field or statistics field, zeros will be used for the missing
bytes, intentionally or unintentionally.
If a variable-length OUTFIL input record is too short to contain an OUTFIL
INCLUDE or OMIT compare field, the action DFSORT takes depends on the
settings for VLSCMP/NOVLSCMP and VLSHRT/NOVLSHRT. For details, see
the discussion of the VLSCMP and NOVLSCMP options in OPTION control
OUTREC or BUILD field, DFSORT will terminate unless the VLFILL=byte
parameter is specified.
OVERLAY or IFTHEN field, blanks will be used for the missing bytes.
If a variable-length OUTFIL output data record is longer than the LRECL of its
OUTFIL data set, the action DFSORT takes depends on the settings for
VLLONG/NOVLLONG. For details, see the discussion of the VLLONG and
NOVLLONG options in OPTION control statement on page 173. Note that
VLLONG can be used to truncate long OUTFIL data records, but has no effect
on long OUTFIL header or trailer records.
If an unedited page number overflows 6 digits, an edited page number
overflows 15 digits, a count or running count overflows 15 digits, or a total or
running total overflows 31 digits, the overflowing value will be truncated to the
number of digits allowed, intentionally or unintentionally.
v Multiple OUTFIL statements can be specified in the same and different sources.
If a ddname occurs more than once in the same source, the ddname is associated
with the first OUTFIL group in which it appears. For example, if the following is
specified in SYSIN:
OUTFIL FNAMES=(OUT1,OUT2),INCLUDE=(1,1,CH,EQ,CA)
OUTFIL FNAMES=(OUT3,OUT1),SAVE
OUT1 and OUT2 are processed as part of the first OUTFIL group, that is, with
INCLUDE. OUT3 is processed as part of the second OUTFIL group, that is, with
SAVE; but OUT1 is not because it is a duplicate ddname.
If a ddname occurs in more than one source, the ddname is associated with the
highest source OUTFIL group in which it appears. For example, if the following
is specified in DFSPARM:
OUTFIL FNAMES=(OUT1,OUT2),INCLUDE=(1,1,CH,EQ,CA)
and the following is specified in SYSIN:
376

OUTFIL FNAMES=(OUT3,OUT1),SAVE
OUT1 and OUT2 are processed as part of the DFSPARM OUTFIL group, that is,
with INCLUDE. OUT3 is processed as part of the SYSIN OUTFIL group, that is,
with SAVE; but OUT1 is not because it is an overridden ddname.
v OUTFIL statements cannot be passed to or returned from an EFS program. The
D2 format cannot be specified in the INCLUDE or OMIT parameter of an
OUTFIL statement.
edited or converted input fields, minimums, maximums, decimal constants and
the results of arithmetic expressions. If NOSZERO is in effect, -0 and +0 are
treated as positive for edited or converted input fields, minimums, maximums,
decimal constants, and the results of arithmetic expressions.
Note: OPTION SZERO or OPTION NOSZERO is ignored for OUTFIL
INCLUDE=(logexp) or OMIT=(logexp), or for OUTFIL
OUTFIL statement in the same source. For example, NOSZERO will be used in
Case 1:
//DFSPARM DD *
OPTION COPY,NOSZERO
/*
//SYSIN DD *
OUTFIL INCLUDE=(1,2,FS,EQ,+0)
/*
Case 2:
//SYSIN DD *
OPTION COPY,NOSZERO
OUTFIL IFTHEN=(WHEN=(1,2,FS,EQ,+0),OVERLAY=(28:CA)),
/*
OUTFIL featuresexamples
Example 1
OPTION
OUTFIL
OUTFIL
OUTFIL
OUTFIL
OUTFIL
COPY
INCLUDE=(15,6,CH,EQ,CMSG005),FNAMES=M005
SAVE,FNAMES=UNKNOWN
This example illustrates how records can be distributed to different OUTFIL data
sets based on criteria you specify:
v Input records with MSG005 in bytes 15 through 20 will be written to the
OUTFIL data set associated with ddname M005.
377

v Input records with anything else in bytes 15 through 20 will be written to the
OUTFIL data set associated with ddname UNKNOWN
Example 2
SORT FIELDS=(18,5,ZD,D)
OUTFIL FNAMES=(V,VBU1,VBU2)
OUTFIL FNAMES=(F,FBU1),
CONVERT,OUTREC=(11,3,X,18,5,X,X0000000F)
OUTFIL FNAMES=VINF,OUTREC=(1,4,C*,5,20,C*,25)
This example illustrates how multiple sorted output data sets can be created and
how a variable-length record data set can be converted to a fixed-length record
data set:
v The first OUTFIL statement writes the variable-length input records to the
variable-length OUTFIL data sets associated with ddnames V, VBU1, and VBU2.
v The second OUTFIL statement reformats the variable-length input records to
fixed-length output records and writes them to the fixed-length OUTFIL data
sets associated with ddnames F and FBU1. CONVERT is used to indicate that a
variable-length data set is to be converted to a fixed-length data set; OUTREC is
used to describe how the variable-length input records are to be reformatted as
fixed-length output records.
v The third OUTFIL statement reformats the variable-length input records and
writes them to the variable-length OUTFIL data set associated with ddname
VINF. OUTREC is used to insert asterisks between fields. 1,4 represents the
RDW. 25 represents the variable part at the end of the input record.
Example 3
OUTFIL FNAMES=USA,
HEADER2=(5:Parts Completion Report for USA,2/,
5:Printed on ,DATE,
at ,TIME=(12:),3/,
5:Part ,20:Completed,35:
Value ($),/,
5:------,20:---------,35:------------),
OUTREC=(5:15,6,ZD,M11,
20:3,4,ZD,M12,LENGTH=9,
35:38,8,ZD,M18,LENGTH=12,
132:X)
OUTFIL FNAMES=FRANCE,
HEADER2=(5:Parts Completion Report for France,2/,
5:Printed on ,DATE=(DM4/),
at ,TIME,3/,
5:Part ,20:Completed,35:
Value (F),/,
5:------,20:---------,35:------------),
20:3,4,ZD,M16,LENGTH=9,
35:38,8,ZD,M22,LENGTH=12,
132:X)
OUTFIL FNAMES=DENMARK,
HEADER2=(5:Parts Completion Report for Denmark,2/,
5:Printed on ,DATE=(DMY-),
at ,TIME=(24.),3/,
5:Part ,20:Completed,35: Value (kr),/,
5:------,20:---------,35:------------),
378

20:3,4,ZD,M13,LENGTH=9,
35:38,8,ZD,M19,LENGTH=12,
132:X)
This example illustrates how reports for three different countries can be produced
from sorted fixed-length input records. The reports differ only in the way that
date, time, and numeric formats are specified:
1. The first OUTFIL statement produces a report that has the date, time, and
numeric formats commonly used in the United States.
2. The second OUTFIL statement produces a report that has the date, time, and
numeric formats commonly used in France.
3. The third OUTFIL statement produces a report that has the date, time, and
numeric formats commonly used in Denmark.
Of course, any one of the three reports can be produced by itself using a single
OUTFIL statement instead of three OUTFIL statements. (This may be necessary if
you are sorting character data according to a specified locale for that country.)
The FNAMES parameter specifies the ddname (USA, FRANCE, DENMARK)
associated with the fixed-length data set for that report.
The HEADER2 parameter specifies the page header to appear at the top of each
page for that report, which will consist of:
v A line of text identifying the report. Note that all English text in the report can
be replaced by text in the language of that country.
v A blank line (2/).
v A line of text showing the date and time. Note that variations of the DATE,
DATE=(abcd), TIME, and TIME=(abc) operands are used to specify the date and
time in the format commonly used in that country.
v Two blank lines (3/).
v Two lines of text showing headings for the columns of data. Note that the
appropriate currency symbol can be included in the text.
The OUTREC parameter specifies the three columns of data to appear for each
input record as follows:
v A 6-byte edited numeric value produced by transforming the ZD value in bytes
15 through 20 according to the pattern specified by M11. M11 is a pattern for
showing integers with leading zeros.
v A 9-byte (LENGTH=9) edited numeric value produced by transforming the ZD
value in bytes 3 through 6 according to the pattern for integer values with
thousands separators commonly used in that country. M12 uses a comma for the
thousands separator. M16 uses a blank for the thousands separator. M13 uses a
period for the thousands separator.
v A 12-byte (LENGTH=12) edited numeric value produced by transforming the
ZD value in bytes 38 through 45 according to the pattern for decimal values
with thousands separators and decimal separators commonly used in that
country. M18 uses a comma for the thousands separator and a period for the
decimal separator. M22 uses a blank for the thousands separator and a comma
for the decimal separator. M19 uses a period for the thousands separator and a
comma for the decimal separator.
Table 46 on page 271 shows the twenty-seven pre-defined edit masks (M0-M26)
from which you can choose.
379

132:X is used at the end of the OUTREC parameter to ensure that the data records
are longer than the report records. This will result in an LRECL of 132 for the
fixed-length OUTFIL data sets (1 byte for the ANSI control character and 131 bytes
for the data).
The three reports might look as follows:
Parts Completion Report for USA
Printed on 03/25/05 at 01:56:20 pm
Part
-----000310
001184
029633
192199
821356
Completed
--------562
1,234
35
3,150
233
Value ($)
-----------8,317.53
23,456.78
642.10
121,934.65
2,212.34
Parts Completion Report for France

Printed on 25/03/2005 at 13:56:20
Part
-----000310
001184
029633
192199
821356
Completed
--------562
1 234
35
3 150
233
Value (F)
-----------8 317,53
23 456,78
642,10
121 934,65
2 212,34
Parts Completion Report for Denmark

Printed on 25-03-05 at 13.56.20
Part
-----000310
001184
029633
192199
821356
Completed
--------562
1.234
35
3.150
233
Value (kr)
-----------8.317,53
23.456,78
642,10
121.934,65
2.212,34
Example 4
SORT FIELDS=(3,10,A,16,13,A),FORMAT=CH
OUTFIL FNAMES=WEST,
INCLUDE=(42,6,CH,EQ,CWest),
HEADER1=(5/,18:
Western Region,3/,
18:Profit and Loss Report,3/,
18:
for ,&DATE,3/,
18:
Page,&PAGE),
OUTREC=(6:16,13,24:31,10,ZD,M5,LENGTH=20,75:X),
HEADER3=(2:Division: ,3,10,5X,Page:,&PAGE,2/,
6:Branch Office,24:
Profit/(Loss),/,
6:-------------,24:--------------------),
TRAILER3=(6:=============,24:====================,/,
6:Total,24:TOTAL=(31,10,ZD,M5,LENGTH=20),/,
6:Lowest,24:MIN=(31,10,ZD,M5,LENGTH=20),/,
6:Highest,24:MAX=(31,10,ZD,M5,LENGTH=20),/,
6:Average,24:AVG=(31,10,ZD,M5,LENGTH=20),/,
3/,2:Average for all Branch Offices so far:,
X,SUBAVG=(31,10,ZD,M5))),
380

TRAILER1=(8:Page ,&PAGE,5X,Date: ,&DATE,5/,
8:Total Number of Branch Offices Reporting:
COUNT,2/,
8:Summary of Profit/(Loss) for all,
Western Division Branch Offices,2/,
12:Total:,
22:TOTAL=(31,10,ZD,M5,LENGTH=20),/,
12:Lowest:,
22:MIN=(31,10,ZD,M5,LENGTH=20),/,
12:Highest:,
22:MAX=(31,10,ZD,M5,LENGTH=20),/,
12:Average:,
22:AVG=(31,10,ZD,M5,LENGTH=20))
This example illustrates how a report can be produced with a header and trailer
page and sections of columns of data, from a sorted subset of fixed-length input
records.
The FNAMES parameter specifies the ddname (WEST) associated with the
fixed-length data set for the report.
The INCLUDE parameter specifies the records to be selected for the report.
The HEADER1 parameter specifies the report header to appear as the first page of
the report, which will consist of five blank lines (5/) followed by four lines of text,
each separated by 2 blank lines (3/). The last two lines of text will show the date
(&DATE) and page number (&PAGE), respectively.
The OUTREC parameter specifies the two columns of data to appear for each
selected input record as follows:
v The character string from bytes 16 through 28 of the input record.
v A 20-byte (LENGTH=20) edited numeric value produced by transforming the
ZD value in bytes 31 through 40 according to the pattern specified by M5.
The SECTIONS parameter specifies the section break field (3,10), page ejects
between sections (SKIP=P), the header (HEADER3) to appear before each section
and the trailer (TRAILER3) to appear after each section. The section header will
consist of a line of text showing the page number, a blank line (2/) and two lines
of text showing the headings for the columns of data. The section trailer will
consist of a line of text separating the data from the trailer, lines of text showing
the total (TOTAL), minimum (MIN), maximum (MAX) and average (AVG) for the
data in the section as edited numeric values, two blank lines, and a line of text
showing the running average (SUBAVG) for all of the data records in the report up
to this point.
The TRAILER1 parameter specifies the report trailer to appear as the last page of
the report, which will consist of a line of text showing the page and date, four
blank lines (5/), a text line showing the count of data records in the report, a blank
line, a line of text, a blank line, and lines of text showing the total, minimum
maximum and average for all of the data in the report as edited numeric values.
are longer than the report records. This will result in an LRECL of 76 for the
fixed-length OUTFIL data set (1 byte for the ANSI control character and 75 bytes
for the data).
The report might look as follows:
381
Western Region
Profit and Loss Report
for
08/20/05
Page
Division:
Chips
Branch Office
------------Gilroy
Los Angeles
Morgan Hill
Oakland
San Francisco
San Jose
San Martin
=============
Total
Lowest
Highest
Average
Page:
Profit/(Loss)
-------------------554,843.42
(22,340.14)
987,322.32
234,124.32
(32,434.31)
1,232,133.35
889,022.03
====================
3,842,670.99
(32,434.31)
1,232,133.35
548,952.99
Average for all Branch Offices so far:
Division:
Ice Cream
Branch Office
------------Marin
Napa
San Francisco
San Jose
San Martin
=============
Total
Lowest
Highest
Average
Page:
548,952.99
Profit/(Loss)
-------------------542,341.23
857,342.83
922,312.45
(234.55)
1,003,467.30
====================
3,325,229.26
(234.55)
1,003,467.30
665,045.85
Division:
Pretzels
Branch Office
------------Marin
Morgan Hill
Napa
San Francisco
San Jose
San Martin
=============
Total
Lowest
Highest
Average
Page:
Profit/(Loss)
-------------------5,343,323.44
843,843.40
5,312,348.56
5,412,300.05
1,234,885.34
(2,343.82)
====================
18,144,356.97
(2,343.82)
5,412,300.05
3,024,059.49
382
597,325.02
1,406,236.51
Page
Date:
08/20/05
Total Number of Branch Offices Reporting:
18
Summary of Profit/(Loss) for all Western Division Branch Offices

Total:
Lowest:
Highest:
Average:
25,312,257.22
(32,434.31)
5,412,300.05
1,406,236.51
Example 5
OUTFIL FNAMES=STATUS,
HEADER2=(1:CPAGE ,&PAGE,C OF STATUS REPORT FOR ,&DATE,2/,
6:CITEM ,16:CSTATUS
,31:CPARTS,/,
6:C-----,16:C------------,31:C-----),
OUTREC=(1,4,
10:6,5,
20:14,1,CHANGE=(12,
C1,CSHIP,
C2,CHOLD,
C3,CTRANSFER),
NOMATCH=(C*CHECK CODE*),
37:39,1,BI,M10,
120:X)
This example illustrates how a report can be produced with a page header and
columns of data from sorted variable-length input records, using a lookup table.
The FNAMES parameter specifies the ddname (STATUS) associated with the
variable-length data set for the report.
The HEADER2 parameter specifies the page header to appear at the top of each
page, which will consist of a line of text showing the page number (&PAGE) and
date (&DATE), a blank line (2/), and two lines of text showing headings for the
columns of data.
The OUTREC parameter specifies the RDW and three columns of data to appear
for each input record as follows (remember that byte 5 is the first byte of data for
variable-length records):
v The character string from bytes 6 through 10 of the input record
v A character string produced by finding a match for byte 14 of the input record
in the table defined by CHANGE (lookup and change). NOMATCH indicates
the character string to be used if byte 14 does not match any of the entries in the
CHANGE table.
v An edited numeric value produced by transforming the BI value in byte 39
according to the pattern specified by M10.
With variable-length input records, you must account for the RDW when
specifying the c: values for OUTREC, but not for headers or trailers. The 1: used
for the first line of HEADER2 causes it to start in the first data byte (by contrast, 5:
must be used to specify the first OUTREC data byte for variable-length records).
Also, because 6: is used for the ITEM heading, 10: must be used for the ITEM data
to get the heading and data to line up in columns.
383

are longer than the report records. This will result in a maximum LRECL of 121 for
the variable-length OUTFIL data set (1 byte for the ANSI control character and a
maximum of 120 bytes for the data).
The first page of the printed report might start as follows:
PAGE
1 OF STATUS REPORT FOR 07/18/05

ITEM
----00082
00123
00300
10321
12140
STATUS
-----------HOLD
SHIP
*CHECK CODE*
TRANSFER
SHIP
PARTS
----36
106
95
18
120
Example 6
OPTION
OUTFIL
OUTFIL
OUTFIL
COPY
FNAMES=(A1,A2,A3),SPLIT
FNAMES=(B1,B2,B3),SPLITBY=25
FNAMES=(C1,C2,C3),SPLIT1R=25
This example illustrates different ways to split 77 input records among three
OUTFIL data sets.
The first OUTFIL statement uses SPLIT to rotate the records among the three
OUTFIL data sets one record at a time. A1 will have records 1, 4, ...,76. A2 will
have records 2, 5, ...,77. A3 will have records 3, 6, ...,75. A1 will have 26 records. A2
will have 26 records. A3 will have 25 records. A1, A2 and A3 will each have
non-contiguous records.
The second OUTFIL statement uses SPLITBY=25 to rotate the records among the
three OUTFIL data sets 25 records at a time. B1 will have records 1-25 and 76-77.
B2 will have records 26-50. B3 will have records 51-75. B1 will have 27 records. B2
will have 25 records. B3 will have 25 records. B1 will have non-contiguous records.
B2 and B3 will have contiguous records.
The third OUTFIL statement uses SPLIT1R=25 to rotate the records once among the
three OUTFIL data sets 25 records at a time. C1 will have records 1-25. C2 will
have records 26-50. C3 will have records 51-77. C1 will have 25 records. C2 will
have 25 records. C3 will have 27 records. C1, C2 and C3 will have contiguous
records.
Example 7
OPTION
OUTFIL
OUTFIL
OUTFIL
OUTFIL
OUTFIL
COPY
FNAMES=RANGE1,ENDREC=1000000
FNAMES=RANGE2,STARTREC=1000001,ENDREC=2000000
FNAMES=(RANGE5,EXTRA),STARTREC=4000001
This example illustrates how specific ranges of output records can be written to
different output data sets. A typical application might be database partitioning.
The first 1 million records will be written to the data set associated with RANGE1,
the second million to RANGE2, the third million to RANGE3, and the fourth
million to RANGE4. The remaining records will be written to both the data set
384

associated with RANGE5 and the data set associated with EXTRA (SAVE or
STARTREC=4000001 will accomplish the same purpose in this case).
Note that the INCLUDE, OMIT, and SAVE parameters of OUTFIL can also be used
to select records to be written to different output data sets, based on criteria you
specify.
Example 8
OPTION COPY,Y2PAST
OUTFIL FNAMES=Y4,
OUTREC=(1,19,
21,2,PD0,M11,C/,
22,2,PD0,M11,C/,
20,2,Y2P,
24,57)
transform mm
transform dd
transform yy to yyyy
This example illustrates how to transform an existing data set with a packed
decimal date field of the form P'yymmdd' (X'0yymmddC') in bytes 20-23 into a
new data set with a character date field of the form C'mm/dd/yyyy' in bytes
20-29. yy represents the two-digit year, yyyy represents the four-digit year, mm
represents the month, dd represents the day, and C represents a positive sign.
The input data set has an LRECL of 80 and the Y4 data set will have an LRECL of
86.
The Y2PAST=26 option sets the century window to be used to transform two-digit
years into four-digit years. If the current year is 2006, the century window will be
1980 to 2079. Using this century window, the input and output fields might be as
follows:
Input Field (HEX)
20
|
0020505F
0950823C
0980316C
0000316F
Output Field (CH)

20
|
05/05/2002
08/23/1995
03/16/1998
03/16/2000
Example 9
OUTFIL FNAMES=SPCL,
OUTREC=(1,14, copy positions 1-14
15,6,Y2T, transform yy to yyyy - allow blanks
21,20)
copy positions 21 - 40
This example illustrates how to transform an existing data set with a character date
field of the form C'yymmdd' and blank special indicators in bytes 15-20, into a
new date set with a character date field of the form C'yyyymmdd' and blank
special indicators in bytes 15-22.
The input data set has an LRECL of 40 and the SPCL data set will have an LRECL
of 42.
The Y2PAST=1996 option sets the century window to 1996-2095. The century
window will be used to transform the two-digit years into four-digit years, but will
not be used for the special blank indicators.
385

MORGAN HILL
SAN JOSE
BOCA RATON
DENVER
960512
000628
951115
CA
CA
FL
CO

MORGAN HILL
SAN JOSE
BOCA RATON
DENVER
19960512
20000628
20951115
CA
CA
FL
CO
Example 10
OPTION COPY
OUTFIL FNAMES=ALL,OUTREC=(CUS ,1,10,C
CWW ,1,10,C
OUTFIL FNAMES=(US,WW),SPLIT,
OUTREC=(1,10,C is
1,10,C is
is in ,11,15,/,
is in ,26,20,2/)
in ,11,15,/,
in ,26,20)
This example illustrates how multiple OUTFIL output and blank records can be
produced from each OUTFIL input record. The input data set has an LRECL of 50
and contains the following three records:
Finance
San Francisco Buenos Aires
Research New York
Amsterdam
Marketing Los Angeles
Mexico City
The first OUTFIL statement creates the data set associated with ddname ALL. This
data set will have an LRECL of 40 (the length of the longest output record; the one
that includes the 26,20 input field). Each input record will result in two data
records followed by two blank records as follows:
ALL data set
US Finance
WW Finance
is in San Francisco
is in Buenos Aires
US Research
WW Research
is in New York
is in Amsterdam
US Marketing
WW Marketing
is in Los Angeles
is in Mexico City
The second OUTFIL statement creates the two data sets associated with ddnames
US and WW. These data sets will have an LRECL of 37 (the length of the longest
output record; the one that includes the 26,20 input field). Each input record will
result in two data records. SPLIT will cause the first data record to be written to
the US data set and the second data record to be written to the WW data set. Thus,
each input record will create one record in each OUTFIL data set as follows:
US data set
Finance
is in San Francisco
Research
is in New York
Marketing is in Los Angeles
WW data set
Finance
is in Buenos Aires
Research
is in Amsterdam
Marketing is in Mexico City
386
Example 11
SORT FIELDS=(6,3,CH,D)
OUTFIL FNAMES=SET60,BUILD=(1,60),VLFILL=C
OUTFIL FNAMES=VARFIX,VTOF,BUILD=(5,20,5X,28,20),VLFILL=C*
This example illustrates how variable-length records that are too short to contain
all OUTFIL BUILD fields can be processed successfully.
The input data set has RECFM=VB and LRECL=80. The records in this data set
have lengths that vary from 15 bytes to 75 bytes.
The first OUTFIL statement creates the data set associated with ddname SET60.
This data set will have RECFM=VB and LRECL=60. Every record in this data set
will have a length of 60. The 1,60 field truncates records longer than 60 bytes to 60
bytes. Because VLFILL=C' ' is specified, the 1,60 field pads records shorter than 60
bytes to 60 bytes using a blank (C' ') as the fill byte.
Note: Without VLFILL=byte, this OUTFIL statement would terminate with an
ICE218A message because some of the input records are too short to contain the
BUILD field.
The second OUTFIL statement creates the data set associated with ddname
VARFIX. This data set will have RECFM=FB and LRECL=45. VTOF changes the
variable-length input records to fixed-length output records according to the fields
specified by BUILD. VLFILL=C'*' allows short input records to be processed. Each
missing byte in an OUTFIL BUILD field is replaced with an asterisk (C'*') fill byte.
Note:
1. CONVERT can be used instead of VTOF.
2. VLFILL=C'*' overrides the default of VLFILL=X'40' for VTOF or CONVERT.
Example 12
OPTION COPY
OUTFIL OUTREC=(SEQNUM,4,BI,Z,8,5,ZD,TO=PD,Z,
31,2,PD,TO=FI,LENGTH=2),Z,
16,3,ZD,ADD,+1,TO=FI,LENGTH=2,Z,
(16,3,ZD,MAX,31,2,PD),MUL,+2,TO=ZD,LENGTH=4)
This example illustrates how a sequence number can be generated, how values in
one numeric format can be converted to another numeric format, and how
arithmetic expressions involving fields and decimal constants can be used.
The input data set has an LRECL of 50 and the SORTOUT data set will have an
LRECL of 19.
The OUTFIL statement creates output records with the following fields:
A binary sequence number in bytes 1-4 that starts at 1 and increments by 1.
X'00' in byte 5.
A PD field in bytes 6-8 containing the converted ZD field from input bytes 8-12
X'00' in position 9.
An FI field in bytes 10-11 containing the converted PD field from input bytes
31-32.
v X'00' in position 12.
v An FI field in bytes 13-14 containing the converted result of the ZD field from
input bytes 16-18 incremented by 1.
v
v
v
v
v
387

v X'00' in position 15.
v A ZD field in bytes 16-19 containing the converted result of the maximum of the
ZD field from input bytes 16-18 and the PD field from input bytes 31-32,
multiplied by 2.
Example 13
SORT FIELDS=COPY
OUTFIL FNAMES=VAROUT1,FTOV
OUTFIL FNAMES=VAROUT2,FTOV,
OUTREC=(20,8,35,10)
OUTFIL FNAMES=VAROUT3,FTOV,VLTRIM=X40
This example illustrates several ways to convert a fixed-length record data set to a
variable-length record data set using the FTOV parameter of OUTFIL.
The input data set has an RECFM=FB and LRECL=60.
v The first OUTFIL statement converts the fixed-length input data set to a
variable-length OUTFIL data set associated with ddname VAROUT1. VAROUT1
will have RECFM=VB and LRECL=64. All of its records will be 64 bytes long
(4-byte RDW plus 60-byte input record).
v The second OUTFIL statement converts the fixed-length input data set to a
variable-length OUTFIL data set associated with ddname VAROUT2. OUTREC is
used to select two input fields for the output records, bytes 20-27 and bytes
35-44. VAROUT2 will have RECFM=VB and LRECL=22. All of its records will be
22 bytes long (4-byte RDW plus 8-byte input field plus 10-byte input field).
v The third OUTFIL statement converts the fixed-length input data set to a
variable-length OUTFIL data set associated with ddname VAROUT3. VAROUT3
will have RECFM=VB and LRECL=64. VLTRIM=X'40' is used to remove the
trailing blanks from the variable-length output records. The records can vary
from 5 bytes long to 64 bytes long depending on the number of trailing blanks
in each record.
Example 14
OPTION
OUTFIL
OUTFIL
OUTFIL
COPY
FNAMES=OUT1,OUTREC=(DATE1(/),X,TIME1(:),X,1,80)
FNAMES=OUT2,OUTREC=(DATE2P,TIME3P,1,80)
FNAMES=OUT3,OUTREC=(DATE3(.),X,TIME2,X,1,80)
This example illustrates several different ways to insert timestamps into your
records.
The input data set has RECFM=FB and LRECL=80.
The first OUTFIL statement creates the data set associated with ddname OUT1.
This data set will have LRECL=100. Each output record will have a timestamp
consisting of the date and time of the run in the form C'yyyy/mm/dd hh:mm:ss '
(20 bytes), followed by the original input record (80 bytes).
The second OUTFIL statement creates the data set associated with ddname OUT2.
consisting of the date of the run in the form P'yyyymm' (4 bytes) and the time of
the run in the form P'hh' (2 bytes), followed by the original input record (80 bytes).
The third OUTFIL statement creates the data set associated with ddname OUT3.
388

consisting of the date and time of the run in the form C'yyyy.ddd hhmm ' (14
bytes), followed by the original input record (80 bytes).
Example 15
OPTION COPY
OUTREC FIELDS=(1,4,11,4,DT1,7,4,TM1,60:X)
OUTFIL NODETAIL,
TRAILER1=(//,
3:Earliest SMF timestamp is ,
MIN=(5,14,ZD,EDIT=(TTTT/TT/TT TT:TT:TT)),/,
3:Latest SMF timestamp is
MAX=(5,14,ZD,EDIT=(TTTT/TT/TT TT:TT:TT)))
This example illustrates how the earliest and latest timestamps from a set of SMF
records can be displayed.
The OUTREC statement uses the DT1 format to convert the SMF date in input
bytes 11-15 to a Z'yyyymmdd' value in bytes 5-12, and uses the TM1 format to
convert the SMF time in input bytes 7-10 to a Z'hhmmss' value in bytes 13-18.
The OUTFIL statement uses the Z'yyyymmddhhmmss' value created by OUTREC
in bytes 5-18 to determine the minimum (earliest) and maximum (latest)
timestamp, and displays those timestamps in a trailer record in the form
C'yyyy/mm/dd hh:mm:ss'.
The report might look as follows:
Earliest SMF timestamp is
Latest SMF timestamp is
2001/01/09 10:27:04
2001/04/24 06:13:22
Example 16
SORT FIELDS=(1,20,BI,A)
OUTFIL FNAMES=FUPPER,OUTREC=(1,80,TRAN=LTOU)
OUTFIL FNAMES=FHEX,OUTREC=(1,80,HEX)
This example illustrates two types of conversion for fixed length records: lowercase
to uppercase conversion and hex conversion.
The input data set has RECFM = FB and LRECL = 80.
The first OUTFIL statement creates the data set associated with ddname FUPPER.
This data set will have RECFM = FB and LRECL = 80. All of the lowercase
EBCDIC characters (a-z) from byte 1 to byte 80 will be converted to uppercase
EBCDIC characters (A-Z). Other characters will remain unchanged. For example,
the characters 'san jose, ca 95193' will be converted to 'SAN JOSE, CA 95193'.
The second OUTFIL statement creates the data set associated with ddname FHEX.
This data set will have RECFM = FB and LRECL = 160 (2 * 80 data bytes). Each
byte from 1 to 80 will be converted to the two bytes representing its hex value. For
example, the three characters 'A12' will be converted to the six characters 'C1F1F2'.
Example 17
OPTION COPY
OUTFIL FNAMES=VUPPER,OUTREC=(1,4,5,TRAN=UTOL)
OUTFIL FNAMES=VHEX,OUTREC=(1,4,5,HEX)
This example illustrates two types of conversion for variable-length records:

uppercase to lowercase conversion and hex conversion.
389

The input data set has RECFM = VB and LRECL = 5000.
The first OUTFIL statement creates the data set associated with ddname VUPPER.
This data set will have RECFM = VB and LRECL = 5000. All of the uppercase
EBCDIC characters (A-Z) from bytes 5 (after the RDW) to the end of each record
will be converted to lowercase EBCDIC characters (a-z). Other characters will
remain unchanged. For example, the characters 'SAN JOSE, CA 95193' will be
converted to 'san jose, ca 95193'.
The second OUTFIL statement creates the data set associated with ddname VHEX.
This data set will have RECFM = VB and LRECL = 9996 (4 for RDW plus 2 * 4996
data bytes). Each byte from 5 (after the RDW) to the end of each record will be
converted to the two bytes representing its hex value. For example, the three
characters 'A12' will be converted to the six characters 'C1F1F2'.
Example 18
OUTFIL FNAMES=SAMP1,REMOVECC,HEADER1=(Sample 1,2/),
STARTREC=120,SAMPLE=(20,4),
OUTREC=(1,8,9,5,PD,ZD,C***,14,50)
OUTFIL FNAMES=SAMP2,REMOVECC,HEADER1=(Sample 2,2/),
SAMPLE=1000,ENDREC=5001
This example illustrates how to take different "samples" of sorted output records.
Sorted records 120, 121, 122, 123, 140, 141, 142, 143, and so on will be reformatted
as indicated by the OUTREC parameter and written to the output data set
associated with SAMP1. The heading 'Sample 1' will appear before the sample
output records.
Sorted records 1, 1001, 2001, 3001, 4001 and 5001 will be written to the output data
set associated with SAMP2. The heading 'Sample 2' will appear before the sample
output records.
Example 19
OPTION COPY
OUTFIL FNAMES=R500,REPEAT=500
OUTFIL FNAMES=R100,OUTREC=(SEQNUM,4,ZD,80XFF),REPEAT=100
This example illustrates how one record can be used to generate many records.
The first OUTFIL statement writes each output record 500 times to the data set
associated with R500. Each set of 500 records will be identical.
The second OUTFIL statement writes each reformatted output record 100 times to
the data set associated with R100. Each set of 100 records will be identical except
for the sequence number. The 100 records written from the first output record will
have sequence numbers 1-100, the 100 records written from the second output
record will have sequence numbers 101-200, and so on.
Example 20
OPTION COPY
OUTFIL OMIT=(56,6,CH,EQ,C******),
OVERLAY=(121:SEQNUM,8,ZD,56:56,6,TRAN=UTOL)
This example illustrates how you can use the OVERLAY parameter with OUTFIL
to add sequence numbers at the end of your records, and to convert uppercase
EBCDIC characters to lowercase EBCDIC characters in certain columns, without
affecting the rest of the record.
390

The input data set has RECFM=FB and LRECL=120. The output data set will have
RECFM=FB and LRECL=128.
The OMIT parameter removes records which have asterisks in positions 56-61.
121:SEQNUM,8,ZD in the OVERLAY parameter adds an 8-byte sequence number
in positions 121-128 of every record. The LRECL is increased from 120 to 128 to
hold the sequence number.
56:56,6,TRAN=UTOL in the OVERLAY parameter converts uppercase EBCDIC
characters in positions 56-61 to lowercase EBCDIC characters in positions 56-61.
Only the two overlaid fields are changed; all of the other data in the records is
unaffected.
Example 21
OPTION COPY
OUTFIL FNAMES=RPT1,
HEADER2=(1:Type,9:Date,21:Time,/,
1:----,9:--------,21:------),
IFTHEN=(WHEN=(5,2,BI,EQ,X0001),
BUILD=(1,4,5:5,2,BI,M11,LENGTH=4,
13:8,4,DT1,25:15,4,TM1)),
13:12,8,DC1,25:12,8,TC1)),
13:17,8,DE1,25:17,8,TE1)),
IFTHEN=(WHEN=NONE,
13:Cn/a,25:Cn/a))
OUTFIL FNAMES=RPT2,
HEADER2=(1:Type,9:Date,21:Time,/,
1:----,9:-------,21:--------),
13:8,4,DT3,25:15,4,TM4)),
13:12,8,DC3,25:12,8,TC4)),
13:17,8,DE3,25:17,8,TE4)),
IFTHEN=(WHEN=NONE,
13:Cn/a,25:Cn/a))
This example illustrates how you can use IFTHEN clauses to reformat different
records in different ways.
The input data set has RECFM=VB and consists of several different types of input
records as follows:
v Type1: Has X'0001' in positions 5-6, a 4-byte SMF date in positions 8-11 and a
4-byte SMF time in positions 15-18.
v Type2: Has X'0002' in positions 5-6 and an 8byte TOD date and time in
positions 1219.
v Type3: Has X'0003' in positions 5-6 and an 8byte ETOD date and time in
positions 1724.
391

v Other types: Do not have X'0001', X'0002' or X'0003' in positions 5-6, and do not
have a date or time value.
The first OUTFIL statement produces a report with RECFM=VBA and LRECL=31
that might look like this:
Type
---0001
0003
0008
0002
0001
0004
Date
-------20040827
20040907
n/a
20040901
20040731
n/a
Time
-----124531
230603
n/a
061559
152201
n/a
The first IFTHEN clause operates only against Type1 records; it converts the SMF
date using DT1 and the SMF time using TM1. The second IFTHEN clause only
operates against Type2 records; it converts the TOD date using DC1 and the TOD
time using TC1. The third IFTHEN clause only operates against Type3 records; it
converts the ETOD date using DE1 and the ETOD time using TE1. The fourth
IFTHEN clause operates against all other types of records; it uses 'n/a' for the date
and time.
The second OUTFIL statement produces a report with RECFM=VBA and
LRECL=33 that looks like this:
Type
---0001
0003
0008
0002
0001
0004
Date
------2004240
2004251
n/a
2004245
2004213
n/a
Time
-------12453184
23060373
n/a
06155903
15220150
n/a
The first IFTHEN clause operates only against Type1 records; it converts the SMF
date using DT3 and the SMF time using TM4. The second IFTHEN clause operates
only against Type2 records; it converts the TOD date using DC3 and the TOD time
using TC4. The third IFTHEN clause operates only against Type3 records; it
converts the ETOD date using DE3 and the ETOD time using TE4. The fourth
IFTHEN clause operates against all other types of records; it uses 'n/a' for the date
and time.
Example 22
OPTION COPY
OUTFIL BUILD=(1,80,SQZ=(SHIFT=LEFT,PAIR=APOST,MID=C,))
This example illustrates how you can create FB output records with comma
separated values from FB input records containing fields in fixed positions.
The 80-byte FB input records might look like this:
John Lewis
Ted Blank
Marilyn Carlson
Rex Otis
-12.83 Research
+128.37 Manufacturing
-282.83 Technical Support
+2.83 Marketing
Note that the data has three fields in fixed positions. The first field is a character
string surrounded by apostrophes. The second field is a numeric value. The third
field is another character string surrounded by apostrophes.
392

The 80-byte FB output records are in the form of comma separated values as
follows:
John Lewis,-12.83,Research
Ted Blank,+128.37,Manufacturing
Marilyn Carlson,-282.83,Technical Support
Rex Otis,+2.83,Marketing
We use OUTFIL BUILD to build the output records with comma separated values.
We use SQZ to squeeze out the blanks between the fields, shift the remaining
characters to the left and insert commas between the fields. SHIFT=LEFT shifts the
characters to the left. PAIR=APOST ensures that blanks within paired apostrophes
are not squeezed out (for example, we want to keep the blank in the 'John Lewis'
field.) MID=C',' inserts a comma for each group of blanks removed between the
fields (for example, between 'John Lewis' and -12.83).
Example 23
OPTION COPY
OUTFIL IFTHEN=(WHEN=INIT,
OVERLAY=(5:5,18,JFY=(SHIFT=LEFT,LEAD=C,TRAIL=C),
34:34,19,JFY=(SHIFT=LEFT,LEAD=C,TRAIL=C))),
IFTHEN=(WHEN=INIT,
BUILD=(1,4,5,60,SQZ=(SHIFT=LEFT,PAIR=APOST,MID=C,))),
VLTRIM=C
This example illustrates how you can create VB output records with comma
separated values from VB input records containing fields in fixed positions.
The 84-byte VB input records might look like this (4-byte RDW followed by data):
Length
45
50
54
51
Data
John Lewis
Ted Blank
Marilyn Carlson
Rex Otis
-12.83 Research
+128.37 Manufacturing
-282.83 Technical Support
+2.83 Marketing
Note that the data has three fields in fixed positions. The first field is a character
string. The second field is a numeric value. The third field is another character
string. (In the previous example, the character strings were surrounded by
apostrophes; in this example the apostrophes must be added around the character
strings.)
The 84-byte VB output records are in the form of comma separated values as
follows:
Length
34
39
49
32
Data
John Lewis,-12.83,Research
Ted Blank,+128.37,Manufacturing
Marilyn Carlson,-282.83,Technical Support
Rex Otis,+2.83,Marketing
We use OUTFIL IFTHEN to ensure that short records are padded on the right with
blanks. (OUTFIL BUILD would terminate due to the short records.) WHEN=INIT
reformats every record. OVERLAY surrounds the character strings with
apostrophes. BUILD builds the output records with comma separated values.
We use JFY to surround the character strings with apostrophes without removing
embedded blanks (for example, John Lewis is changed to 'John Lewis'). After the
first IFTHEN, the records look like this:
393

Length
52
52
54
52
Data
John Lewis
Ted Blank
Marilyn Carlson
Rex Otis
-12.83
+128.37
-282.83
+2.83
Research
Manufacturing
Technical Support
Marketing
We then use SQZ to squeeze out the blanks between the fields, shift the remaining
characters to the left and insert commas between the fields. SHIFT=LEFT shifts the
characters to the left. PAIR=APOST ensures that blanks within paired apostrophes
are not squeezed out (for example, we want to keep the blank in the 'John Lewis'
field.) MID=C',' inserts a comma for each group of blanks removed between the
fields (for example, between 'John Lewis' and -12.83).
Finally, we use VLRTIM=C' ' to remove trailing blanks at the end of each VB
record by adjusting its length appropriately.
Example 24
OPTION COPY
OUTFIL REMOVECC,
HEADER1=(C<?xml version="1.0"?>,/,
3:C<booklist>),
BUILD=(5:C<book>,/,
7:1,20,JFY=(SHIFT=LEFT,LEAD=C<title>,TRAIL=C</title>,
LENGTH=36),/,
7:24,15,SQZ=(SHIFT=LEFT,LEAD=C<author>,MID=C, ,
TRAIL=C</author>,LENGTH=33),/,
5:C</book>),
TRAILER1=(3:C</booklist>)
This example illustrates how you can generate XML statements from FB input
records containing fields in fixed positions.
Modern Poetry
Intro to Computers
Marketing
Friedman
KR
Chatterjee CL
Maxwell
G
Note that the data has three character fields in fixed positions.
The 42-byte FB output records look like this:
<?xml version="1.0"?>
<booklist>
<book>
<title>Modern Poetry</title>
<author>Friedman, KR</author>
</book>
<book>
<title>Intro to Computers</title>
<author>Chatterjee, CL</author>
</book>
<book>
<title>Marketing</title>
<author>Maxwell, G</author>
</book>
</booklist>
We use OUTFIL HEADER1 to generate the xml and booklist starting tags that
precede the set of tags for each record. We use OUTFIL BUILD to generate the set
of tags for each record as follows:
v A constant is used to generate the book starting tag.
394

v JFY is used to generate the title tags and data from the first input field. LEAD
generates the title starting tag before the input field from the record. TRAIL
generates the title ending tag after the last nonblank character from the input
field. JFY keeps embedded blanks between the first nonblank character and the
last nonblank character of the input field. LENGTH ensures that the addition of
the LEAD and TRAIL strings does not cause truncation by increasing the output
length to 36 bytes (overriding the default of 20 bytes from the input field).
v SQZ is used to generate the author tags and data from the second and third
input fields. LEAD generates the author starting tag before the input field from
the record. MID replaces the blanks between the second and third input fields
with a comma and one blank. TRAIL generates the author ending tag after the
last nonblank character from the input fields. LENGTH ensures that the addition
of the LEAD, MID and TRAIL strings does not cause truncation by increasing
the output length to 33 bytes (overriding the default of 15 bytes from the input
field).
We use OUTFIL TRAILER1 to generate the booklist ending tag that follows the set
of tags for each record.
Example 25
OPTION COPY
OUTFIL HEADER2=(1:First,13:Initial,22:Last,37:Score,/,
1:10-,13:7-,22:11-,35:7-),
PARSE=(%00=(STARTAFT=CLAST-> ,ENDBEFR=C ,FIXLEN=11),
%01=(STARTAFT=CFIRST-> ,ENDBEFR=C ,FIXLEN=11),
%02=(STARTAFT=CINITIAL-> ,ENDBEFR=C ,FIXLEN=7),
%03=(STARTAFT=CSCORE-> ,FIXLEN=7)),
BUILD=(1:%01,13:%02,22:%00,
35:%03,SFF,EDIT=(SIIIT.T),SIGNS=(,-))
This example illustrates how you can create a report from FB input records with
variable position/length fields, such as keyword delimited values.
LAST-> Clark
FIRST-> Oscar
INITIAL-> D
SCORE-> +98.2
LAST-> Roberts FIRST-> Harriet
INITIAL->
SCORE-> -152.6
LAST-> Stein FIRST-> Gertrude INITIAL-> V SCORE-> +5.1
Note that each record has four variable fields each identified by a specific
keyword. The fields do not start and end in the same position in every record and
they have different lengths in different records.
The output report has RECFM=FBA and LRECL=42 and looks like this:
First
Initial Last
---------- ------- ----------Oscar
D
Clark
Harriet
Roberts
Gertrude
V
Stein
Score
------98.2
-152.6
5.1
In order to create a report like this from variable position/length fields, we use
HEADER2 to create the page header, and PARSE and BUILD to create a fixed
parsed field from each variable field. We use %00 to create an 11-byte fixed parsed
field with the extracted 'LAST-> ' value. We use %01 to create an 11-byte fixed
parsed field with the extracted 'FIRST-> ' value. We use %02 to create a 7-byte
fixed parsed field with the extracted 'INITIAL-> ' value. We use %03 to create a
7-byte fixed parsed field with the extracted 'SCORE-> ' value. Since the fourth field
(%03) is extracted as signed and padded on the right with blanks, we treat it as
SFF format in order to edit it.
395
Example 26
INREC IFTHEN=(WHEN=INIT,
PARSE=(%1=(FIXLEN=10,STARTAFT=CFirst=",ENDBEFR=C"))),
IFTHEN=(WHEN=INIT,
PARSE=(%2=(FIXLEN=10,STARTAFT=CMiddle=",ENDBEFR=C"))),
IFTHEN=(WHEN=INIT,
PARSE=(%3=(FIXLEN=10,STARTAFT=CLast=",ENDBEFR=C"))),
IFTHEN=(WHEN=INIT,
PARSE=(%4=(FIXLEN=10,STARTAFT=CWife=",ENDBEFR=C")),
BUILD=(%1,13:%2,25:%3,37:%4))
SORT FIELDS=(25,10,CH,A,1,10,CH,A,13,10,CH,A)
OUTFIL HEADER2=(First,13:Middle,
25:CLast,37:Wife,/,10C-,13:10C-,
25:10C-,37:10C-)
This example illustrates how you can create a sorted report from FB input records
with keyword values that can occur in any order or not occur at all.
Last="Buchanan" First="James"
Wife="Louisa" First="John" Middle="Quincy" Last="Adams"
First="George" Last="Washington" Wife="Martha"
Last="Clinton" Wife="Hillary" Middle="Jefferson" First="William"
First="John" Wife="Abigail" Last="Adams"
Note that each record has up to four variable fields each identified by a specific
keyword (First, Middle, Last, Wife). In each record, the fields can be in any order,
do not start and end in the same position and can have different lengths.
We want to sort the records by last name, first name and middle name, and create
a report with fixed headings and values for the first name, middle name, last name
and wife's name that looks like this:
First
Middle
---------- ---------John
John
Quincy
James
William
Jefferson
George
Last
---------Adams
Adams
Buchanan
Clinton
Washington
Wife
---------Abigail
Louisa
Hillary
Martha
In order to extract and sort variable fields like these, we use a separate IFTHEN
PARSE clause for each keyword. This allows us to find a particular keyword
anywhere in the record or ignore it if it is not in the record. If we used PARSE
instead of IFTHEN PARSE, a missing keyword would cause any keywords that
followed to be ignored. But because each IFTHEN PARSE starts scanning at
position 1 by default, we can look for each keyword independently of the others. If
a keyword is missing, the parsed field is set to all blanks (for example, %2 and %4
are set to blanks for the James Buchanan record).
We use %1 to create a 10-byte fixed parsed field into which we extract the first
name. We use %2 to create a 10-byte fixed parsed field into which we extract the
middle name. We use %3 to create a 10-byte fixed parsed field into which we
extract the last name. We use %4 to create a 10-byte fixed parsed field into which
we extract the wife's name. We use BUILD to reformat the records to contain the
fixed parsed fields. Then we SORT on the fixed parsed fields. Finally, we use
OUTFIL to create the headings.
396
Example 27
OPTION COPY
OUTFIL FINDREP=(INOUT=(CAM,CIN THE MORNING,
CPM,CIN THE EVENING),MAXLEN=70)
This example illustrates how you can replace values in FB or VB records with
larger values and shift the rest of the bytes to the right.
The 40 byte FB input records might look like this:
COFFEE AT 12:28 AM, TOAST AT 06:15 PM
MILK AT 03:17 PM, BAGELS AT 05:03 PM
PUDDING AT 09:32 AM
We want to replace each instance of 'AM' with 'IN THE MORNING' and each
instance of 'PM' with 'IN THE EVENING' and shift the bytes after 'AM' or 'PM' to
the right. We use INOUT to indicate find and replace pairs. Since replacing the
smaller values with the larger values can cause the remaining bytes to be shifted
beyond the end of the 40 byte record, we use MAXLEN to set the output record to
70 bytes to allow for the expansion, overriding the default of using the input
length for the output record.
The 70 byte FB output records look like this:
COFFEE AT 12:28 IN THE MORNING, TOAST AT 06:15 IN THE EVENING
MILK AT 03:17 IN THE EVENING, BAGELS AT 05:03 IN THE EVENING
PUDDING AT 09:32 IN THE MORNING
Example 28
OPTION COPY
OUTREC IFTHEN=(WHEN=INIT,BUILD=(1,4,8:5)),
IFTHEN=(WHEN=GROUP,BEGIN=(8,5,CH,EQ,CPAGE:),
PUSH=(5:SEQ=3))
OUTFIL INCLUDE=(5,3,ZD,EQ,2,OR,5,3,ZD,EQ,3),
BUILD=(1,4,5:8)
This example illustrates how you can INCLUDE specific relative records from
groups of VB records. We insert a sequence number between the RDW and the
first data byte of each record. The sequence number restarts at 1 for the first record
in each group. We INCLUDE on the sequence number and then remove it.
Len|Data
12|PAGE: 1
22|LINE 1 OF REPORT A
|...
12|PAGE: 2
|...
12|PAGE: 3
|...
397

We use an IFTHEN WHEN=INIT clause to reformat each record so it has room for
the 3-byte sequence number between the RDW and the first data byte. After the
WHEN=INIT clause is executed, the intermediate records look like this:
Len|Data
15|
PAGE: 1
25|
LINE 1 OF REPORT A
25|
LINE 2 OF REPORT A
25|
LINE 3 OF REPORT A
25|
LINE 4 OF REPORT A
|
...
15|
PAGE: 2
26|
LINE 66 OF REPORT A
26|
LINE 67 OF REPORT A
26|
LINE 68 OF REPORT A
26|
LINE 69 OF REPORT A
|
...
15|
PAGE: 3
27|
LINE 131 OF REPORT A
27|
27|
27|
Note that positions 5-7 are blank and the 'PAGE:' characters have been shifted over
to positions 8-12.
We use an IFTHEN WHEN=GROUP clause to put a 3-byte sequence number in
each record. BEGIN indicates a group starts with a record that has 'PAGE:' in
positions 8-12. PUSH overlays a 3-byte sequence number at positions 5-7 of each
record. The sequence number starts at 1 for the first record of a group and is
incremented by 1 for each subsequent record of the group. After the IFTHEN
GROUP clause is executed, the intermediate records look like this:
Len|Data
15|001PAGE: 1
25|002LINE 1 OF REPORT A
|006...
15|001PAGE: 2
|006...
15|001PAGE: 3
|006...
Note that the records in each group are numbered starting with 001 for the first
record of each group. The records we want have sequence numbers 002 and 003.
We use an OUTFIL statement to only INCLUDE the records with sequence number
002 or 003, and to remove the sequence numbers so the included output records
will be identical to the input records. After the OUTFIL statement is executed, the
final output records look like this:
Len|Data
398

23|LINE
23|LINE
24|LINE
24|LINE
66 OF REPORT A
67 OF REPORT A
131 OF REPORT A
132 OF REPORT A
Example 29
OPTION COPY
OUTFIL FNAMES=OUT1,INCLUDE=(11,3,CH,EQ,CD51),ACCEPT=3
OUTFIL FNAMES=OUT2,STARTREC=2,ACCEPT=5
This example illustrates how you can stop processing OUTFIL input records for
various groups in different ways.
The input data set has these records:
HEADER 2010/06/30
FRANK
D51
ED
D52
VICKY
D51
MARTIN
D52
LILY
D50
MARC
D51
JUNE
D51
LUCY
D51
TRAILER
8
The first OUTFIL statement uses INCLUDE to extract the D51 records and
ACCEPT=3 to only write the first three D51 records to OUT1. Thus, OUT1 will
have these records:
FRANK
VICKY
MARC
D51
D51
D51
The second OUTFIL statement uses STARTREC=2 to skip the HEADER record and
ACCEPT=5 to only write the next five records to OUT2. Thus, OUT2 will have
these records:
FRANK
ED
VICKY
MARTIN
LILY
D51
D52
D51
D52
D50
Example 30
OPTION COPY
OUTFIL FNAMES=OUT1,
INCLUDE=(3,4,CH,EQ,Ckey1),
IFTRAIL=(HD=YES,TRLID=(1,1,CH,EQ,CT),
14:TOT=(10,4,ZD,TO=ZD,LENGTH=6)))
OUTFIL FNAMES=OUT2,
INCLUDE=(3,4,CH,EQ,Ckey2),
IFTRAIL=(HD=YES,TRLID=(1,1,CH,EQ,CT),
14:TOT=(10,4,ZD,TO=ZD,LENGTH=6)))
This example illustrates how you can split an input file with a header, and a trailer
containing a count and total for the input data records, into two output files each
of which has the header and a trailer with accurate count and total for the
included data records.
The input data set has RECFM=FB and LRECL=80 with these input records:
399

H
D
D
D
D
D
D
D
T
2010/07/06
key1
0100
key2
0200
key2
0118
key1
0150
key1
0025
key2
1000
key2
0310
X 0007 QR 001903
D52-007-321-7526
The first record is a header record with a date. The records with D in position 1
are data records; they have key1 or key2 in positions 3-6 and an amount field in
positions 10-13.
The trailer record is identified by a 'T' in position 1. Positions 6-9 contain a count
(0007) of the data records, and positions 14-19 contain a total (001903) for the
amount fields in the data records. In addition, there are other static fields in the
trailer record that must not be changed. Note that IFTRAIL with HD=YES does not
process the header or trailer records as data records, so they are not included in
the count or total.
The first OUTFIL statement writes the header record, data records with key1, and
trailer record, in the OUT1 data set. The IFTRAIL operand is used to identify the
trailer (TRLID), indicate the first record is a header (HD=YES), and update the
count and total in the trailer (TRLUPD) to reflect the input data records included
in this output data set.
OUT1 will have these records:
H
D
D
D
T
2010/07/06
key1
0100
key1
0150
key1
0025
X 0003 QR 000275
D52-007-321-7526
The second OUTFIL statement writes the header record, data records with key2,
and trailer record, in the OUT2 data set. The IFTRAIL operand is used to identify
the trailer (TRLID), indicate the first record is a header (HD=YES), and update the
count and total in the trailer (TRLUPD) to reflect the input data records included
in this output data set. Note that IFTRAIL with HD=YES does not process the
header or trailer records as data records, so they are not included in the count or
total.
OUT2 will have these records:
H
D
D
D
D
T
2010/07/06
key2
0200
key2
0118
key2
1000
key2
0310
X 0004 QR 001628
D52-007-321-7526
Example 31
OPTION Y2PAST=1990
SORT FIELDS=(1,6,Y2W,A)
OUTFIL REMOVECC,
HEADER1=(1:Input,15:NEXTDFRI,25:PREVDSUN,35:LASTDAYQ),
BUILD=(1:1,6,Y2W,TOJUL=Y4W(-),
15:1,6,Y2W,NEXTDFRI,TOJUL=Y4W(-),
25:1,6,Y2W,PREVDSUN,TOJUL=Y4W(-),
35:1,6,Y2W,LASTDAYQ,TOJUL=Y4W(-))
400

This example illustrates how you can sort by a date, and calculate a specific day
after and before a date, and the last day of the quarter for a date. The input date is
in the form C'mmddyy' and the output dates will be in the form 'ddd-yyyy'.
The SORTIN data set has these input records with a C'mmddyy' date field in
positions 1-6:
010105
120699
021610
999999
092810
031500
000000
032505
110210
The SORT statement sorts by the C'mmddyy' date. We use 1,6,Y2W for the sort
field to match the C'mmddyy' date.
The OUTFIL statement writes a heading and performs four operations on each
date. It converts the input date to C'ddd-yyyy' form, gets the next Friday date in
C'ddd-yyyy' form, gets the previous Sunday date in C'ddd-yyyy' form, and gets
the end of the quarter date in C'ddd-yyyy' form. We use 1,6,Y2W for the input
field to match the C'mmddyy' date, and we use TOJUL=Y4W(-) to give us a
C'ddd-yyyy' output date.
Input
000-0000
340-1999
075-2000
001-2005
084-2005
047-2010
271-2010
306-2010
999-9999
NEXTDFRI PREVDSUN LASTDAYQ

000-0000 000-0000 000-0000
344-1999 339-1999 365-1999
077-2000 072-2000 091-2000
007-2005 361-2004 090-2005
091-2005 079-2005 090-2005
050-2010 045-2010 090-2010
274-2010 269-2010 273-2010
309-2010 304-2010 365-2010
999-9999 999-9999 999-9999
Note that the '000000' and '999999' input values are treated as special indicators for
output.
401
OUTREC Control Statement
OUTREC control statement

,
OUTREC
PARSE=( definition
,
FIELDS=
BUILD=
( item
,
OVERLAY=( item
,
FINDREP=( item
,
IFTHEN=(clause)
IFOUTLEN=n
The OUTREC control statement allows you to reformat the input records after they
are sorted, merged or copied.
The OUTREC control statement supports a wide variety of parsing, editing, and
reformatting tasks, including:
records.
v Sophisticated editing capabilities, such as control of the way numeric fields are
so on.
editing patterns are available via the user-defined editing masks.
v Transformation of SMF, TOD, and ETOD date and time values to more usable
forms.
v Various types of arithmetic operations for input date fields.
402

v Conversion of input date fields of one type (CH, ZD, PD, 2-digit year, 4-digit
lookup table, based on a character, hexadecimal, or bit string as input (that is,
lookup and change).
You can create the reformatted OUTREC records in one of the following three
ways using unedited, edited, or converted input fields (p,m for fixed fields, or %nn
for parsed fields - see PARSE), and a variety of constants:
v BUILD or FIELDS: Reformat each record by specifying all of its items one by
one. Build gives you complete control over the items you want in your
reformatted OUTREC records and the order in which they appear. You can
delete, rearrange and insert fields and constants. Example:
OUTREC BUILD=(1,20,CABC,26:5C*,
15,3,PD,EDIT=(TTT.TT),21,30,80:X)
OUTREC OVERLAY=(45:45,8,TRAN=LTOU)
OUTREC FINDREP=(IN=CMr.,OUT=CMister)
v IFTHEN clauses: Reformat different records in different ways by specifying how

build, overlay, find/replace, or group operation items are applied to records that
meet given criteria. IFTHEN clauses let you use sophisticated conditional logic
to choose how different record types are reformatted. Example:
OUTREC IFTHEN=(WHEN=(1,5,CH,EQ,CTYPE1),
BUILD=(1,40,C**,+1,TO=PD)),
OUTREC records:
v Fixed position/length fields or variable position/length fields. For fixed fields,
you specify the starting position and length of the field directly. For variable
fields, such as delimited fields, comma separated values (CSV), tab separated
values, blank separated values, keyword separated fields, null-terminated strings
(and many other types), you define rules that allow DFSORT to extract the
relevant data into fixed parsed fields, and then use the parsed fields as you
would use fixed fields.
v Blanks, binary zeros, character strings, and hexadecimal strings
v Current date, future date, past date and current time in various forms
boundaries
v Hexadecimal or bit representations of binary input fields
v Left-justified, right-justified, left-squeezed, or right-squeezed input fields.
403

v Numeric input fields of various formats converted to different numeric formats,
or to character format edited to contain signs, thousands separators, decimal
points, leading zeros or no leading zeros, and so on.
v Decimal constants converted to different numeric formats, or to character format
edited to contain signs, thousands separators, decimal points, leading zeros or
converted to different numeric formats, or to character format edited to contain
signs, thousands separators, decimal points, leading zeros or no leading zeros,
and so on.
v SMF, TOD and ETOD date and time fields converted to different numeric
formats, or to character format edited to contain separators, leading zeros or no
Gregorian) converted to corresponding output date fields of another type or to a
corresponding day of the week.
v A character constant, hexadecimal constant or input field selected from a lookup
table, based on a character, hexadecimal or bit constant as input.
v A zoned decimal group identifier, a zoned decimal group sequence number, or a
field propagated from the first record of a group to all of the records of a group.
For information concerning the interaction of INREC and OUTREC, see INREC
statement notes on page 148 and OUTREC statement notes on page 424.
See OUTFIL control statements on page 223 for complete details on the OUTFIL
OUTREC parameter.
PARSE
404

,
,
PARSE=(
%n=
%nn=
%nnn=
%=
FIXLEN=m
ABSPOS=p
ADDPOS=x
SUBPOS=y
,
STARTAFT=string
STARTAFT=an
STARTAFT=BLANKS
STARTAT=string
STARTAT=an
STARTAT=BLANKS
STARTAT=NONBLANK
,
ENDBEFR=string
ENDBEFR=an
ENDBEFR=BLANKS
ENDAT=string
ENDAT=an
ENDAT=BLANKS
PAIR=APOST
PAIR=QUOTE
REPEAT=v
position/length fields (p,m) can be used in the BUILD (or FIELDS) or
want to extract.
See PARSE under "OUTFIL Control Statements" for complete details.
Sample Syntax:
OUTREC PARSE=(%200=(ENDBEFR=C*,FIXLEN=3),
405

BUILD=(%203,X,%203,HEX,21:%202,31:%201,SFF,M26,LENGTH=7,
18,6,%200,UFF,M11,LENGTH=3,%204,JFY=(SHIFT=RIGHT))
Default for PARSE: None; must be specified. See Appendix B,

details.
FIELDS or BUILD
,
FIELDS=
BUILD=
(
c:
s
p,m
,a
%nn
p
p,m
%nn
p
,TRAN
LTOU
UTOL
ALTSEQ
ATOE
ETOA
HEX
UNHEX
BIT
UNBIT
p,m
,HEX
%nn
p
p,m,f
,edit
%nn,f
,to
(p,m,f)
(%nn,f)
deccon
,edit
(deccon)
,to
arexp
,edit
(arexp)
,to
p,m
,Y2x
%nn
,Y4x
,edit
,to
,todate
,dateop
,Y2x(s)
,Y4x(s)
,Y2xP
p,m
,lookup
%nn
p,m
,justify
%nn
p,m
,squeeze
%nn
seqnum
Specifies all of the items in the reformatted OUTREC record in the order in
which they are to be included. The reformatted OUTREC record consists of the
for parsed fields - see PARSE), edited decimal constants, edited results of
406

For variable-length records, the first item in the BUILD or FIELDS parameter
position (p) as the last item in the BUILD or FIELDS parameter. For example:
OUTREC FIELDS=(1,4,
C|,
5)
unedited RDW
| separator
The BUILD or FIELDS parameter of the OUTREC statement differs from the
BUILD or OUTREC parameter of the OUTFIL statement in the following ways:
v The BUILD or FIELDS parameter of the OUTREC statement applies to all
input records; the BUILD or OUTREC parameter of the OUTFIL statement
only applies to the OUTFIL input records for its OUTFIL group.
v The BUILD or OUTREC parameter of the OUTFIL statement supports the
slash (/) separator for creating blank records and new records; the BUILD or
FIELDS parameter of the OUTREC statement does not.
c:
specifies the position (column) for a separation field, input field decimal
constant, arithmetic expression, or sequence number, relative to the start of
the reformatted output record. Unused space preceding the specified
column is padded with EBCDIC blanks. The following rules apply:
v c: must be followed by a separation field, input field, decimal constant,
or arithmetic expression.
v For variable-length records, c: must not be specified before the first input
field (the record descriptor word) nor after the variable part of the input
record.
See Table 29 on page 130 for examples of valid and invalid column
alignment.
s
specifies that a separation field (blanks, zeros, character string, hexadecimal

string, current date, future date, past date, or current time) is to appear in
the reformatted output record. It can be specified before or after any input
field. Consecutive separation fields may be specified. For variable-length
records, separation fields must not be specified before the first input field
(the record descriptor word) or after the variable part of the input record.
Permissible values are nX, nZ, nC'xx...x', nX'yy...yy', and various date and
time constants.
nX
Blank separation. n bytes of EBCDIC blanks (X'40') are to appear in the
reformatted output records. n can range from 1 to 4095. If n is omitted,
407

1 is used.
See Table 30 on page 131 for examples of valid and invalid blank
separation.
nZ
Binary zero separation. n bytes of binary zeros (X'00') are to appear in
the reformatted output records. n can range from 1 to 4095. If n is
omitted, 1 is used.
See Table 31 on page 131 for examples of valid and invalid binary zero
separation.
nC'xx...x'
constant (C'xx...x') are to appear in the reformatted output records. n
can range from 1 to 4095. If n is omitted, 1 is used. x can be any
EBCDIC character. You can specify from 1 to 256 characters.
If you want to include a single apostrophe in the character string, you
must specify it as two single apostrophes:
Required:
ONEILL
Specify:
CONEILL
See Table 32 on page 131 for examples of valid and invalid character
string separation.
nX'yy...yy'
Hexadecimal string separation. n repetitions of the hexadecimal string
constant (X'yy...yy') are to appear in the reformatted output records. n
can range from 1 to 4095. If n is omitted, 1 is used.
specify from 1 to 256 pairs of hexadecimal digits.
See Table 33 on page 132 for examples of valid and invalid
hexadecimal string separation.
DATEn, DATEn(c), DATEnP

Constant for current date. The date of the run is to appear in the
reformatted output records. See DATEn, DATEn(c), DATEnP under
OUTFIL OUTREC for details.
&DATEn, &DATEn(c), &DATEnP
can be used instead of DATEn, DATEn(c), and DATEnP, respectively.
Constant for future date. A future date relative to the current date of
the run is to appear in the reformatted output records. See DATEn+r,
DATEn(c)+r, DATEnP+r under OUTFIL OUTREC for details.
can be used instead of DATEn+r, DATEn(c)+r, and DATEnP+r,
respectively.
Constant for past date. A past date relative to the current date of the
run is to appear in the reformatted output records. See DATEn-r,
DATEn(c)-r, DATEnP-r under OUTFIL OUTREC for details.
408

Can be used instead of DATEn-r, DATEn(c)-r, and DATEnP-r,
respectively

reformatted output records. See TIMEn, TIMEn(c), TIMEnP under
OUTFIL OUTREC for details.
Can be used instead of TIMEn, TIMEn(c), and TIMEnP, respectively.
DATE
specifies that the current date is to appear in the reformatted output
records in the form 'mm/dd/yy'. See DATE under OUTFIL OUTREC
for details.
&DATE
DATE=(abcd)
records in the form 'adbdc'. See DATE=(abcd) under OUTFIL OUTREC
for details.
&DATE=(abcd)
DATENS=(abc)
records in the form 'abc'. See DATENS=(ab) under OUTFIL OUTREC
for details.
&DATENS=(abc)
YDDD=(abc)
records in the form 'acb'. See YDDD=(abc) under OUTFIL OUTREC for
details.
&YDDD=(abc)
YDDDNS=(ab)
records in the form 'ab'. See YDDDNS=(ab) under OUTFIL OUTREC
for details.
&YDDDNS=(ab)
TIME
specifies that the current time is to appear in the reformatted output
records in the form 'hh:mm:ss'. See TIME under OUTFIL OUTREC for
details.
&TIME
409

TIME=(abc)
records in the form 'hhcmmcss' (24-hour time) or 'hhcmmcss xx'
(12-hour time). See TIME=(abc) under OUTFIL OUTREC for details.
&TIME=(abc)
TIMENS=(ab)
record in the form 'hhmmss' (24-hour time) or 'hhmmss xx' (12-hour
time). See TIMENS=(ab) under OUTFIL OUTREC for details.
&TIMENS=(ab)
p,m,a
specifies that an unedited input field is to appear in the reformatted output
record.
p
input record.13 The first data byte of a fixed-length record has relative
position 1. The first data byte of a variable-length record has relative
position 5, because the first four bytes are occupied by the RDW. All fields
must start on a byte boundary, and no field may extend beyond byte
32752. See OUTREC statement notes on page 424 for special rules
specifies the length of the input field. It must include the sign if the data is
signed and must be a whole number of bytes. See OUTREC statement
notes on page 424 for more information.
specifies the alignment (displacement) of the input field in the reformatted

output record relative to the start of the reformatted output record.
The permissible values of a are:
H
Halfword aligned. The displacement (p-1) of the field from the

beginning of the reformatted input record, in bytes, is a multiple of 2
(that is, position 1, 3, 5, and so forth).
Fullword aligned. The displacement is a multiple of 4 (that is, position

1, 5, 9, and so forth).
Doubleword aligned. The displacement is a multiple of 8 (that is,

Alignment can be necessary if, for example, the data is to be used in a

COBOL application program where items are aligned through the
SYNCHRONIZED clause. Unused space preceding aligned fields are
always padded with binary zeros.
%nn
output record. See PARSE for details of parsed fields. See p,m,a for further
details. Note that alignment (H, F, D) is not permitted for %nn fields (for
example, %nn,F results in an error message and ermination).
13. If INREC is specified, p must refer to the record as reformatted by INREC. If your E15 user exit reformats the record, and
INREC is not specified, p must refer to the record as reformatted by your E15 user exit.
410

p
specifies that the unedited part of the variable input record (that part beyond
the minimum record length), is to appear in the reformatted output record, as
the last field. p without m can only be used for variable-length records; not for
fixed-length records.
Attention: If 1,4,p is specified (only RDW and variable part of record), null
records containing only an RDW will result if the record length is less than p.
record length (RECORD statement L4 value) plus 1 byte.
p,m,TRAN=keyword
Specifies that an input field is to be translated as indicated by the keyword.
See p,m,TRAN=keyword under OUTFIL OUTREC for details.
%nn,TRAN=keyword
keyword. See %nn,TRAN=keyword under OUTFIL OUTREC for details.
p,TRAN=keyword
indicated by the keyword. See p,TRAN=keyword under OUTFIL OUTREC for
details.
p,m,HEX
Can be used instead of p,m,TRAN=HEX. See p,m,HEX under OUTFIL
OUTREC for details.
%nn,HEX
Can be used instead of %nn,TRAN=HEX. See %nn,HEX under OUTFIL
OUTREC for details.
p,HEX
Can be used instead of p,TRAN=HEX. See p,HEX under OUTFIL OUTREC for
details.
output record. You can edit BI, FI, PD, PD0, ZD, FL, CSF, FS, UFF, SFF, DC1,
DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3, TC4, TE1, TE2,
TE3, TE4, TM1, TM2, TM3 or TM4 fields using either pre-defined edit masks
(M0-M26) or specific edit patterns you define. You can control the way the
edited fields look with respect to length, leading or suppressed zeros,
thousands separators, decimal points, leading and trailing positive and
negative signs, and so on.
See p,m,f,edit under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(5:(21,8,ZD),M19,X,46,5,ZD,M13,
31:35,4,FL,SIGNS=(,,+,-),LENGTH=10,
51:8,4,PD,EDIT=(**II,IIT.TTXS),SIGNS=(,,+,-))
reformatted output record. See PARSE for details of parsed fields. See
p,m,f,edit or (p,m,f),edit for further details.
specifies that a converted numeric input field is to appear in the reformatted
output record. You can convert BI, FI, PD, PD0, ZD, FL, CSF, FS, UFF, SFF,
411

DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2, DT3, TC1, TC2, TC3, TC4, TE1,
TE2, TE3, TE4, TM1, TM2, TM3, or TM4 fields to BI, FI, PD, PDC, PDF, ZD,
ZDF, ZDC, CSF, or FS fields.
See p,m,f,to under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(21,4,BI,TO=PDF,X,8,4,ZD,FI,LENGTH=2)
reformatted output record. See PARSE for details of parsed fields. See p,m,f,to
or (p,m,f),to for further details.
specifies that an edited decimal constant is to appear in the reformatted output
record. The decimal constant must be in the form +n or -n where n is 1 to 31
decimal digits. The sign (+ or -) must be specified. A decimal constant
produces a signed, 31-digit zoned decimal (ZD) result to be edited as specified.
See deccon,edit under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(-5000,EDIT=(-T,TT.T),21:(+0),M11,LENGTH=7)
output record. The decimal constant must be in the form +n or -n where n is 1
to 31 decimal digits. The sign (+ or -) must be specified. A decimal constant
produces a signed, 31-digit zoned decimal (ZD) result to be converted as
specified.
See deccon,to under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(+1,PD,LENGTH=6,(-1),PD,LENGTH=6,
-50,TO=ZD,LENGTH=8)
specifies that the edited result of an arithmetic expression is to appear in the
reformatted output record. The arithmetic expression can consist of input
fields, decimal constants, operators and parentheses. An arithmetic expression
produces a signed, 31-digit zoned decimal (ZD) result to be edited as specified.
See arexp,edit under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(C**,27,2,FI,SUB,
83,4,PD,EDIT=(STTTTTTT),SIGNS=(+,-),
25:(((15,5,ZD,MUL,+2),ADD,+100),MAX,62,2,PD),M25,LENGTH=12)
specifies that the converted result of an arithmetic expression is to appear in
the reformatted output record. The arithmetic expression can consist of input
fields, decimal constants, operators and parentheses. An arithmetic expression
produces a signed, 31-digit zoned decimal (ZD) result to be converted as
specified.
See arexp,to under OUTFIL OUTREC for details.
Sample Syntax:
412

OUTREC FIELDS=((15,6,FS,DIV,+5),ADD,(-1,ADD,36,6,FS),TO=FI,X,
3,2,FI,MAX,-6,LENGTH=4,TO=FS)
p,m,Y2x or p,m,Y4x
using the century window established by the Y2PAST option in effect. Y2x and
Y4x dates with special indicators are expanded appropriately (for example,
p,6,Y2T transforms C'000000' to C'00000000').
See "p,m,Y2x or p,m,Y4x" under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC BUILD=(21,3,Y2U,X,3,8,Y4W)
%nn,Y2x or %nn,Y4x
Specifies that a parsed input date field is to be edited. See PARSE for details of
parsed fields. See "p,m,Y2x or p,m,Y4x" for further details.
Specifies that the output for a p,m,Yxx input date field is to be edited
according to the edit parameters you specify. For example, if you specify:
OUTREC BUILD=(28,5,Y4V,EDIT=(TTTT-TT-TT)

%nn,Y2x,edit or %nn,Y4x,edit
Specifies that the output for a parsed Yxx input date field is to be edited
according to the edit parameters you specify. See PARSE for details of parsed
fields. See "p,m,Y2x,edit or p,m,Y4x,edit" for further details.
Specifies that an input date field is to be converted according to the to
parameters you specify. For example, if you specify:
OUTREC BUILD=(5,6,Y2W,TO=PD,LENGTH=5)
The C'mmddyy' date value will be transformed to a Z'mmddccyy' date value

and then converted to a P'mmddccyy' (X'0mmddccyyC') date value.
Specifies that a parsed input date field is to be converted according to the to
parameters you specify. See PARSE for details of parsed fields. See "p,m,Y2x,to
or p,m,Y4x,to" for further details.
corresponding output date field of another type. You can convert date fields
between combinations of 2-digit and 4-digit year dates, CH/ZD and PD dates,
and Julian and Gregorian dates. You can also convert a date field to a
corresponding day of the week in several forms.
See "p,m,Y2x,todate or p,m,Y4x,todate" under OUTFIL OUTREC for details.
Sample Syntax:
413

OUTREC BUILD=(21,3,Y2X,TOGREG=Y4T(/),X,
corresponding output date field of another type. See PARSE for details of
parsed fields. See "p,m,Y2x,todate or p,m,Y4x,todate" for further details.
Specifies an arithmetic operation for an input date field. See "p,m,Y2x,dateop
or p,m,Y4x,dateop" under OUTFIL OUTREC for details.
Specifies an arithmetic operation for a parsed input date field. See
"%nn,Y2x,dateop or %nn,Y4x,dateop" under OUTFIL OUTREC for details.
Specifies that an input date field is to be edited with separators. s can be any
character except a blank. Real Y2x dates are edited using the century window
indicators are expanded appropriately (for example, p,8,Y4T(s) transforms
C'00000000' to C'0000/00/00').
See "p,m,Y2x(s) or p,m,Y4x(s)" under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC BUILD=(19,7,Y4W(/),X,
43,5,Y4V(-))
PARSE for details of parsed fields. See "p,m,Y2x(s) or p,m,Y4x(s)" for further
details.
p,m,Y2xP
Specifies that an input date field is to be converted to a packed decimal output
date field. Real Y2x dates are edited using the century window established by
the Y2PAST option in effect. Y2x and Y4x dates with special indicators are
expanded appropriately (for example, p,6,Y2TP transforms C'000000' to
P'00000000').
See "p,m,Y2xP" under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC BUILD=(11,3,Y2XP,X,21,4,Y2WP)
%nn,Y2xP
Specifies that a parsed input date field is to be converted to a packed decimal
output date field. See PARSE for details of parsed fields. See "p,m,Y2xP" for
further details.
specifies that a character constant, hexadecimal constant, input field (p,m) or
parsed input field (%nn) from a lookup table is to appear in the reformatted
output record. You can use p,m,lookup or %nn,lookup to select a specified
character set constant (that is, a character or hexadecimal string) or set field
414

(that is, an input field or parsed input field) based on matching an input value
against find constants (that is, character, hexadecimal, or bit constants).
See p,m,lookup or %nn,lookup under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(11,1,
CHANGE=(6,
CR,CREAD,
CU,CUPDATE,
XFF,CEMPTY,
CA,CALTER),
NOMATCH=(11,6),
4X,
21,1,
CHANGE=(10,
B.1......,CVSAM,
p,m,justify
reformatted output record. For a left-justified field, leading blanks are removed
and the characters from the first nonblank to the last nonblank are shifted left,
with blanks inserted on the right if needed. For a right-justified field, trailing
blanks are removed and the characters from the last nonblank to the first
nonblank are shifted right, with blanks inserted on the left if needed.
Optionally:
See p,m,justify under OUTFIL OUTREC for details.
Sample Syntax:
21,20,JFY=(SHIFT=LEFT),5X,
52,12,JFY=(SHIFT=RIGHT,PREBLANK=C+,LEAD=C$,TRAIL=C.00))
%nn,justify
specifies that a left-justified or right-justified parsed input field is to appear in
the reformatted output record. See PARSE for details of parsed fields. See
p,m,justify for further details.
p,m,squeeze
specifies that a left-squeezed or right-squeezed input field is to appear in the
reformatted output record. For a left-squeezed field, all blanks are removed
and the characters from the first nonblank to the last nonblank are shifted left,
with blanks inserted on the right if needed. For a right-justified field, all blanks
are removed and the characters from the last nonblank to the first nonblank
are shifted right, with blanks inserted on the left if needed. Optionally:
v specific characters can be changed to blanks before squeezing begins
v a string (for example, a comma delimiter) can be inserted wherever a group
of blanks is removed between the first nonblank and the last nonblank.
v blanks can be kept as is between paired apostrophes ('AB CD EF') or paired
quotes ("AB CD EF")
415

See p,m,squeeze under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(21,20,SQZ=(SHIFT=RIGHT),5X,
152,18,SQZ=(SHIFT=LEFT,PREBLANK=X05,
LEAD=CVOL=SER=,MID=C,,PAIR=QUOTE,LENGTH=30))
%nn,squeeze
specifies that a left-squeezed or right-squeezed parsed input field is to appear
in the reformatted output record. See PARSE for details of parsed fields. See
p,m,squeeze for further details.
seqnum
specifies that a sequence number is to appear in the reformatted output record.
The sequence numbers are assigned in the order in which the records are
received for OUTREC processing. You can create BI, PD, ZD, CSF, or FS
sequence numbers and control their lengths, starting values and increment
values. You can restart the sequence number at the start value each time a
specified input field (p,m) or parsed input field (%nn) changes.
See seqnum under OUTFIL OUTREC for details.
Sample Syntax:
OUTREC FIELDS=(SEQNUM,6,ZD,START=1000,INCR=50,RESTART=(21,5),1,60)
Default for BUILD or FIELDS: None; must be specified. See Appendix B,

details.
OVERLAY
416

,
OVERLAY= (
c:
s
p,m
,a
%nn
p,m
%nn
,TRAN
LTOU
UTOL
ALTSEQ
ATOE
ETOA
HEX
UNHEX
BIT
UNBIT
p,m
,HEX
%nn
p,m,f
,edit
%nn,f
,to
(p,m,f)
(%nn,f)
deccon
,edit
(deccon)
,to
arexp
,edit
(arexp)
,to
p,m
,Y2x
%nn
,Y4x
,edit
,to
,todate
,dateop
,Y2x(s)
,Y4x(s)
,Y2xP
p,m
,lookup
%nn
p,m
,justify
%nn
p,m
,squeeze
%nn
seqnum

rearrange, or delete fields, use BUILD or FIELDS rather than OVERLAY. Use
record. OVERLAY can be easier to use then BUILD or FIELDS when you just
OUTREC OVERLAY=(25,2,11:CA,15,3,C**)
417

and C'**' is placed at output positions 15-16. The rest of the record remains
unchanged.
OUTREC OVERLAY=(21:8,4,ZD,MUL,+10,TO=ZD,LENGTH=6,
5:5,1,TRAN=UTOL,
variable-length records, the RDW length is also increased to correspond to the
For example, if your input record has a length of 40 and you specify:
OUTREC OVERLAY=(16:CABC,51:5C*,35:15,2)
the output record is given a length of 55. Blanks are filled in from columns
41-50. For variable-length records, the length in the RDW is changed from 40
to 55 after all of the OVERLAY items are processed.
The OVERLAY parameter of the OUTREC statement applies to all input
records whereas the OVERLAY parameter of the OUTFIL statement only
applies to the OUTFIL input records for its OUTFIL group.
See OUTREC FIELDS for details of the items listed in the OVERLAY syntax
diagram shown previously in this section. You can specify all of the items for
OVERLAY in the same way that you can specify them for BUILD or FIELDS
with the following exceptions:
v For p,m,H or p,m,F or p,m,D specified for OVERLAY, fields are aligned as
necessary without changing the preceding bytes.
for OVERLAY, so be sure to specify the first column (c:) as 5 or greater. If
you do not specify the first column, it will default to 1: which is invalid for
variable-length records with OVERLAY. Whereas FIELDS=(1,m,...) is
required, OVERLAY=(1,m) is not allowed since it would overlay the RDW.
Sample Syntax:
OUTREC OVERLAY=(21:21,4,ZD,TO=PD,LENGTH=4,
2:5,8,HEX,45:C*,32,4,C*,81:SEQNUM,5,ZD)

OUTREC OVERLAY=(5:X0001,28:CDate is ,YDDDNS=(4D),
17:5C*)
Default for OVERLAY: None; must be specified.
418

FINDREP
FINDREP
IN=incon,OUT=outcon
,
IN=( incon
,
,OUT=outcon
)

,
STARTPOS=p
ENDPOS=q
DO=n
MAXLEN=n
OVERRUN=ERROR
OVERRUN=TRUNC
SHIFT=YES
SHIFT=NO
See FINDREP under "OUTFIL Control Statements" for complete details.
Sample Syntax:
OUTREC FINDREP=(INOUT=(CGoodbye,CBye,CHello,CHi))
Default for FINDREP: None; must be specified. See Appendix B,

"Specification/Override of DFSORT Options".
IFTHEN
419

,
IFTHEN=(
WHEN=INIT
,BUILD=(items)
,OVERLAY=(items)
,FINDREP=(items)
,KEYBEGIN=(p,m)
,END=(logexp)
,RECORDS=n
,BUILD=(items)
,HIT=NEXT
,OVERLAY=(items)
,FINDREP=(items)
,BUILD=(items)
,HIT=NEXT
,OVERLAY=(items)
,FINDREP=(items)
,BUILD=(items)
,OVERLAY=(items)
,FINDREP=(items)

reformatted.
record, use BUILD or FIELDS rather than IFTHEN. If you want to do find and
replace operations in the same way for every record, use FINDREP rather than
IFTHEN. Use IFTHEN clauses if you want to insert, rearrange, delete or
overlay fields, or perform find/replace operations in different ways for
different records, or if you want to perform operations on groups of records.
v WHEN=INIT: Use one or more WHEN=INIT clauses to apply build, overlay
clauses.
evaluates as true.
420

v WHEN=INIT clauses or WHEN=GROUP clauses
v WHEN=NONE clauses
not specified.
Example:
OUTREC IFTHEN=(WHEN=(12,1,BI,ALL,X3F),OVERLAY=(18:CYes)),

processing stops.
processing stops.
processing stops.
processing stops.
OUTREC IFTHEN=(WHEN=INIT,OVERLAY=(8:8,4,ZD,ADD,+1,TO=ZD,LENGTH=4)),
421

DFSORT sets an appropriate LRECL (or reformatted record length if the
OUTREC record is further modified) for the output records based on the build,
overlay, find/replace and group operation items specified by the IFTHEN
clauses. However, DFSORT does not analyze the possible results of
you use OUTREC IFTHEN clauses, you can override the OUTREC LRECL
determined by DFSORT with the OUTREC IFOUTLEN parameter.
input is:
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
A
B
B
C
A
C
B
D
1
1
2
1
2
2
3
1
and you specify:

OUTREC IFTHEN=(WHEN=(8,1,CH,EQ,CA),OVERLAY=(15:SEQNUM,4,ZD)),

RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
RECORD
A
B
B
C
A
C
B
D
1
1
2
1
2
2
3
1
0001
0001
0002
0001
0002
0002
0003
0003
The IFTHEN clauses of the OUTREC statement apply to all input records
whereas the IFTHEN clauses of the OUTFIL statement only apply to the
WHEN=INIT clause
See "WHEN=INIT clause" under OUTFIL IFTHEN for details. Note that /
cannot be used to create blank records or new records.
Sample Syntax:
422

OUTREC IFTHEN=(WHEN=INIT,
BUILD=(1,20,21:CDepartment,31:3X,21,60)),
WHEN=GROUP clause
See "WHEN=GROUP clause" under OUTFIL IFTHEN for details.
Sample Syntax
OUTREC IFTHEN=(WHEN=GROUP,
PUSH=(41:ID=5))
See "WHEN=(logexp) clause" under OUTFIL IFTHEN for details. Note that
although / can be used to create blank records and new records with
OUTFIL, it cannot be used with OUTREC.
Sample Syntax:
OUTREC IFTHEN=(WHEN=(1,3,CH,EQ,CT01,AND,
BUILD=(1,21,42,13)),
BUILD=(1,25,42,13))
WHEN=ANY clause
See "WHEN=ANY clause" under OUTFIL IFTHEN for details. Note that
Sample Syntax:
OUTREC IFTHEN=(WHEN=(1,3,SS,EQ,CT01,T02,T03),
IFTHEN=(WHEN=ANY,OVERLAY=(16:CGroup Found)
WHEN=NONE clause
See "WHEN=NONE clause" under OUTFIL IFTHEN for details. Note that
Sample Syntax:
OUTREC IFTHEN=(WHEN=INIT,BUILD=(1,20,21:CDepartment,31:3X,21,60)),
Default for IFTHEN clauses: None; must be specified.

IFOUTLEN
IFOUTLEN=n
Overrides the OUTREC LRECL (or reformatted record length if the OUTREC
record is further modified) determined by DFSORT from your OUTREC
423

IFTHEN clauses. DFSORT sets an appropriate LRECL for the output records
based on the build, overlay, find/replace and group operation items specified
by the IFTHEN clauses. However, DFSORT does not analyze the possible
results of WHEN=(logexp) conditions when determining an appropriate
OUTREC LRECL. When you use OUTREC IFTHEN clauses, you can override
the OUTREC LRECL determined by DFSORT with the OUTREC IFOUTLEN
parameter.
n
specifies the length to use for the OUTREC LRECL (or for the reformatted
record length if the OUTREC record is further modified) . The value for n
must be between 1 and 32767, but must not be larger than the maximum
LRECL allowed for the RECFM, and must not conflict with the specified or
retrieved LRECL for the fixed-length output data set.
Sample Syntax:
OUTREC IFOUTLEN=70,
OUTREC statement notes

v
v If input records are reformatted by INREC or E15, OUTREC must refer to fields
in the appropriate reformatted record (see INREC statement notes on page
148).
v When you specify OUTREC, you must be aware of the change in record size and
layout of the resulting reformatted output records.
v If the SORTOUT LRECL is specified or available, DFSORT will use it even if it
does not match the reformatted OUTREC record length; this can cause padding
or truncation of the reformatted OUTREC records, or termination. If the
SORTOUT LRECL is not specified or available, DFSORT can automatically use
the reformatted OUTREC record length as the SORTOUT LRECL, when
appropriate. See the discussion of the SOLRF and NOSOLRF options in
OPTION control statement on page 173 for details.
For VSAM data sets, the maximum record size defined in the cluster is
equivalent to the LRECL when processing fixed-length records, and is four bytes
more than the LRECL when processing variable-length records. See VSAM
considerations on page 15 for more information.
v For variable-length records, the first entry in the FIELDS, BUILD, or IFTHEN
BUILD parameter must specify or include the unedited 4-byte RDW, that is, the
first field must be 1,4 or 1,m with m greater than 4. DFSORT sets the length of
the reformatted record in the RDW.
the reformatted record immediately following the RDW, the entry in the FIELDS,
BUILD, or IFTHEN BUILD parameter can specify both RDW and data field in
one. (1,m,...). Otherwise, the RDW must be specifically included in the
reformatted record (for example, 1,4,1,4,HEX)..
424

v With FIELDS, BUILD or IFTHEN BUILD, the variable part of the input record
reformatted record as the last part. In this case, a value must be specified for pn
that is less than or equal to the minimum record length (RECORD statement L4
value) plus 1 byte, and mn and an must be omitted. For example,
OUTREC FIELDS=(1,8,20C*,9)
v
v
v
If INREC with FIELDS or BUILD and OUTREC with FIELDS and BUILD are
specified, either both must specify position-only for the last part, or neither must
specify position-only for the last part. OVERLAY or IFTHEN, and FIELDS or
BUILD, can differ with respect to position-only. See INREC statement notes on
page 148 for more details.
If the reformatted record includes only the RDW and the variable part of the
input record, null records containing only an RDW may result.
The reformatted output records are in the format specified by OUTREC
regardless of whether INREC was specified.
Fields referenced in OUTREC statements can overlap each other or control fields.
v If input is variable records, the output is also variable. This means that each
record is given the correct RDW by DFSORT before output.
v When OUTREC is specified, your E35 user exit routine must refer to fields in the
v DFSORT issues a message and terminates processing if an OUTREC statement is
specified for a tape work data set sort or conventional merge application.
v When you specify OUTREC, VLSHRT is not used. If VLSHRT is specified, it is
ignored.
edited or converted input fields, decimal constants, and the results of arithmetic
expressions. If NOSZERO is in effect, -0 and +0 are treated as positive for edited
or converted input fields, decimal constants, and the results of arithmetic
expressions.
Note: OPTION SZERO or OPTION NOSZERO is ignored for OUTREC
OUTREC statement in the same source. For example, NOSZERO will be used in
Case 1:
//DFSPARM DD *
OPTION COPY,NOSZERO
/*
//SYSIN DD *
OUTREC IFTHEN=(WHEN=(1,2,FS,EQ,+0),OVERLAY=(22:CYes)),
IFTHEN=(WHEN=NONE,OVERLAY=(22:CNo ))
/*
Case 2:
425

//SYSIN DD *
OPTION COPY,NOSZERO
OUTREC IFTHEN=(WHEN=(1,2,FS,EQ,+0),OVERLAY=(28:CA)),
/*
Reformatting records after processing examples

Both INREC and OUTREC can be used to reformat records, either separately or
together. The two control statements perform similar functions, although INREC
reformats records before sort, merge, or copy processing, whereas OUTREC
reformats records after sort, merge, or copy processing. This section shows
OUTREC examples. See Reformatting records before processing examples on
page 150 for INREC examples.
Example 1
OUTREC FIELDS=(11,32)
This statement specifies that the output record is to contain 32 bytes beginning
with byte 11 of the input record. This statement can be used only with fixed-length
input records, because it does not include the first 4 bytes.
Example 2
OUTREC FIELDS=(1,4,11,32,D,101)
This statement is for variable-length records of minimum length 100 bytes, and
specifies that the output record is to contain an RDW plus 32 bytes of the input
record starting at byte 11 (aligned on a doubleword boundary, relative to the start
of the record) plus the entire variable portion of the input record.
Example 3
OUTREC FIELDS=(1,42,D,101)
This statement is for variable-length records of minimum length 100 bytes, and
specifies that the output record should contain an RDW plus the first 38 data bytes
of the input record plus the entire variable portion of the input record.
The 'D' parameter has no effect because the first field is always placed at the
beginning of the output record.
Example 4
SORT FIELDS=(20,4,CH,D,10,3,CH,D)
OUTREC FIELDS=(7:20,4,C FUTURE ,20,2,10,3,1Z,1,9,13,7,
24,57,TRAN=LTOU,6XFF)
This example illustrates how a fixed-length input data set can be sorted and
reformatted for output. The SORTIN LRECL is 80 bytes.
The reformatted output records are fixed length with a record size of 103 bytes.
of 103. The reformatted records look as follows:
Position
Contents
426
1-6
EBCDIC blanks for column alignment
7-10

11-18
Character string: C' FUTURE '
19-20
21-23
24
Binary zero
25-33
34-40
41-97
Input positions 24 through 80 with lowercase EBCDIC letters converted to

uppercase EBCDIC letters
98-103 Hexadecimal string: X'FFFFFFFFFFFF'
Example 5
SORT FIELDS=(12,4,PD,D)
RECORD TYPE=V,LENGTH=(,,,100)
OUTREC FIELDS=(1,7,5Z,5X,28,8,6X,101)
This example illustrates how a variable-length input data set can be sorted and
reformatted for output. The variable part of the input records is included in the
output records. The minimum input record size is 100 bytes and the maximum
input record size (SORTIN LRECL or maximum record size for VSAM) is 200
bytes.
The reformatted output records are variable-length with a maximum record size of
131 bytes. The reformatted records look as follows:
Position
Contents
1-4
RDW (input positions 1 through 4)
5-7
8-12
Binary zeros
13-17
EBCDIC blanks
18-25
26-31
EBCDIC blanks
32-n
Input positions 101 through n (variable part of input records)
Example 6
MERGE FIELDS=(28,4,BI,A)
OUTREC BUILD=(1,4,5Z,5X,5,3,28,8,6Z,DATE3-1)
This example illustrates how input files can be merged and reformatted for output,
with yesterday's date included. The variable part of the input records is not to be
included in the output records. The SORTINnn LRECL is 50 bytes.
The reformatted output records are variable-length with a maximum record size of
38 bytes. The reformatted records look as follows:
Position
Contents
1-4
RDW (input positions 1 through 4)
5-9
Binary zeros
427

10-14
EBCDIC blanks
15-17
18-25
26-31
Binary zeros
32-38
Yesterday's date in the form C'yyyyddd'
Example 7
OUTREC FIELDS=(SEQNUM,8,ZD,START=1000,INCR=100,
11:8,4,PD,M12,
31:15,4,Y2V(/),
51:2,1,CHANGE=(3,
X01,CL92,X02,CM72,X03,CJ42),
NOMATCH=(C???))
This example illustrates how a sequence number can be generated, how numeric
and date values can be edited, and how a lookup table can be used.
The reformatted output records look as follows:
Position
Contents
1-8
A zoned decimal sequence number that starts at 1000 and increments by

100.
1120
A CH field containing the PD field from input positions 8 through 11

edited according to the M12 edit mask.
31-40
A C'yyyy/mm/dd' date field containing the P'yymmdd' date field from

input positions 15-18 transformed according to the specified century
window of 1985-2084.
5153
A CH field containing C'L92', C'M72', C'J42' or C'???' as determined by

using a lookup table for the input field in position 2.
Example 8
(5,4,FI,ADD,3,2,FI,ADD,23,2,FI),DIV,+1000,
EDIT=(STTTTTTT),SIGNS=(,-),2X,
9,5,ZD,MIN,16,5,FS,TO=ZD,LENGTH=5,2X,
21,40)
This example illustrates how input records can be reformatted for output to
contain the results of arithmetic expressions involving input fields, decimal
constants, operators and parentheses.
The reformatted output records look as follows:
Position
Contents
428
1-20
21-28
A CH field containing the total of the FI fields from positions 5 through 8,

3 through 4 and 23 through 24, divided by 1000, and edited according to
the specified edit pattern.
29-30
EBCDIC blanks

31-35
A ZD field containing the minimum of the ZD field in positions 9 through

13 and the FS field in positions 16 through 20.
36-37
EBCDIC blanks
38-77
Example 9
OPTION COPY
OUTREC IFTHEN=(WHEN=(3,2,SS,EQ,CFR,MX,GR),
OVERLAY=(11:DATE=(DM4.),TIME1(.))),
IFTHEN=(WHEN=(3,2,SS,EQ,CCN,US,EN),
OVERLAY=(11:DATE=(4MD/),TIME1(:)))
This example illustrates how you can use IFTHEN clauses with OUTREC to
reformat different records in different ways.
For records with 'FR', 'GR' or 'MX' in positions 3-4, positions 11-20 of the
reformatted output records are overlaid with a date of the form 'dd.mm.yyyy' and
positions 21-28 are overlaid with a time of the form 'hh.mm.ss'. The data before
positions 11-28 and after positions 11-28 are not affected.
For records with 'CN', 'US' or 'EN' in positions 3-4, positions 11-20 of the
reformatted output records are overlaid with a date of the form 'yyyy/mm/dd'
and positions 21-28 are overlaid with a time of the form 'hh:mm:ss'. The data
before positions 11-28 and after positions 11-28 are not affected.
Since an IFTHEN clause with WHEN=NONE is not specified, records without 'FR',
'GR', 'MX', 'CN', 'US' or 'EN' in positions 3-4 are not changed.
Example 10
OPTION COPY
OUTREC OVERLAY=(31:11,10,ZD,DIV,+1200,TO=PD,LENGTH=6,
37:11,10,ZD,MOD,+1200,TO=PD,LENGTH=4)
This example illustrates how you can use the OVERLAY parameter with OUTREC
to change certain columns in your records without affecting other columns.
Positions 31-36 of the reformatted input records are overlaid with a 6-byte PD
value equal to the quotient of the 10-byte ZD value at positions 11-20 divided by
+1200. Positions 37-40 of the reformatted input records are overlaid with a 4-byte
PD value equal to the remainder of the 10-byte ZD value at positions 11-20 divided
by +1200. The data before positions 31-40 and after positions 31-40 are not affected.
Example 11
OPTION COPY
OVERLAY=(81:11,10,
11:81,10,ZD,DIV,+1200,TO=PD,LENGTH=6,
17:81,10,ZD,MOD,+1200,TO=PD,LENGTH=4)),
IFOUTLEN=80
In the previous example (Example 10), we used OVERLAY to overlay positions

31-40 with PD fields for the quotient and remainder of the 10-byte ZD value at
positions 11-20 divided by +1200. In this example, we want to overlay positions
11-20 with the quotient and remainder. The input records are 80 bytes and
fixed-length and we want the output records to be 80 bytes and fixed-length as
well.
429

If we just overlaid the fields directly as before, we would "ruin" the ZD value
before we could use it to get the remainder; the PD quotient would overlay
positions 11-16, so positions 11-20 would no longer contain the ZD value we need
to get the remainder.
In order to overlay the ZD value itself with the PD values, we make a copy of the
10 byte ZD value after the end of the record, and then overlay the original ZD
value with the quotient and remainder derived from the copy of the ZD value.
By making a copy of the 10 byte ZD value at the end of the record, we extend the
record length from 80 bytes to 90 bytes. Since we want the final record length to be
80 bytes, we must remove the extra 10 bytes. So instead of just using the
OVERLAY parameter, we use an IFTHEN WHEN=INIT clause with OVERLAY, and
IFOUTLEN=80. Alternatively, we could use an OVERLAY parameter on the
OUTREC statement, followed by an OUTFIL statement, to remove the extra 10
bytes like this:
OPTION COPY
OUTREC OVERLAY=(81:11,10,
11:81,10,ZD,DIV,+1200,TO=PD,LENGTH=6,
17:81,10,ZD,MOD,+1200,TO=PD,LENGTH=4)
OUTFIL OUTREC=(1,80)
Example 12
OPTION COPY
OUTREC OVERLAY=(11:11,16,JFY=(SHIFT=RIGHT,LEAD=C(,
TRAIL=C),LENGTH=18))
This example illustrates how you can right-justify fields within your records.
0001
0002
0003
0004
9-1-632-731
011-276-321-7836
753-218-307
528-314
Note that the second field has left-justified numeric values in various forms. We
want to right-justify these values and surround them with a left parenthesis and
right parenthesis.
The 50-byte FB output records look like this:
0001
0002
0003
0004
(9-1-632-731)
(011-276-321-7836)
(753-218-307)
(528-314)
We use OUTFIL OVERLAY to limit the changes to the second field. We use JFY to
right-justify the second field with surrounding parentheses. SHIFT=RIGHT shifts
the characters to the right. LEAD=C'(' adds a left parenthesis before the first
non-blank character. TRAIL=C')' adds a right parenthesis after the last non-blank
character. LENGTH=18 increases the output length by 2 bytes to allow for the
parentheses (overriding the default of 16 from the input length).
Example 13
OPTION COPY
OUTREC PARSE=(%01=(ABSPOS=2,FIXLEN=13,ENDBEFR=C","),
%=(ENDBEFR=C","),
%03=(FIXLEN=6,ENDBEFR=C","),
430

%04=(FIXLEN=6,ENDBEFR=C","),
%05=(FIXLEN=12,ENDBEFR=C")),
BUILD=(%01,20:%03,SFF,ADD,%04,SFF,EDIT=(SIIT.T),SIGNS=(,-),
31:%05)
This example illustrates how you can reformat records containing variable
position/length fields, such as comma separated values.
"Buffy
"Bruce
"Clark
"Diana
Summers","F","+725.8","-27.3","Sunnydale"
Wayne","M","-5.3","-173.2","Gotham City"
Kent","M","+21.3","-15.8","Metropolis"
Prince","F","-16.4","+128.9","Gateway City"
Note that each record has five variable fields, each enclosed in quotes and
separated by a comma. The fields do not start and end in the same position in
every record and they have different lengths in different records.
The 42-byte output records should look like this:
Buffy
Bruce
Clark
Diana
Summers
Wayne
Kent
Prince
698.5
-178.5
5.5
112.5
Sunnydale
Gotham City
Metropolis
Gateway City
The first fixed-length output field corresponds to the first variable input field. The
second fixed-length output field corresponds to the total of the third and fourth
variable input fields. The third fixed-length output field corresponds to the fifth
variable input field.
In order to reformat the input records for output, we use PARSE and BUILD to
create the fixed parsed fields we need from the variable position/length fields; the
quotes around each value are removed. %00 is used for the first input field. % is
used to ignore the second input field. %03 and %04 are used for the third and
fourth input fields which are added together using SFF format. %05 is used for the
fifth input field.
Example 14
OPTION COPY
PARSE=(%100=(FIXLEN=3,ENDBEFR=C.,REPEAT=3),
%103=(FIXLEN=3)),
BUILD=(%100,4:C.,5:%101,8:C.,9:%102,12:C.,13:%103)),
IFTHEN=(WHEN=(1,3,CH,EQ,C167,AND,13,3,CH,EQ,C99),
OVERLAY=(5:C113,9:C75 ,1:1,15,SQZ=(SHIFT=LEFT))),
IFTHEN=(WHEN=NONE,
BUILD=(1:1,15,SQZ=(SHIFT=LEFT)))
This example illustrates how you can modify variable position/length fields, such
as ftp addresses.
167.113.117.99
167.90.18.99
167.80.118.98
165.250.89.562
167.125.890.95
168.250.89.99
167.125.890.99
0.0.0.0
167.90.580.99
431

We want to convert each FTP address of the form 167.x.y.99 to 167.113.75.99. x and
y can be any 1-3 digit value. Thus, the 15-byte output records should look like this:
167.113.75.99
167.113.75.99
167.80.118.98
165.250.89.562
167.125.890.95
168.250.89.99
167.113.75.99
0.0.0.0
167.113.75.99
In order to reformat the input records for output, we use IFTHEN clauses as
follows:
v IFTHEN WHEN=INIT clause: We PARSE the four variable numeric values into
four 3-byte fixed parsed fields using %100, %101, %102 and %103 respectively.
Note that we use REPEAT=3 with the %100 parsed field to repeat it for the %101
and %102 parsed fields. REPEAT=n is useful when you have several contiguous
parsed fields with the same operands).
We reformat the record with %100, a period, %101, a period, %102, a period and
%103. At this point, the reformatted records look like this:
167.113.117.99
167.90 .18 .99
167.80 .118.98
165.250.89 .562
167.125.890.95
168.250.89 .99
167.125.890.99
0 .0 .0 .0
167.90 .580.99
v IFTHEN WHEN=(logexp) clause: If the first fixed-length field is '167' and the
fourth fixed-length field is '99 ', we overlay the second fixed-length field with
'113' and the third fixed-length field with '75'. Then we squeeze the fields to the
left to remove the blanks.
v IFTHEN WHEN=NONE clause: If the first fixed-length field is not '167' or the
fourth fixed-length field is not '99 ', we squeeze the fields to the left to remove
the blanks.
Example 15
OUTREC FINDREP=(IN=C*,OUT=C )
This example illustrates how you can replace a character with another character
anywhere in FB or VB records.
The FB input records might look like this:
05 ***** JUNE ***************
02
* APRIL *
01* * * * DAISY * * * *
03BETTY ***********
We want to replace every asterisk with a blank. We use IN=C'*' to indicate we

want to find each asterisk, and OUT=C' ' to indicate we want to replace it with a
blank.
The sorted output records look like this:
432

01
DAISY
02
APRIL
03BETTY
05
JUNE
Example 16
OPTION COPY
OUTREC IFTHEN=(WHEN=GROUP,BEGIN=(1,3,CH,EQ,CHDR),
END=(1,3,CH,EQ,CTRL),PUSH=(31:ID=1))
OUTFIL INCLUDE=(31,1,CH,NE,C ),BUILD=(1,30)
This example illustrates how you can INCLUDE groups of FB records between a
header and a trailer. We add an ID after the end of each record to indicate whether
it's part of a group or not, INCLUDE on the ID, and then remove it.
C33
HDR
A01
B02
C03
TRL
R24
T02
HDR
D04
E05
TRL
F97
Not in a group
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Not in a group
Not in a group
Start Group 2
Group 2 record
Group 2 record
End Group 2
Not in a group
In the output data set we only want to include groups of records that start with
'HDR' and end with 'TRL'.
We use an IFTHEN WHEN=GROUP clause to put a non-blank character in each
record that is part of a group. BEGIN indicates a group starts with a record that
has 'HDR' in positions 1-3. END indicates a group ends with a record that has
'TRL' in positions 1-3. PUSH overlays a 1-byte ID character at position 31 in each
record of a group (after the end of the record). After the IFTHEN GROUP clause is
C33
HDR
A01
B02
C03
TRL
R24
T02
HDR
D04
E05
TRL
F97
Not in a group
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Not in a group
Not in a group
Start Group 2
Group 2 record
Group 2 record
End Group 2
Not in a group
1
1
1
1
1
2
2
2
2
Note that the records within a group have a non-blank character in position 31
whereas the records outside groups have a blank character in position 31. The ID
starts at 1 for the first group and is incremented by 1 for each subsequent group.
Since we are only allowing one character for the ID, when the ID counter gets to
10, a '0' will appear in position 31. That's fine since we are just looking for a
non-blank to indicate a record within a group, or a blank to indicate a record
outside of a group.
433

We use an OUTFIL statement to only INCLUDE records with a non-blank in
position 31, and to remove the ID character so the included output records will be
identical to the input records. After the OUTFIL statement is executed, the final
output records look like this:
HDR
A01
B02
C03
TRL
HDR
D04
E05
TRL
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Start Group 2
Group 2 record
Group 2 record
End Group 2
Example 17
OPTION COPY
OUTREC IFTHEN=(WHEN=INIT,BUILD=(1,4,6:5)),
IFTHEN=(WHEN=GROUP,BEGIN=(6,3,CH,EQ,CHDR),
END=(6,3,CH,EQ,CTRL),PUSH=(5:ID=1))
OUTFIL INCLUDE=(5,1,CH,NE,C ),BUILD=(1,4,5:6)
This example illustrates how you can INCLUDE groups of VB records between a
header and a trailer. It's similar to Example 16, but here the records are
variable-length. For the FB records, we could add the ID after the end of each
record and then remove it without changing the records. But we can't add the ID
at the end of each VB record because that would pad all of the records to a fixed
length. So, instead we insert the ID between the RDW and the first data byte of
each record, and later remove it.
Len|Data
23|C33
23|HDR
25|A01
25|B02
25|C03
21|TRL
23|R24
23|T02
23|HDR
25|D04
25|E05
21|TRL
25|F97
Not in a group
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Not in a group
Not in a group
Start Group 2
Group 2 record
Group 2 record
End Group 2
Not in a group
In the output data set we only want to include groups of records that start with
'HDR' and end with 'TRL'.
We use an IFTHEN WHEN=INIT clause to reformat each record so it has room for
the ID byte between the RDW and the first data byte. After the WHEN=INIT
clause is executed, the intermediate records look like this:
Len|Data
24| C33
24| HDR
26| A01
26| B02
26| C03
22| TRL
24| R24
24| T02
24| HDR
434
Not in a group
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Not in a group
Not in a group
Start Group 2

26|
26|
22|
26|
D04
E05
TRL
F97
Group 2 record
Group 2 record
End Group 2
Not in a group
Note that position 5 is blank and the 'HDR' and 'TRL' characters have been shifted
over to positions 6-8.
We use an IFTHEN WHEN=GROUP clause to put a non-blank character in each
record that is part of a group. BEGIN indicates a group starts with a record that
has 'HDR' in positions 6-8. END indicates a group ends with a record that has
'TRL' in positions 6-8. PUSH overlays a 1-byte ID character at position 5 in each
record of a group. After the IFTHEN GROUP clause is executed, the intermediate
records look like this:
Len|Data
24| C33
24|1HDR
26|1A01
26|1B02
26|1C03
22|1TRL
24| R24
24| T02
24|2HDR
26|2D04
26|2E05
22|2TRL
26| F97
Not in a group
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Not in a group
Not in a group
Start Group 2
Group 2 record
Group 2 record
End Group 2
Not in a group
Note that the records within a group have a non-blank character in position 5
whereas the records outside groups have a blank character in position 5. The ID
starts at 1 for the first group and is incremented by 1 for each subsequent group.
Since we are only allowing one character for the ID, when the ID counter gets to
10, a '0' will appear in position 5. That's fine since we are just looking for a
non-blank to indicate a record within a group, or a blank to indicate a record
outside of a group.
We use an OUTFIL statement to only INCLUDE records with a non-blank in
position 5, and to remove the ID character so the included output records will be
identical to the input records. After the OUTFIL statement is executed, the final
output records look like this:
Len|Data
23|HDR
25|A01
25|B02
25|C03
21|TRL
23|HDR
25|D04
25|E05
21|TRL
Start Group 1
Group 1 record
Group 1 record
Group 1 record
End Group 1
Start Group 2
Group 2 record
Group 2 record
End Group 2
Example 18
OUTREC OVERLAY=(30:16,8,Y4T,TOGREG=Y4T)
OUTFIL INCLUDE=(30,1,CH,EQ,C*)
This example illustrates how to list records with dates outside of the valid range
(for example, a month not between 01-12).
435

Note: Dates with an invalid digit (A-F) can result in a data exception (0C7
ABEND).
Betten
Vezinaw
Casad
Boenig
Kolusu
Yaeger
20091021
20091101
00000000
20091325
20090931
20090731
The SORT statement sorts the records by the name in positions 1-12. After the
SORT statement is processed, the sorted records will look like this:
Betten
Boenig
Casad
Kolusu
Vezinaw
Yaeger
20091021
20091325
00000000
20090931
20091101
20090731
The OUTREC statement uses TOGREG to convert each ccyymmdd value in

positions 16-23 to a ccyymmdd value in positions 30-37; the third column will be
identical to the second column for valid dates and special indicators, but will
contain asterisks for invalid dates. After the OUTREC statement is processed, the
reformatted records will look like this:
Betten
Boenig
Casad
Kolusu
Vezinaw
Yaeger
20091021
20091325
00000000
20090931
20091101
20090731
20091021
********
00000000
********
20091101
20090731
The OUTFIL statement selects the records that have an asterisk in position 30, that
is, the records with an invalid date. The Boenig record is invalid because mm is 13,
and the Kolusu record is invalid because mm is 09 but dd is 31 (September only
has 30 days). Note that the special indicator of all 0s for the Casad record is valid.
Boenig
Kolusu
20091325
20090931
********
********
Example 19
BUILD=(1:1,8,UFF,TO=ZD,LENGTH=6,8:11,6,UFF,TO=ZD,LENGTH=5)),
IFTHEN=(WHEN=INIT,
BUILD=(1,6,Y2W,DATEDIFF,8,5,Y2T))
This example illustrates how you can calculate the difference in days between two
different types of date fields, each of which has separators.
The SORTIN data set has these input records with a C'mm/dd/yy' date field in
positions 1-8 and a C'yy/ddd' date field in positions 11-16:
03/05/07
12/13/07
02/19/08
09/01/08
11/22/09
436
07/052
08/193
08/365
09/001
09/015

07/22/98
01/15/09
09/15/10
06/30/10
01/121
09/015
09/322
08/050
The OUTREC statement calculates the number of days for date1-date2 and puts the
result in the output record in positions 1-8.
The first IFTHEN clause removes the / separators from date1 and date2 so we can
use them in DATEDIFF. After the first IFTHEN clause, the records look like this,
with the C'mmddyy' date in positions 1-6 and the C'yyddd' date in positions 8-12:
030507
121307
021908
090108
112209
072298
011509
091510
063010
07052
08193
08365
09001
09015
01121
09015
09322
08050
The second IFTHEN clause uses DATEDIFF to get the number of days between
date1 and date2. We use 1,6,Y2W to match the C'mmddyy' date and 8,5,Y2T to
match the C'yyddd' date.
+0000012
-0000211
-0000315
-0000122
+0000311
-0001014
+0000000
+0000301
+0000862
Note that when date1>=date2, the result is a positive value, and when
date1<date2, the result is a negative value.
437
RECORD Control Statement
RECORD control statement

RECORD TYPE=
F
V
D
LENGTH=( L1
TYPE =F,
, L2
L1
, L3
,
L2
L1
LENGTH=( L1
TYPE= V ,
D
, L2
L1
, L3
L2
L1
,
L2
L1
,
,
L2
L1
,
,
L1
,
L3
L2
L1
, L4
L3
,
L3
,
L2
, L5
L4
,
L3
, L6
L5
L4
,
L4
,
L5
, L7
L6
The RECORD control statement can be used to specify the type and lengths of the
records being processed, and the minimum and average record lengths for a
variable-length sort.
The RECORD control statement is required when:
v A user exit changes record lengths.
v A user exit supplies all of the input records.
v A Conventional merge or tape work data set sort uses VSAM input.
TYPE
TYPE=x
Can be used to specify the record type when input is VSAM, or an E15 or E32
exit supplies all of the input records. The record type can be:
v Fixed-length (F). The records are processed without an RDW, so the data
starts in position 1. Control statement positions should be specified
accordingly.
An RRDS can always be processed as fixed-length. A KSDS, ESDS or VRRDS
used for input should only be processed as fixed-length if all of its records
have a length equal to the maximum record size defined for the cluster.
Otherwise, input records which are shorter than the maximum record size
are padded with bytes that may or may not be zeros (that is, "garbage"
bytes).
438

v Variable-length (V). The records are processed with an RDW in positions
1-4, so the data starts in position 5. Control statement positions should be
specified accordingly.
An RRDS, KSDS, ESDS or VRRDS can always be processed as
variable-length. For VSAM input, DFSORT reads each record and prepends
an RDW to it. For VSAM output, DFSORT removes the RDW before writing
each record.
TYPE is only required for a Conventional merge or tape work data set sort that
uses VSAM input or an E15 or E32 exit that supplies all of the input records.
If input is non-VSAM, DFSORT determines the record type from the RECFM of
the input data set and ignores TYPE.
If input is VSAM, or an E15 or E32 exit supplies all of the input records,
DFSORT determines or assigns the record type as follows, using the
information in the order listed:
1. F or V from RECORD TYPE if specified.
2. F or V from SORTOUT RECFM if available.
3. V if OUTFIL VTOF, CONVERT or VLFILL is specified, or F if OUTFIL
FTOV is specified.
4. F or V from OUTFIL RECFM if available.
5. V if SORTIN is VSAM and SORTOUT is VSAM; otherwise F.
Note:
a. If the selected record type is not what you want DFSORT to use, specify
RECORD TYPE=F or RECORD TYPE=V as appropriate.
b. For a Conventional merge or tape work data set sort, you must specify
RECORD TYPE=F or RECORD TYPE=V as appropriate.
x can be one of the following:
F
fixed-length record processing.

Note: FB can be used instead of F.
variable-length record processing.

Note: VB can be used instead of V.
ASCII variable-length record processing.

Note: DB can be used instead of D.
Default: F or V as described previously in this section. See Appendix B,

details.
LENGTH
LENGTH= (
439

L1
)
,
L2
L1
,
L1
,
L1
,
L2
,
L1
,
L3
,
L2
L4
L3
L2
L1
L1
L3
L2
,
L3
,
L2
,
,
L4
,
L3
L5
L4
,
,
L4
L6
L5
,
L5
L7
L6
Can be used to specify various record lengths. L1 through L3 apply to

fixed-length and variable-length record processing. L4 and L5 apply to
variable-length record processing. L6 and L7 are accepted, but not used.
LENGTH is required only if:
v A user exit changes record lengths.
v A user exit supplies all of the input records.
L1 Input record length. For variable-length records, maximum input record
length.
Note:
1. L1 is ignored if the input record length is available from SORTIN.
2. L1 is required if there is no SORTIN or SORTINnn data set, unless L2 is
specified.
Default: The SORTIN or SORTINnn record length. For VSAM data sets, the
maximum record size (RECSZ value).
L2 Record length after E15. For variable-length records, maximum record
length after E15.
Note:
1. L2 is ignored if E15 is not used.
2. An accurate value for L2 must be specified if E15 changes the record
length.
3. L2 must be at least 18 bytes if tape work data sets are used.
4. L2 is ignored if there is no SORTIN or SORTINnn data set, unless L1 is
not specified.
Default: L1.
L3 Output record length. For variablelength records, maximum output
record length.
Note: L3 is ignored if the record length (LRECL or VSAM RECSZ) is
available from SORTOUT, or if NOSOLRF is in effect and E35, INREC,
OUTREC, and OUTFIL are not used.
Default: One of the following, in the order listed:
1. SORTOUT record length if available
2. OUTREC record length if SOLRF is in effect
3. INREC record length if SOLRF is in effect
4. L2 if specified providing an E15 is used
440

5. SORTIN or SORTINnn record length if available
6. L1
L4 Minimum record length.
Note:
1. L4 is not used if the Blockset technique is selected
2. L4 is only used for variable-length record sort applications.
3. Specifying L4 may improve performance, but if L4 is too large,
DFSORT could fail with message ICE015A.
Default: The minimum length needed to contain all control fields. This
number must be at least 18 bytes if the maximum input record length is
greater than 18 bytes; otherwise, DFSORT sets L4 to 18 bytes.
L5 Average record length.
Note:
1. L5 is not used if the Blockset technique is selected
2. L5 is overridden by the AVGRLEN parameter if both are specified
3. L5 is only used for variable-length sorts.
Default: None; optional.
L6, L7
Record lengths that are accepted but are reserved for future use.
Note:
1.
2. You can drop values from the right. For example, LENGTH=(80,70,70,70).
3. You can omit values from the middle or left, provided you indicate their
omission by a comma or semicolon. For example, LENGTH=(,,,30,80).
4. Parentheses are optional when L1 alone is specified. If any of L2 through L7 is
specified, with or without L1, parentheses are required.
Applicable Functions: See Appendix B, Specification/override of DFSORT options,
on page 863.
Describing the record format and lengthexamples

Example 1
MODS E15=(INEX,1000,EXIT),E35=(OUTEX,2000,EXIT)
RECORD LENGTH=(,175,180)
This example illustrates how the RECORD statement can be used to indicate that
E15 and E35 exits change the record length. The record type (F) and input record
length (200) are obtained automatically from the RECFM and LRECL of the input
data set, respectively.
LENGTH
L2 specifies that the E15 exit passes back 175 byte records. L3 specifies that
the E35 exit passes back 180 byte records.
Example 2
MODS E15=(E15ONLY,1000,EXIT)
RECORD TYPE=V,LENGTH=60
441

This example illustrates how the RECORD statement can be used to set the record
type and maximum input record length when an E15 exit supplies all of the input
as variable-length records.
TYPE
V specifies that the E15 exit inserts variable-length records, that is, the inserted
records contain an RDW in positions 1-4 and the data starts in position 5.
LENGTH
L1 specifies that the E15 exit inserts records with a maximum length of 60
bytes.
REFORMAT control statement

application for joining two files, on page 457 for details.
SORT control statement

,
SORT FIELDS=
( p,m,f,s
,
( p,m,
FORMAT=f
f,
COPY

,
,
CKPT
DYNALLOC
=
d
(,n)
(d,n)
OFF
EQUALS
NOEQUALS
FILSZ=
x
Ex
Ux
SIZE=
y
Ey
Uy
SKIPREC=z
STOPAFT=n
Y2PAST=
s
f
The SORT control statement must be used when a sorting application is performed;
this statement describes the control fields in the input records on which the
program sorts. A SORT statement can also be used to specify a copy application.
User labels will not be copied to the output data sets.
442
SORT Control Statement

The way in which DFSORT processes short SORT control fields depends on the
setting for VLSHRT/NOVLSHRT. A short field is one where the variable-length
record is too short to contain the entire field, that is, the field extends beyond the
record. For details about sorting short records, see the discussion of the VLSHRT
and NOVLSHRT options in OPTION control statement on page 173.
The options available on the SORT statement can be specified in other sources as
well. A table showing all possible sources for these options and the order of
override is given in Appendix B, Specification/override of DFSORT options, on
page 863.
When an option can be specified on either the SORT or OPTION statement, it is
preferable to specify it on the OPTION statement.
DFSORT accepts but does not process the following SORT operands: WORK=value
and ORDER=value.
The active locale's collating rules affect SORT processing as follows:
v DFSORT produces sorted records for output according to the collating rules
defined in the active locale. This provides sorting for single- or multi-byte
character data, based on defined collating rules that retain the cultural and local
characteristics of a language.
character (CH) control fields.
173.
FIELDS
,
FIELDS=( p,m,f,s
Requires four facts about each control field in the input records: the position of
the field within the record, the length of the field, the format of the data in the
field, and the sequence into which the field is to be sorted. These facts are
communicated to DFSORT by the values of the FIELDS operand, represented
by p, m, f, and s.
The value for f can optionally be specified by the FORMAT=f parameter as
explained later in this section.
All control fields must be located within the first 32752 bytes of a record.
The maximum length of the collected control fields for which Blockset can be
used is 4088 bytes. However, the maximum length can be less than 4088 for
various situations, such as the use of certain formats (for example, PD), the
EQUALS option, and so on. If this maximum is exceeded, Blockset cannot be
used.
The FIELDS operand can be written in two ways.
The program examines the major control field first, and it must be specified
first. The minor control fields are specified following the major control field. p,
m, f, and s describe the control fields. The text that follows gives specifications
in detail.
443

p
specifies the first byte of a control field relative to the beginning of the
input record. 14
The first data byte of a fixed-length record has relative position 1. The first
data byte of a variable-length record has relative position 5. The first 4
bytes contain the record descriptor word. All control fields, except binary,
must begin on a byte boundary. The first byte of a floating-point field is
interpreted as a signed exponent; the rest of the field is interpreted as the
fraction.
Fields containing binary values are described in a bytes.bits notation as
follows:
1. First, specify the byte location relative to the beginning of the record
and follow it with a period.
2. Then, specify the bit location relative to the beginning of that byte.
Remember that the first (high-order) bit of a byte is bit 0 (not bit 1); the
remaining bits are numbered 1 through 7.
Thus, 1.0 represents the beginning of a record. A binary field beginning on
the third bit of the third byte of a record is represented as 3.2. When the
beginning of a binary field falls on a byte boundary (say, for example, on
the fourth byte), you can write it in one of three ways:
4.0
4.
4
Other examples of this notation are shown in Figure 8:

byte 1
bits 0 - 7
1.0
1.
1
byte 3
bits 0 - 7
byte 2
bits 0 - 7
1.6
2.2
3.0 3.1
3.
3
Figure 8. Examples of Notation for Binary Fields
specifies the length of the control field. Values for all control fields except
binary fields must be expressed in integer numbers of bytes. Binary fields
can be expressed in the notation bytes.bits. The length of a binary control
field that is an integer value (d) can be expressed in one of three ways:
444

d.0
d.
d
The number of bits specified must not exceed 7. A control field 2 bits long
would be represented as 0.2.
The total number of bytes occupied by all control fields must not exceed
4092 (or, when the EQUALS option is in operation, 4088 bytes). When you
determine the total, count a binary field as occupying an entire byte if it
occupies any part of it. For example, a binary field that begins on byte 2.6
and is 3 bits long occupies two bytes. All fields must be completely
contained within the first 32752 bytes of the record.
f
specifies the format of the data in the control field. Acceptable control field
lengths (in bytes) and available formats are shown in Table 60.
Table 60. Control Field Formats and Lengths

Length
Description
CH
1 to 4092 bytes
Character
AQ
1 to 4092 bytes
Character with alternate

collating sequence
ZD
1 to 256 bytes
PD
1 to 256 bytes
PD0
2 to 8 bytes
Packed decimal with sign

and first digit ignored
FI
1 to 256 bytes
Signed fixed-point
BI
1 bit to 4092 bytes
Unsigned binary
FL
1 to 256 bytes
Signed hexadecimal
floating-point
15
AC
1 to 4092 bytes
ASCII character
CSF or FS
1 to 32 bytes

UFF
1 to 44 bytes
SFF
1 to 44 bytes
CSL or LS
2 to 256 bytes

separate sign
CST or TS
2 to 256 bytes

separate sign
CLO or OL
1 to 256 bytes

overpunch sign
CTO or OT
1 to 256 bytes

overpunch sign
ASL
2 to 256 bytes

leading separate sign
AST
2 to 256 bytes

trailing separate sign
D1
1 to 4092 bytes
User-defined data type

(requires an EFS program)
15. If CHALT is in effect, CH is treated as AQ.

445

Table 60. Control Field Formats and Lengths (continued)
Length
Description
Y2T
3 to 6 bytes
Character or zoned yyx...x

full date format with special
indicators
Y2U
2 or 3 bytes
Packed decimal yyx and

yyxxx full date format with
special indicators
Y2V
3 or 4 bytes
Packed decimal yyxx and

yyxxxx full date format with
special indicators
Y2W
3 to 6 bytes
Character or zoned x...xyy

full date format with special
indicators
Y2X
2 or 3 bytes
Packed decimal xyy and

xxxyy full date format with
special indicators
Y2Y
3 or 4 bytes
Packed decimal xxyy and

xxxxyy full date format with
special indicators
Y2C or Y2Z
2 bytes
Two-digit character or
zoned-decimal year data
Y2P
2 bytes
Two-digit packed-decimal
year data
Y2D
1 byte
Two-digit decimal year data
Y2S
2 bytes
Two-digit character or
zoned-decimal year data with
special indicators
Y2B
1 byte
Two-digit binary year data
descriptions.
CSF, FS, UFF, SFF, Y2 and PD0 format fields can only be used if Blockset is
selected.
For Y2 format fields, real dates are collated using the century window
established by the Y2PAST option in effect, but the century window is not used
for special indicators. Thus the Y2 formats will collate real dates and special
indicators as follows:
v Y2T and Y2W:
Ascending:
BI zeros, blanks, CH/ZD zeros, lower century dates (for example,
19yy), upper century dates (for example, 20yy), CH/ZD nines, BI
ones.
Descending:
BI ones, CH/ZD nines, upper century dates (for example, 20yy),
lower century dates (for example, 19yy), CH/ZD zeros, blanks, BI
zeros.
v Y2U, Y2V, Y2X and Y2Y:
Ascending:
PD zeros, lower century dates (for example, 19yy), upper century
dates (for example, 20yy), PD nines.
446

Descending:
PD nines, upper century dates (for example, 20yy), lower century
dates (for example, 19yy), PD zeros.
v Y2C, Y2Z, Y2P, Y2D and Y2B:
Ascending:
Lower century years (for example, 19yy), upper century years (for
example, 20yy).
Descending:
Upper century years (for example, 20yy), lower century years (for
example, 19yy).
v Y2S:
Ascending:
BI zeros, blanks, lower century years (for example, 19yy), upper
century years (for example, 20yy), BI ones.
Descending:
BI ones, upper century years (for example, 20yy), lower century
years (for example, 19yy), blanks, BI zeros.
The AC format sequences EBCDIC data using the ASCII collating sequence
shown in Appendix D. For example, if AC is used with ascending sequence,
the EBCDIC numbers (0-9) will collate before the EBCDIC uppercase letters
(A-Z) which in turn will collate before the EBCDIC lowercase letters (a-z).
You can use p,m,s rather than p,m,f,s if you use FORMAT=f to supply the
format for the field, as described later in this section.
All floating-point data must be normalized before the program can collate it
properly. You can use an E15 or E61 user exit to do this during processing. If
you use E61, specify the E option for the value of s in the FIELDS operand for
each control field you are going to modify with this user exit.
s
specifies how the control field is to be ordered. The valid codes are:
A
ascending order
descending order
control fields to be modified
Specify E if you include an E61 user exit to modify control fields before the
program sorts them. After an E61 user exit modifies the control fields,
DFSORT collates the records in ascending order using the formats
specified. 16
For information on how to add a user exit, see Chapter 5, Using your own
user exit routines, on page 487.
Default: None; must be specified. See Appendix B, Specification/override of
DFSORT options, on page 863 for full override details.
FORMAT
16. With a conventional merge or a tape work data set sort, control fields for which E is specified are treated as binary byte format
regardless of the actual formats specified.
447

FORMAT=f
FORMAT=f can be used to specify a particular format for one or more control
fields. f from FORMAT=f is used for p,m,s fields. f from FORMAT=f is ignored
for p,m,f,s fields. For example, the following are all equivalent:
SORT FIELDS=(5,5,ZD,A,12,6,PD,D,21,3,PD,A,35,7,ZD,A)
SORT FORMAT=ZD,FIELDS=(5,5,A,12,6,PD,D,21,3,PD,A,35,7,A)
SORT FIELDS=(5,5,ZD,A,12,6,D,21,3,A,35,7,ZD,A),FORMAT=PD
The permissible field formats are shown under the description of 'f' for fields.
If you have specified the COPY operand, FORMAT=f cannot be specified.
Default: None; FORMAT=f must be specified if any field is specified as p,m,s
rather than p,m,f,s. See Appendix B, Specification/override of DFSORT
Note: DFSORT issues an informational message and ignores FORMAT=f if all
of the fields are specified as p,m,f,s.
FIELDS=COPY
FIELDS=COPY
See the discussion of the COPY option discussed in OPTION control

CKPT
CKPT
See the discussion of this option discussed in OPTION control statement on

page 173.
DYNALLOC
DYNALLOC

=
d
(,n)
(d,n)
OFF
See the discussion of this option in OPTION control statement on page 173.
EQUALS or NOEQUALS
EQUALS
NOEQUALS
See the discussion of these options in OPTION control statement on page

173.
FILSZ or SIZE
448

FILSZ=
SIZE=
x
Ex
Ux
y
Ey
Uy
See the discussion of these options in OPTION control statement on page

173.
SKIPREC
SKIPREC=z
STOPAFT
STOPAFT=n
Y2PAST
Y2PAST=
s
f
Y2PAST=value.
SORT/MERGE statement notes

v If records are reformatted by INREC (SORT and MERGE) or E15 (SORT),
FIELDS must refer to fields in the appropriate reformatted records.
v If SZERO is in effect, -0 collates before +0 in ascending order and after +0 in
descending order when numeric fields are sorted or merged. If NOSZERO is in
effect, -0 collates equally with +0 when numeric fields are sorted or merged.
However, SZERO is always used for a conventional merge or tape work data set
sort application.
Specifying a SORT or COPYexamples

Example 1
SORT
FIELDS=(2,5,FS,A),FILSZ=29483
FIELDS
The control field begins on the second byte of each record in the input data set,
is five bytes long, and contains floating sign data. It is to be sorted in
ascending order.
FILSZ
The data set to be sorted contains exactly 29483 records.
Example 2
SORT
FIELDS=(7,3,CH,D,1,5,FI,A,398.4,7.6,BI,D,99.0,230.2,
BI,A,452,8,FL,A),DYNALLOC=(3390,4)
449

FIELDS
The first four values describe the major control field. It begins on byte 7 of
each record, is 3 bytes long, and contains character (EBCDIC) data. It is to be
sorted in descending order.
The next four values describe the second control field. It begins on byte 1, is 5
bytes long, contains fixed-point data, and is to be sorted in ascending order.
The third control field begins on the fifth bit (bits are numbered 0 through 7)
of byte 398. The field is 7 bytes and 6 bits long (occupies 9 bytes), and contains
binary data to be placed in descending order.
The fourth control field begins on byte 99, is 230 bytes and 2 bits long, and
contains binary data. It is to be sorted in ascending order.
The fifth control field begins on byte 452, is 8 bytes long and contains
normalized hexadecimal floating-point data, which is to be sorted in ascending
order. If the data in this field were not normalized, you could specify E instead
of A and include your own E61 user exit routine to normalize the field before
the program examined it.
DYNALLOC
Four work data sets are allocated on 3390.
Example 3
SORT
FIELDS=(3,8,ZD,E,40,6,CH,D)
FIELDS
The first four values describe the major control field. It begins on byte 3 of
each record, is 8 bytes long, and contains zoned decimal data that is modified
by your routine before sort examines the field.
The second field begins on byte 40, is 6 bytes long, contains character
(EBCDIC) data, and is sorted in descending sequence.
Example 4
SORT
FIELDS=(7025,4,A,5048,8,A),FORMAT=ZD,EQUALS
FIELDS
The major control field begins on byte 7025 of each record, is 4 bytes long,
contains zoned decimal data (FORMAT=ZD), and is to be sorted in ascending
sequence.
The second control field begins on byte 5048, is 8 bytes long, has the same data
format as the first field, and is also to be sorted in ascending order.
FORMAT
FORMAT=ZD is used to supply ZD format for the p,m,s fields and is
equivalent to specifying p,m,ZD,s for these fields.
With FORMAT=f, you can mix p,m,s and p,m,f,s fields when that's convenient
such as when all or most of the fields have the same format (although you can
always code p,m,f,s for all fields and not use FORMAT=f, if you prefer). For
example, the following are also valid uses of the FORMAT=f parameter:
SORT FORMAT=BI,FIELDS=(21,4,A,5,4,PD,A,31.3,1.4,A,52,20,A)
SORT FIELDS=(16,4,A,22,8,BI,D,3,2,A),FORMAT=FI
EQUALS
specifies that the sequence of equal collating records is to be preserved from
input to output.
450
Example 5
SORT
FIELDS=COPY
FIELDS
The input data set is copied to the output data set without sorting or merging.
Example 6
OPTION Y2PAST=1950
SORT FIELDS=(21,6,Y2T,A,13,3,Y2X,D)
Y2PAST
Sets a century window of 19502049.
FIELDS
Sorts on a C'yymmdd' (or Z'yymmdd') date in positions 21-26 in ascending
order, and on a P'dddyy' date in positions 13-15 in descending order. "Real"
dates are sorted using the century window of 1950-2049. Special indicators are
sorted correctly relative to the "real" dates.
SUM control statement

,
SUM FIELDS=
p,m,f
,
p,m
FORMAT=f
,f
NONE
The SUM control statement specifies that, whenever two records are found with
equal sort or merge control fields, the contents of their summary fields are to be
added, the sum is to be placed in one of the records, and the other record is to be
deleted. If the EQUALS option is in effect the first record of summed records is
kept. If the NOEQUALS option is in effect, the record to be kept is unpredictable.
For further details, see SUM statement notes on page 453.
If the ZDPRINT option is in effect, positive summed ZD values are printable. If the
NZDPRINT option is in effect, positive summed ZD values are not printable. For
further details, see SUM statement notes on page 453.
The way in which DFSORT processes short SUM summary fields depends on
whether the VLSHRT or NOVLSHRT option is in effect. A short field is one where
the variable-length record is too short to contain the entire field; that is, the field
extends beyond the record. For details about sorting, merging and summing short
records, see the discussion of the VLSHRT and NOVLSHRT options in OPTION
FIELDS
,
FIELDS= (
p,m,f
Designates numeric fields in the input record as summary fields.

p
specifies the first byte of the field relative to the beginning of the input
record. 17 The first data byte of a fixed-length record has relative position 1.
451
SUM Control Statement

The first data byte of a variable-length record has relative position 5, as the
first four bytes are occupied by the RDW. All fields must start on a byte
boundary and no field can extend beyond byte 32752
m
specifies the length in bytes of the summary fields to be added. See

Table 61for permissible length values.
specifies the format of the data in the summary field:
Table 61. Summary Field Formats and Lengths

Length
Description
BI
2, 4, or 8 bytes
Unsigned binary
FI
2, 4, or 8 bytes
Signed fixed-point
FL
4, 8, or 16 bytes
Signed hexadecimal
floating-point
PD
1 to 16 bytes
ZD
1 to 31 bytes
The value for f can optionally be specified by the FORMAT=f parameter as

explained later in this section.
Note: See Appendix C, Data format descriptions, on page 891 for
detailed format descriptions.
NONE
eliminates records with duplicate keys. Only one record with each key is
kept and no summing is performed.
Note: The FIRST operand of ICETOOL's SELECT operator can be used to
perform the same function as SUM FIELDS=NONE with OPTION EQUALS.
Additionally, SELECT's FIRSTDUP, ALLDUPS, NODUPS, HIGHER(x),
LOWER(y), EQUAL(v), LASTDUP, and LAST operands can be used to select
records based on other criteria related to duplicate and non-duplicate keys.
SELECT's DISCARD(savedd) operand can be used to save the records
discarded by FIRST, FIRSTDUP, ALLDUPS, NODUPS, HIGHER(x), LOWER(y),
EQUAL(v), LASTDUP, or LAST. See SELECT operator on page 668 for
complete details on the SELECT operator.
Default: None; must be specified.
FORMAT
FORMAT=f
FORMAT=f can be used to specify a particular format for one or more

summary fields. f from FORMAT=f is used for p,m fields. f from FORMAT=f is
ignored for p,m,f fields. For example, the following are all equivalent:
452

SUM FIELDS=(5,5,ZD,12,6,PD,21,3,PD,35,7,ZD)
SUM FORMAT=ZD,FIELDS=(5,5,12,6,PD,21,3,PD,35,7)
SUM FIELDS=(5,5,ZD,12,6,21,3,35,7,ZD),FORMAT=PD
The permissible field formats are shown under the description of 'f' for fields.
Default: None. FORMAT=f must be specified if any field is specified as p,m
rather than p,m,f. See Appendix B, Specification/override of DFSORT
Note: DFSORT issues an informational message and ignores FORMAT=f if all of
the fields are specified as p,m,f.
SUM statement notes

v If overflow might occur during summation, INREC can be used to create a
larger SUM field in the reformatted input record (perhaps resulting in a larger
record for sorting or merging) so that overflow does not occur. Example 5 on
page 455 illustrates this technique.
v An invalid PD or ZD sign or digit results in a data exception (0C7 ABEND); 0-9
are invalid for the sign and A-F are invalid for the digit. For example, a ZD
value such as 3.5 (X'F34BF5') results in an 0C7 because "." (X'4B') is treated as an
invalid digit. ICETOOL's DISPLAY or VERIFY operator can be used to identify
decimal values with invalid digits. ICETOOL's VERIFY operator can be used to
identify decimal values with invalid signs.
v Whether or not positive summed ZD results have printable numbers depends on
whether NZDPRINT or ZDPRINT is in effect (as set by the ZDPRINT
installation option and the NZDPRINT and ZDPRINT parameters of the
OPTION statement):
If NZDPRINT is in effect, positive summed ZD results do not consist of
printable numbers, regardless of whether the original values consisted of
printable numbers or not. For example, if X'F2F3F1' (prints as '231') and
X'F3F0F6' (prints as '306') are summed, the result with NZDPRINT in effect is
X'F5F3C7' (prints as '53G').
If ZDPRINT is in effect, positive summed ZD results consist of printable
numbers, regardless of whether the original values consisted of printable
numbers or not. For example, if X'F2F3C1' (prints as '23A') and X'F3F0F6'
(prints as '306') are summed, the result with ZDPRINT in effect is X'F5F3F7'
(prints as '537').
Thus, ZDPRINT must be in effect to ensure that positive summed ZD results are
printable.
Unsummed positive ZD values retain their original signs, regardless of whether
NZDPRINT or ZDPRINT is in effect. For example, if X'F2F8C5' is not summed,
it remains X'F2F8C5' (prints as '28E'). OUTFIL's OUTREC parameter can be used
to ensure that all summed or unsummed ZD values are printable, as illustrated
by Example 4 later in this section.
v If input records are reformatted by INREC or E15, SUM must refer to fields in
the appropriate reformatted record (see the preceding description of p).
v Summary fields must not be control fields. They must not overlap control fields,
or each other, and must not overlap the RDW.
453

v FL (hexadecimal floating-point) values to be summed can be normalized or
unnormalized. However, the resulting FL values are always normalized.
Normalization processing by the hardware can produce different sums for FL
values summed in different orders.
v Exponent overflow for summed FL values results in an exponent overflow
exception (0CC ABEND)
v Exponent underflow for summed FL values results in a true zero result.
v When records are summed, you can predict which record is to receive the sum
(and be retained) and which record is to be deleted only when EQUALS is in
effect, overflow does not occur, and the BLOCKSET technique is used. In this
case, the first record (based on the sequence described under the discussion of
the EQUALS or NOEQUALS parameter of the OPTION control statement on
page 173) is chosen to contain the sum.
Fields other than summary fields remain unchanged and are taken from the
record that receives the sum.
v You can control the action that DFSORT takes when overflow occurs for BI, FI,
PD or ZD values with the OVFLO parameter as described in OPTION control
v
v DFSORT issues a message and terminates processing if a SUM statement is
specified for a tape work data set sort or Conventional merge.
v DFSORT does not support the XSUM parameter provided by a competitive sort
product to write records deleted by SUM processing to a SORTXSUM DD data
set. However, ICETOOL's SELECT operator can perform the same function as
XSUM with FIELDS=NONE. For example, this ICETOOL job:
//S1EXEC PGM=ICETOOL
//TOOLMSG DD SYSOUT=*
//DFSMSG DD SYSOUT=*
//SORTIN DD DSN=...
//SORTOUT DD DSN=...
//SORTXSUM DD DSN=...
//TOOLIN DD *
SELECT FROM(SORTIN) TO(SORTOUT)ON(5,4,CH) FIRST DISCARD(SORTXSUM)
/*
is equivalent to this XSUM job:

//S1 EXEC PGM=ICEMAN
//SORTIN DD DSN=...
//SORTOUT DD DSN=...
//SORTXSUM DD DSN=...
//SYSIN DD *
SUM FIELDS=NONE,XSUM
/*
Tip: You can also perform additional functions with ICETOOL's SELECT
operator that are not available with XSUM. See Chapter 7, Using ICETOOL, on
page 563 for complete details of ICETOOL's SELECT operator.
Adding summary fieldsexamples

Example 1
SUM FIELDS=(15021,8,PD,15011,4,FI)
454

This statement designates an 8-byte packed decimal field at byte 15021, and a
4-byte fixed-integer field at byte 15011, as summary fields.
Example 2
SUM FIELDS=NONE
This statement illustrates the elimination of duplicate records.
Example 3
SUM FIELDS=(41,8,49,4),FORMAT=ZD
OPTION ZDPRINT
These statements illustrate the use of the FORMAT operand and the ZDPRINT
option. The SUM statement designates two zoned decimal fields, one 8 bytes long
starting at byte 41, and the other 4 bytes long starting at byte 49. As a result of the
ZDPRINT option, the positive summed ZD values will be printable. Note,
however, that the ZDPRINT option does not affect ZD values which are not
summed due to overflow or unique keys. The next example shows how to use
OUTFIL to make all summary fields printable.
Example 4
SUM FIELDS=(41,8,49,4),FORMAT=ZD
OUTFIL OUTREC=(1,40,41,8,ZD,M11,49,4,ZD,M11,53,28)
These statements illustrate the use of the OUTFIL statement to ensure that all
positive ZD summary fields in the output data set are printable. Whereas the
ZDPRINT option affects only positive summed ZD fields, OUTFIL can be used to
edit positive or negative BI, FI, PD, or ZD values, whether they are summed or
not. OUTFIL can also be used to produce multiple output data sets, reports, and so
on. See OUTFIL control statements on page 223 for complete details about
OUTFIL processing.
Note: For purposes of illustration, this example assumes that the input records are
80 bytes long.
Example 5
* Add Z0 before the ZD SUM field to prevent overflow.
* Add P0 before the PD SUM field to prevent overflow.
INREC FIELDS=(1,10, Copy bytes before ZD SUM field
11:C0,12:11,4,
Add Z0 before ZD SUM field
16:15,6,
Copy bytes after ZD SUM field
22:X00,23:21,2, Add P0 before PD SUM field
25:23,5,
Copy SORT field
30:28,53)
Copy bytes after SORT field
* Sort on key in its new position.
* Sum on the expanded ZD and PD fields in
* their new positions.
SUM FIELDS=(11,5,ZD,22,3,PD)
These statements illustrate a technique for preventing overflow of summed fields

by using INREC to make the fields larger before they are summed.
The fields that might overflow when they are summed are a 4 byte ZD field
starting at position 11 and a 2 byte PD field starting at position 21. In order to
prevent them from overflowing, we expand each field on the left with an
455

appropriate zero byte; C'0' (Z'0') for the ZD field and P'0' (X'00') for the PD field.
We can then sum on the new 5 byte ZD field and on the new 3 byte PD field.
Note that adding these extra bytes increases the length of the record and changes
the starting position of various fields. In the SORT and SUM statements, we must
specify the starting positions of the fields in the reformatted record rather than the
starting positions of the fields in the input record. For example, although the SORT
field starts in position 23 in the input record, we must use its starting position of
25 in the reformatted record.
456
Chapter 4. Using a JOINKEYS application for joining two files

Overview
You can perform various types of "join" applications on two files (F1 and F2) by
one or more keys with DFSORT using the following statements:
JOINKEYS
You must specify two JOINKEYS statements; one for the F1 file and
another for the F2 file. A separate subtask will be used to process each file
and a main task will be used to process the joined records from the two
files.
Each JOINKEYS statement must specify the ddname of the file it applies to
and the starting position, length and sequence of the keys in that file. You
can also optionally specify if the file is already sorted by the keys and if
sequence checking of the keys is not needed; if the file has fixed-length or
variable-length records; to stop reading the file after n records; a 2-byte id
to be used for the message and control data set for the subtask used to
process the file, and if a subset of the records is to be processed based on a
logical expression.
JOIN
If you don't specify a JOIN statement, only paired records from F1 and F2
are kept and processed by the main task as the joined records (inner join).
You can optionally specify a JOIN statement to have the main task keep
and process: unpaired F1 records as well as paired records (left outer join);
unpaired F2 records as well as paired records (right outer join); unpaired
F1 and F2 records as well as paired records (full outer join); only unpaired
F1 records; only unpaired F2 records, or only unpaired F1 and F2 records.
REFORMAT
You would normally specify a REFORMAT statement to indicate the F1
and/or F2 fields you want in the joined records. You can optionally specify
an indicator of where the key was found, and a FILL character to be used
for missing bytes. If a JOIN statement with ONLY is specified, the
REFORMAT statement is optional.
F1 and F2 can be any type of sequential or VSAM file supported by DFSORT for
SORTIN and can have different attributes (for example, F1 can have RECFM=FB
and LRECL=100 and F2 can have RECFM=VB and LRECL=254, or F1 can be a
VSAM ESDS and F2 can be a PDS member).
F1 will be sorted or copied by "subtask1". An E35 exit will be used to pass the
needed fields from the F1 records to the "main task" (an intermediate output data
set is not used or required). A subset of the DFSORT statements, such as
INCLUDE, OMIT, INREC, SUM and OPTION, will be available for processing the
F1 records.
F2 will be sorted or copied by "subtask2". An E35 exit will be used to pass the
needed fields from the F2 records to the "main task" (an intermediate output data
set is not used or required). A subset of the DFSORT statements, such as
INCLUDE, OMIT, INREC, SUM and OPTION, will be available for processing the
F2 records.
457
Overview
The "main task" will use an E15 to join the records passed from the E35 of
subtask1 and E35 of subtask2. The joined records will be processed as the input
records for a sort or copy application. Most of the control statements and options
available for a DFSORT sort or copy application will be available for processing the
joined records.
Note: Since a JOINKEYS application uses three tasks, it can require more storage
than a regular DFSORT application. You may need to use REGION=0M for some
JOINKEYS applications.
JOINKEYS application processing

Figure 9 is a pictorial representation of the processing performed for a JOINKEYS
application, and the order in which the various functions are performed.
SORTJNF1 (F1 file)
Subtask1 for F1
Messages: JNF1JMSG
Control: JNF1CNTL
<SKIPREC>
<E15>
<INCLUDE/OMIT>
<STOPAFT>
<INREC>
COPY or SORT (1)
<SUM>
E35 passes F1 fields
to main task's E15 (3)
SORTJNF2 (F2 file)
Subtask2 for F2
Messages: JNF2JMSG
Control: JNF2CNTL
<SKIPREC>
<E15>
<INCLUDE/OMIT>
<STOPAFT>
<INREC>
COPY or SORT (2)
<SUM>
E35 passes F2 fields
to main task's E15 (4)
Main task for joined records

Messages: SYSOUT
Control: SYSIN, DFSPARM,
SORTCNTL, parmlist
E15 joins fields passed by
subtask1's E35 and subtask2's E35
as directed by JOINKEYS, JOIN and
REFORMAT statements (5)
<INCLUDE/OMIT>
<STOPAFT>
<INREC>
<SORT or COPY - one required>
<SUM>
<OUTREC>
<E35/SORTOUT/OUTFIL - one required>
Figure 9. JOINKEYS Application Processing
Legend for Figure 9
458
JOINKEYS Application Processing

v (1) COPY is used automatically for subtask1 if the SORTED operand is specified
in the F1 JOINKEYS statement. Otherwise, SORT is used automatically with the
binary keys specified in the FIELDS operand of the F1 JOINKEYS statement.
v (2) COPY is used automatically for subtask2 if the SORTED operand is specified
in the F2 JOINKEYS statement. Otherwise, SORT is used automatically with the
binary keys specified in the FIELDS operand of the F2 JOINKEYS statement.
v (3) An E35 exit is used for subtask1 automatically. (An intermediate output file is
not required or used.)
v (4) An E35 exit is used for subtask2 automatically. (An intermediate output file is
not required or used.)
v (5) An E15 exit is used for the main task automatically. (Intermediate input files
are not required or used.)
For direct invocation of DFSORT (for example, PGM=SORT), the JOINKEYS, JOIN
and REFORMAT statements can be specified in SYSIN or DFSPARM. For program
invocation of DFSORT (for example, LINK EP=SORT), the JOINKEYS, JOIN and
REFORMAT statements can be specified in the caller's parameter list, in
SORTCNTL or in DFSPARM.
Subtask1 reads the F1 file. It uses COPY or SORT, and an E35 exit (with no output
data set) automatically to pass the needed F1 fields to the main task's E15 exit.
Subtask1 can optionally use the other listed control statements from JNF1CNTL.
Subtask1 messages are displayed in JNF1JMSG.
Subtask2 reads the F2 file. It uses COPY or SORT, and an E35 exit (with no output
data set) automatically to pass the needed F2 fields to the main task's E15 exit.
Subtask2 can optionally use the other listed control statements from JNF2CNTL.
Subtask2 messages are displayed in JNF2JMSG.
The main task uses an E15 exit automatically. The E15 creates the joined records by
accessing the F1 fields passed by subtask1's E35 and the F2 fields passed by
subtask2's E35. The main task writes the output to SORTOUT and/or OUTFIL or
uses a supplied E35 exit to "delete" all of the records. For direct invocation of
DFSORT, the main task can optionally use the other listed control statements from
SYSIN or DFSPARM. For program invocation of DFSORT, the main task can
optionally use the other listed control statements from the caller's parameter list, in
SORTCNTL or in DFSPARM. The main task's messages are displayed in SYSOUT.
The starting position in the fields you specify for the subtasks or main task must
reflect any reformatting of the records you do at each stage. This includes using
the starting positions of the joined records for main task functions.
Examples:
v If you specify INREC and SUM statements in JNF1CNTL for subtask1, the
FIELDS operands in the JOINKEYS, REFORMAT, and SUM statements must
specify the starting positions in the F1 records as reformatted by INREC.
v If you specify an INCLUDE statement in SYSIN for the main task, the COND
operand must specify the starting positions in the joined records as reformatted
by REFORMAT.
Sample JOINKEYS applications

Here are the JCL and control statements for a simple JOINKEYS application to do
an inner join (also known as a cartesian join) which joins all paired records.
459

//S1
EXEC PGM=SORT
//SORTJNF1 DD DSN=INPUT1,DISP=SHR
//SORTJNF2 DD DSN=INPUT2,DISP=SHR
//SYSIN DD *
* Control statements for JOINKEYS application
JOINKEYS FILE=F1,FIELDS=(15,2,A,7,4,A)
REFORMAT FIELDS=(F2:1,70,F1:1,60)
* Control statements for main task (joined records)
SORT FIELDS=COPY
/*
Here are the JCL and control statements for a JOINKEYS application which omits
certain records and normalizes keys for F1 and F2, and retains only sorted,
non-duplicate, unpaired records from F1. The F1 and F2 files are already in order
by the specified JOINKEYS FIELDS. DFSORT symbols are used for various fields
and constants.
//S2
EXEC PGM=SORT
//SYMNAMES DD *
IN1_dept,11,3,ch
IN1_target,J82
IN1_normal_key,27,5,ZD
IN1_output,1,60
IN2_dept,14,3,ch
IN2_target,M25
IN2_PD_key,21,3,PD
IN2_normal_key,=,5,ZD
join_sort_key,45,8,UFF
//FIX DD DSN=F.INPUT,DISP=SHR
//VAR DD DSN=V.INPUT,DISP=SHR
//SORTOUT DD DSN=FIXOUT1,DISP=(NEW,CATLG,DELETE),
// SPACE=(CYL,(5,5)),UNIT=SYSDA
//JNF1CNTL DD *
* Control statements for subtask1 (F1)
OMIT COND=(IN1_dept,EQ,IN1_target)
INREC OVERLAY=(IN1_normal_key:IN1_normal_key,TO=ZDF,LENGTH=5)
//JNF2CNTL DD *
OMIT COND=(IN2_dept,EQ,IN2_target)
INREC OVERLAY=(IN2_PD_key:IN2_PD_key,TO=ZDF,LENGTH=5)
//SYSIN DD *
JOINKEYS F1=FIX,FIELDS=(IN1_normal_key,D),SORTED
JOINKEYS F2=VAR,FIELDS=(IN2_normal_key,D),SORTED
JOIN UNPAIRED,F1,ONLY
REFORMAT FIELDS=(F1:IN1_output)
SORT FIELDS=(join_sort_key,A)
SUM FIELDS=NONE
/*
JCL for a JOINKEYS application

The required JCL statements used for a JOINKEYS application are as follows. See
z/OS DFSORT Application Programming Guide for general information on DFSORT
message, control, input, output, work and symbols data sets.
//stepname EXEC PGM=SORT
Invokes DFSORT. PGM=ICEMAN can also be used.
460

Note: Since a JOINKEYS application uses three tasks, it can require more
storage than a regular DFSORT application. You may need to use REGION=0M
for some JOINKEYS applications.
//SYSOUT DD
Messages from the main task. An alternate ddname can be supplied with
MSGDDN=ddname in //DFSPARM.
//SORTJNF1 DD
F1 input file (read by subtask1). The ddname is SORTJNF1 (or ccccJNF1 if
SORTDD=cccc is in effect) if FILE=F1 or FILES=F1 is specified on the
JOINKEYS statement. An alternate ddname can be supplied with F1=ddname
on the JOINKEYS statement.
Note: The F1 data set is treated as a SORTIN data set by subtask1 and is
subject to the rules for a SORTIN data set documented in Chapter 2, Invoking
DFSORT with Job Control Language, on page 27.
//SORTJNF2 DD
F2 input file (read by subtask2). The ddname is SORTJNF2 (or ccccJNF2 if
SORTDD=cccc is in effect) if FILE=F2 or FILES=F2 is specified on the
JOINKEYS statement. An alternate ddname can be supplied with F2=ddname
on the JOINKEYS statement.
Note: The F2 data set is treated as a SORTIN data set by subtask2 and is
subject to the rules for a SORTIN data set documented in Chapter 2, Invoking
DFSORT with Job Control Language, on page 27.
//SORTOUT DD or //outfil DD
Output of joined records (written by main task). An alternate ddname can be
supplied with SORTOUT=ddname or SORTDD=cccc in //DFSPARM or with
the OUTFIL statement in //SYSIN or //DFSPARM.
//DFSPARM DD, //SYSIN DD or //SORTCNTL DD
Control statements for the main task including JOINKEYS, JOIN, REFORMAT,
OPTION, SORT, INCLUDE or OMIT, SUM, OUTREC, RECORD, ALTSEQ,
MODS and OUTFIL. If SKIPREC or E15 is specified, it will not be used.
DFSPARM can be used if DFSORT is invoked directly or called from a
program. SYSIN can be used if DFSORT is invoked directly. SORTCNTL can be
used if DFSORT is called from a program; an alternate ddname of ccccCNTL
can be supplied with SORTDD=cccc in DFSPARM.
Note: If DFSORT is called from a program, the control statements for the main
task can be supplied using the parameter list.
The optional JCL statements used for a JOINKEYS application are as follows:
SYMNAMES DD
DFSORT symbols for the main task, subtask1 and subtask2.
SYMNOUT DD
Listing of symbols supplied via the SYMNAMES DD.
JNF1JMSG DD
Messages from subtask1. This message data set will be dynamically allocated
with SYSOUT=*, RECFM=FBA, LRECL=121 and BLKSIZE=121 if a JNF1JMSG
DD statement is not supplied. If a JNF1JMSG DD statement is supplied, this
461

message data set will be given the same attributes as a //SYSOUT message
data set. An alternate ddname of idF1JMSG can be supplied with TASKID=id
on the JOINKEYS statement for F1.
JNF2JMSG DD
Messages from subtask2. This message data set will be dynamically allocated
with SYSOUT=*, RECFM=FBA, LRECL=121 and BLKSIZE=121 if a JNF2JMSG
DD statement is not supplied. If a JNF2JMSG DD statement is supplied, this
message data set will be given the same attributes as a //SYSOUT message
data set. An alternate ddname of idF2JMSG can be supplied with TASKID=id
on the JOINKEYS statement for F2.
JNF1CNTL DD
Control statements for subtask1 including INCLUDE or OMIT, OPTION,
MODS, RECORD, ALTSEQ, INREC and SUM. If E35 is specified, it will not be
used. The following control statements must not be specified: JOINKEYS,
JOIN, REFORMAT, MERGE, OUTFIL, OUTREC or SORT.
An alternate ddname of idF1CNTL can be supplied with TASKID=id on the
JOINKEYS statement for F1.
JNF2CNTL DD
Control statements for subtask2 including INCLUDE or OMIT, OPTION,
MODS, RECORD, ALTSEQ, INREC and SUM. If E35 is specified, it will not be
used. The following control statements must not be specified: JOINKEYS,
JOIN, REFORMAT, MERGE, OUTFIL, OUTREC or SORT.
An alternate ddname of idF2CNTL can be supplied with TASKID=id on the
JOINKEYS statement for F2.
SORTWKdd DD
Work data sets for the main task. Not recommended since dynamic allocation
of work data sets is preferred. Alternate ddnames of ccccWKdd can be
supplied with SORTDD=cccc in //DFSPARM.
JNF1WKdd DD
Work data sets for subtask1. Not recommended since dynamic allocation of
work data sets will be enabled. Alternate ddnames of idF1WKdd can be
supplied with TASKID=id on the JOINKEYS statement for F1.
JNF2WKdd DD
Work data sets for subtask2. Not recommended since dynamic allocation of
work data sets will be enabled. Alternate ddnames of idF2WKdd can be
supplied with TASKID=id on the JOINKEYS statement for F2.
JOINKEYS statements
,
JOINKEYS
462
FILE=F1
FILES=F1
F1=ddname
FILE=F2
FILES=F2
F2=ddname
, FIELDS= (
p,m,s

,
SORTED
, NOSEQCK
F
V
STOPAFT=n
TASKID=id
INCLUDE=
ALL
NONE
OMIT=
ALL
NONE
TYPE=
Two JOINKEYS statements are required for a JOINKEYS application; one for the F1
file and the other for the F2 file:
v FILE=F1 or F1=ddname must be used to indicate that the JOINKEYS statement
applies to the F1 input file. FILE=F1 associates the F1 file with a ddname of
SORTJNF1. You can use a different ddname for the F1 file by specifying
F1=ddname. For simplicity, we will use SORTJNF1 when referring to the
ddname for the F1 file.
v FILE=F2 or F2=ddname must be used to indicate that the JOINKEYS statement
applies to the F2 input file. FILE=F2 associates the F2 file with a ddname of
SORTJNF2. You can use a different ddname for the F2 file by specifying
F2=ddname. For simplicity, we will use SORTJNF2 when referring to the
ddname for the F2 file.
FILE=F1, FILES=F1 or F1=ddname
FILE=F1
FILES=F1
F1=ddname
Must be used for the JOINKEYS statement associated with the F1 file. FILE=F1
(or FILES=F1) specifies a ddname of SORTJNF1 for the F1 file. F1=ddname can
be used to specify any valid ddname for the F1 file. Do not use the same
ddname for the F1 file and the F2 file.
When FILE=F1, FILES=F1 or F1=ddname is specified, the other operands of the
JOINKEYS statement apply to the F1 file.
FILE=F2, FILES=F2 or F2=ddname
FILE=F2
FILES=F2
F2=ddname
Must be used for the JOINKEYS statement associated with the F2 file. FILE=F2
(or FILES=F2) specifies a ddname of SORTJNF2 for the F2 file. F2=ddname can
be used to specify any valid ddname for the F2 file. Do not use the same
ddname for the F1 file and the F2 file.
463

When FILE=F2, FILES=F2 or F2=ddname is specified, the other operands of the
JOINKEYS statement apply to the F2 file.
FIELDS
,
FIELDS= (
p,m,s
Must be specified to indicate the starting position, length and order (ascending
or descending) of the keys in the input file. The keys will be treated as binary,
so they must be "normalized". For example, if the keys are actually zoned
decimal, they must have all C and D signs, or all F and D signs. You can use
an INREC statement in JNF1CNTL and/or JNF2CNTL to normalize the keys
for the F1 file and/or F2 file, respectively, if appropriate.
Each pair of keys for the F1 and F2 files must match with respect to length and
order, but can start in different positions. For example, if the first key for the
F1 file is 5 bytes ascending and the second key for the F1 file is 3 bytes
descending, the first key for the F2 file must be 5 bytes ascending and the
second key for the F2 file must be 3 bytes descending.
If a variable-length record is too short to contain a key you specify, the short
key value will be compared using binary zeros for the missing bytes.
p
specifies the starting position of the key. The first data byte of a
fixed-length record is in position 1. The first data byte of a
variable-length record is in position 5 after the 4-byte RDW. p can be 1
to 32752 but all fields must be completely contained within the first
32752 bytes of the record.
specifies the length of the key. The total length of all keys must not
exceed 4080 bytes. All fields must be completely contained within the
first 32752 bytes of the record.
The length for each pair of F1 and F2 keys must match.
specifies the order of the key. Use A for ascending or D for descending.
The order for each pair of F1 and F2 keys must match.

JOINKEYS F1=IN1,FIELDS=(22,3,A,55,9,D)
File F1 is processed using the ddname IN1, the ascending key in positions
22-24 and the descending key in positions 55-63. File F2 is processed using the
ddname IN2, the ascending key in positions 15-17 and the descending key in
positions 1-9.
SORTED
SORTED
By default, DFSORT will sort the input file by the specified keys. If the records
of the input file are already in sorted order by the specified keys, you can use
the SORTED operand to tell DFSORT to copy the records rather than sort
them. This can improve performance. DFSORT will terminate if the copied
records are not in the order specified by the keys unless you specify the
NOSEQCK operand.
464

JOINKEYS FILE=F1,FIELDS=(22,3,A),SORTED
JOINKEYS FILE=F2,FIELDS=(15,3,A)
File F1 is copied using the ddname SORTJNF1 and the ascending key in
positions 22-24. The SORTJNF1 records will be checked for the correct key
order. File F2 is sorted using the ddname SORTJNF2 and the ascending key in
positions 15-17.
If you use the SORTED operand, statements and options only available for a
sort application, such as SUM, will be ignored for the subtask that copies the
input file.
NOSEQCK
NOSEQCK
If you specify the SORTED operand and know that the records of the input file
are already in sorted order by the specified keys, you can use the NOSEQCK
operand to tell DFSORT not to check the order of the records. This can
improve performance.
JOINKEYS FILE=F1,FIELDS=(22,3,A),SORTED,NOSEQCK
JOINKEYS FILE=F2,FIELDS=(15,3,A),SORTED
File F1 is copied using the ddname SORTJNF1 and the ascending key in
positions 22-24. The SORTJNF1 records will be not be checked for the correct
key order. File F2 is copied using the ddname SORTJNF2 and the ascending
key in positions 15-17. The SORTJNF2 records will be checked for the correct
key order.
If the records are not actually in order by the specified keys and you use
NOSEQCK, the output may be incorrect.
The NOSEQCK operand is ignored if the SORTED operand is not specified.
TYPE
TYPE=
F
V
TYPE=V can be used to tell DFSORT to use variable-length processing for a

VSAM input file. TYPE=F (the default) can be used to tell DFSORT to use
fixed-length processing for a VSAM input file.
JOINKEYS F1=VSAM1,FIELDS=(22,3,A),TYPE=V
JOINKEYS F2=VSAM2,FIELDS=(15,3,A),TYPE=F
VSAM file F1 is sorted as variable-length records using the ddname VSAM1

and the ascending key in positions 22-24. (Remember that for TYPE=V VSAM
processing, DFSORT adds an RDW in positions 1-4 which you must account
for when specifying the starting position.) VSAM file F2 is sorted as
fixed-length records using the ddname VSAM2 and the ascending key in
positions 15-17. (Remember that for TYPE=F VSAM processing, DFSORT does
not add an RDW.)
STOPAFT
465

STOPAFT=n
Can be used to specify the maximum number of records (n) you want the
subtask for the input file to accept for sorting or copying (accepted means read
from the input file and not deleted by INCLUDE or OMIT).
can be up to 28 digits with up to 15 significant digits.

JOINKEYS FILE=F1,STOPAFT=5,FIELDS=(32,4,A)
JOINKEYS FILE=F2,STOPAFT=10,FIELDS=(32,4,A)
The first 5 input records from SORTJNF1 and the first 10 input records from
SORTJNF2 will be processed.
You can use STOPAFT=n on the OPTION statement in JNF1CNTL (for the F1
file) or JNF2CNTL (for the F2 file) instead of specifying STOPAFT=n on the
JOINKEYS statement.
For example, instead of the STOPAFT operands in the JOINKEYS statements
shown previously, you could specify:
//SYSIN DD *
...
//JNF1CNTL DD *
OPTION STOPAFT=5
//JNF2CNTL DD *
OPTION STOPAFT=10
TASKID
TASKID=id
By default, DFSORT uses the following ddnames for the subtasks:

v JNF1JMSG for the subtask1 (F1 file) message data set.
v JNF1CNTL for the subtask1 (F1 file) control data set.
v JNF1WKdd for the subtask1 (F1 file) work data sets.
v JNF2JMSG for the subtask2 (F2 file) message data set.
v JNF2CNTL for the subtask2 (F2 file) control data set.
v JNF2WKdd for the subtask2 (F2 file) work data sets.
The TASKID=id operand can be used to change the first two characters from
JN to the specified id characters. The same id can be used for the F1 and F2
ddnames, or a different id can be used for each.
JOINKEYS F1=IN1,FIELDS=(1,10,A),TASKID=C1
JOINKEYS F2=IN2,FIELDS=(22,10,A),TASKID=C1
C1F1JMSG, C1F1CNTL and C1F1WKdd will be used for subtask1 (F1 file).
C1F2JMSG, C1F2CNTL and C1F2WKdd will be used for subtask2 (F2 file).
If you specify:
JOINKEYS F1=IN1,FIELDS=(1,10,A),TASKID=I1
JOINKEYS F2=IN2,FIELDS=(22,10,A),TASKID=I2
I1F1JMSG, I1F1CNTL and I1F1WKdd will be used for subtask1 (F1 file).
I2F2JMSG, I2F2CNTL and I2F2WKdd will be used for subtask2 (F2 file).
466

The TASKID=id operand can be useful when you are doing multiple
JOINKEYS applications and want to separate the messages for each application
and/or specify different control statements or work data sets for different
subtasks.
INCLUDE or OMIT
INCLUDE
OMIT
ALL
NONE
Can be used to specify criteria for the records you want the subtask for the
input file to include or omit for sorting or copying. You can use the same
logical expressions, ALL or NONE in the same way as for the INCLUDE or
OMIT operand of the OUTFIL statement. See OUTFIL control statements on
JOINKEYS FILE=F1,FIELDS=(35,8,A),
OMIT=(5,4,CH,EQ,CABCD)
JOINKEYS FILE=F2,FIELDS=(37,8,A),
INCLUDE=(1,20,SS,EQ,CNO)
Only records without 'ABCD' in positions 5-8 will be processed from file F1.
Only records with 'NO' somewhere in positions 1-20 will be processed from
file F2.
Although the INCLUDE and OMIT operands are available on the JOINKEYS
statement, it is recommended that you specify an INCLUDE or OMIT
statement in JNF1CNTL or JNF2CNTL instead for ease of use.
For example, instead of the OMIT and INCLUDE operands in the JOINKEYS
statements shown previously, you could specify:
//SYSIN DD *
...
//JNF1CNTL DD *
OMIT COND=(5,4,CH,EQ,CABCD)
//JNF2CNTL DD *
INCLUDE COND=(1,20,SS,EQ,CNO)
JOIN statement
,
JOIN UNPAIRED

F1
F2
ONLY
If you don't specify a JOIN statement for a JOINKEYS application, only paired
records from F1 and F2 are kept and processed by the main task as the joined
records. This is known as an inner join.
You can change which records are kept and processed by the main task as the
joined records by specifying a JOIN statement. You must specify the UNPAIRED
467

operand. The F1, F2 and ONLY operands are optional. The JOIN operands you
specify indicate the joined records to be kept and processed by the main task as
follows:
UNPAIRED,F1,F2 or UNPAIRED
Unpaired records from F1 and F2 as well as paired records. This is known as a
full outer join.
UNPAIRED,F1
Unpaired records from F1 as well as paired records. This is known as a left
outer join.
UNPAIRED,F2
Unpaired records from F2 as well as paired records. This is known as a right
outer join.
UNPAIRED,F1,F2,ONLY or UNPAIRED,ONLY
Unpaired records from F1 and F2.
UNPAIRED,F1,ONLY
Unpaired records from F1.
UNPAIRED,F2,ONLY
Unpaired records from F2.
REFORMAT statement
,
,
REFORMAT FIELDS= (
F1:
F2:
p,m
F1:
F2:
)
p,m

,FILL=byte
A REFORMAT statement can always be used for a JOINKEYS application, and is

required unless a JOIN statement with the ONLY operand is specified. The
REFORMAT statement indicates the fields from the F1 file and/or the F2 file you
want to include in the joined records, and the order in which you want the fields
to appear. You can also include an indicator of where the key was found in the
joined records ('B' for key found in F1 and F2, '1' for key found in F1 only, or '2' for
key found in F2 only).
If the REFORMAT statement only defines position with length (p,m) fields, each
joined record will be fixed-length (TYPE=F) with a LENGTH equal to the total
length of all of the p,m fields. The maximum length for TYPE=F joined records is
32760 bytes. The F1 and F2 files can both be fixed-length, both be variable-length,
or can be mixed fixed-length and variable-length.
If the REFORMAT statement defines a position without a length (p without m)
field, each joined record will be variable-length (TYPE=V) with a LENGTH equal
to the total length of all of the p,m fields plus the variable length of each p field
(one for F1 and/or one for F2). The maximum length for TYPE=V joined records is
32767 bytes. The F1 and F2 files can both be variable-length or can be mixed
fixed-length and variable-length. However:
v The first field must contain the RDW (1,n with n equal to or greater than 4) and
must be from a variable-length file (F1 or F2).
468

v The position without length fields (one from the F1 file and/or one from the F2
file) must be the last fields specified and must be from a variable-length file (F1
and/or F2).
For joined records created from paired records, any F1 fields specified will be
extracted from the F1 record and any F2 fields specified will be extracted from the
F2 record.
For joined records created from unpaired F1 records, any F1 fields specified will be
extracted from the F1 record and any F2 fields specified will be filled with the
specified FILL character or blanks by default. However, if the F2 file is
variable-length, any specified F2 RDW fields (1,n with n equal to or less than 4)
will be filled with binary zeros.
For joined records created from unpaired F2 records, any F2 fields specified will be
extracted from the F2 record and any F1 fields specified will be filled with the
specified FILL character or blanks by default. However, if the F1 file is
variable-length, any specified F1 RDW fields (1,n with n equal to or less than 4)
will be filled with binary zeros.
If a JOIN statement with the ONLY operand is specified, the REFORMAT
statement does not have to be specified. In this case, the layout of the joined
records will depend on the specified JOIN statement operands as follows:
The joined records will be the original unpaired F1 records. If the F1
records are fixed-length, the joined records will be fixed-length. If the F1
records are variable-length, the joined records will be variable-length.
The joined records will be the original unpaired F2 records. If the F2
records are fixed-length, the joined records will be fixed-length. If the F2
records are variable-length, the joined records will be variable-length.
JOIN UNPAIRED,F1,F2,ONLY or JOIN UNPAIRED,ONLY
The joined records will be variable-length. If the F1 records are
fixed-length, each unpaired F1 record will be variable-length with an RDW
followed by the original F1 record. If the F1 records are variable-length,
each unpaired F1 record will be the original F1 record. If the F2 records are
fixed-length, each unpaired F2 record will be variable-length with an RDW
followed by the original F2 record. If the F2 records are variable-length,
each unpaired F2 record will be the original F2 record.
In all cases, the TYPE and LENGTH will be set as appropriate for the joined
records.
For joined records with TYPE=F, the maximum LENGTH is 32760.
For joined records with TYPE=V, the maximum LENGTH is 32767. Note that the
RECFM of the output file must be VS or VBS in order to allow output records
greater than 32756.
FIELDS
469

,
,
FIELDS= (
F1:
F2:
p,m
F1:
F2:
p,m
Must be specified to indicate the starting position and length of each field from
the F1 file and/or the F2 file to be included in the joined records, and
optionally an indicator of where the key was found. The fields and indicator
will be included in the joined records in the order in which they are specified.
F1:
indicates the following fields up to the next Fn: or end of the FIELDS
operand are from the F1 file.
F2:
indicates the following fields up to the next Fn: or end of the FIELDS
operand are from the F2 file.
p,m
specifies the starting position and length of a fixed field. p specifies the
starting position of the field. The first data byte of a fixed-length
record is in position 1. The first data byte of a variable-length record is
in position 5 after the 4-byte RDW. p can be 1 to 32767. m specifies the
length of the field. m can be 1 to 32760. All fields must be completely
contained within the first 32767 bytes of the record.
indicates a 1-byte indicator is to be included in each joined record. The

indicator will be set to one of the following values in each paired or
unpaired joined record, as appropriate:
v 'B' - the key was found in F1 and F2.
v '1' - the key was found in F1, but not in F2.
v '2' - the key was found in F2, but not in F1.
Only one ? can be specified in the FIELDS operand. If ? is not the last
item, it must be followed by F1: or F2:.
For TYPE=F joined records, the indicator can appear anywhere in the
record. For example, each of the following is valid:
* Put indicator in position 1 of each joined record.
REFORMAT FIELDS=(?,F1:1,20,F2:5,8)
REFORMAT FIELDS=(F1:1,20,?,F1:31,9)
REFORMAT FIELDS=(F2:11,20,6,4,?)
For TYPE=V joined records, the indicator must appear in the fixed part
of the record, that is, after the RDW and before the position without
length fields. For example, the following is valid:
REFORMAT FIELDS=(F1:1,4,?,F1:11)
p without m gives the starting position of a variable field. Only one p

without m field can be specified for F1 and only one p without m field
can be specified for F2.
If either or both p without m fields are specified, they must be the last
fields in the FIELDS operand. In this case, the first field in the FIELDS
470

operand must be from a variable-length file (F1 or F2) and must
include the RDW (1,n where n is equal to or greater than 4).
Example with just p,m fields:
REFORMAT FIELDS=(F1:27,5,1,8,F2:19,20,F1:1201,15)
Example with one p without m field:

REFORMAT FIELDS=(F1:1,4,F2:6,25,92,2,F1:8,9,32)
Example with two p without m fields:

REFORMAT FIELDS=(F2:1,9,21,3,F1:101,7,28,9,122,F2:26)
FILL
FILL=byte
The FILL operand can be used to override DFSORT's default fill byte of a
blank (X'40'). The fill byte is used in the following situations:
v A p,m (fixed) field is specified for a file (F1 or F2) with variable-length
records, and the field extends beyond the end of a record. Each missing byte
is replaced with the fill byte. For example, if a variable-length F1 record is 20
bytes long and the REFORMAT statement has:
REFORMAT FIELDS=(F1:1,30,41),FILL=C*
bytes 21-30 of the joined record are filled with asterisks.

v The REFORMAT statement has a p,m (fixed) field from the F1 file and a
joined record is being created from an unpaired F2 record. For example if
the following are specified:
JOIN UNPAIRED,F2
REFORMAT FIELDS=(F1:21,5,F2:1,10),FILL=X00
and an unpaired F2 record is found, the joined record will have 5 binary
zero bytes followed by bytes 1-10 from the F2 record.
v The REFORMAT statement has a p,m (fixed) field from the F2 file and a
joined record is being created from an unpaired F1 record. For example if
the following are specified:
JOIN UNPAIRED,F1,F2
REFORMAT FIELDS=(F1:21,5,F2:1,10),FILL=X00
and an unpaired F1 record is found, the joined record will have bytes 21-25
of the F1 record followed by 10 binary zero bytes. (Since UNPAIRED,F1,F2 is
specified, if an unpaired F2 record is found, the joined record will have 5
binary zero bytes followed by bytes 1-10 of the F2 record.)
byte
specifies the fill byte. Permissible values are C'x' and X'yy'.
C'x'
Character fill byte. x must be one EBCDIC character. If you want to use
an apostrophe as the fill byte, you must specify it as C''''.
X'yy'
Hexadecimal fill byte. yy must be one pair of hexadecimal digits

(00-FF).
JOINKEYS application notes

In the notes later in this section, "subtaskn" refers to subtask1 for the F1 file or
subtask2 for the F2 file.
v Since a JOINKEYS application uses three tasks, it can require more storage than
a regular DFSORT application. You may need to use REGION=0M for some
JOINKEYS applications.
471

v
DFSORT's normal syntax and continuation rules are used for the JOINKEYS,
JOIN and REFORMAT statements. See Chapter 3, Using DFSORT program
control statements, on page 81 for details.
v Control statements from DFSPARM, SYSIN (direct invocation) or SORTCNTL

(program invocation) will be used for the main task, but not for the subtasks.
Control statements from JNF1CNTL will be used for subtask1. Control
statements from JNF2CNTL will be used for subtask2.
v For the main task, the normal options in effect for a DFSORT run will be used.
The run-time options will be those from DFSPARM, SYSIN, SORTCNTL, EXEC
PARM or the caller's parameter list, as appropriate. The installation options will
be those for direct invocation or program invocation of DFSORT, as appropriate.
v For subtaskn, the options in effect for the subtask will be used with the
exceptions noted later in this section. The run-time options will be those from
JNFnCNTL. The installation options will be those for a calling program (INV,
TSOINV, TD1-TD4).
v TYPE=F processing is used for a VSAM F1 or F2 file by default. TYPE=V can be
specified in JNFnCNTL, or on the JOINKEYS statement for Fn, to override the
default of TYPE=F.
v For subtaskn, if an INCLUDE or OMIT statement or a STOPAFT or TYPE option
is specified in JNFnCNTL, it will override the corresponding option in a
JOINKEYS statement for Fn. For ease of use, it is recommended that you use an
INCLUDE or OMIT statement, and a STOPAFT or TYPE option, in JNFnCNTL
rather than an INCLUDE, OMIT, STOPAFT or TYPE operand in a JOINKEYS
statement.
v For the main task, override of statements and options is done in the normal way,
that is, DFSPARM overrides SYSIN for direct invocation of DFSORT, and
DFSPARM overrides SORTCNTL overrides the caller's parameter list for
program invocation of DFSORT. See Chapter 3, Using DFSORT program control
statements, on page 81 for details.
v The following options will be used for subtaskn and cannot be overridden: LIST,
MSGPRT=ALL, NOABEND, ESTAE, NOCHECK, EQUALS and RESINV=0.
DYNALLOC will be used automatically for subtaskn, but can be overridden by
a DYNALLOC option in JNFnCNTL.
v If an INCLUDE or OMIT statement is specified in JNFnCNTL, DFSORT will use
the following options in effect for subtaskn processing: ALTSEQ, CHALT or
NOCHALT, SZERO or NOSZERO and VLSCMP or NOVLSCMP. These options
can be specified in JNFnCNTL. The subtaskn LOCALE installation option in
effect will also be used if appropriate.
v If an INCLUDE or OMIT operand is specified on the JOINKEYS statement for
Fn, DFSORT will use these main task options in effect for subtaskn processing:
ALTSEQ, CHALT or NOCHALT, SZERO or NOSZERO and VLSCMP or
NOVLSCMP. If these options are specified in JNFnCNTL, they will be ignored
for subtaskn processing. LOCALE will not be used.
v For subtaskn, the following statements are not allowed in JNFnCNTL:
JOINKEYS, JOIN, MERGE, OUTFIL, OUTREC, REFORMAT and SORT.
v For the main task, a MERGE statement other than MERGE FIELDS=COPY is not
allowed in any source.
v If appropriate, options such as FILSZ, DYNALLOC, AVGRLEN, and so on can
be specified in JNFnCNTL for subtaskn, or in DFSPARM for the main task.
v If tape data sets are used for the F1 and F2 files, they must be on different
drives so DFSORT can read them in parallel.
472
JOINKEYS application examples

Example 1 - Paired F1/F2 records without duplicates
//JKE1 EXEC PGM=SORT
//SYSOUT
DD SYSOUT=*
//SORTJNF1 DD *
Roses
03 Red
Daisies
06 Orange
Roses
04 Pink
Daisies
02 Yellow
Roses
06 Yellow
Daisies
12 Lilac
Roses
09 Blue
/*
//SORTJNF2 DD *
Red
Lilies
InStock
Red
Roses
InStock
Orange
Daisies
SoldOut
Pink
Roses
SoldOut
Yellow
Daisies
InStock
Yellow
Roses
Ordered
Lilac
Daisies
SoldOut
White
Daisies
InStock
/*
//SYSIN
DD
*
REFORMAT FIELDS=(F1:20,8,1,15,F2:26,10,F1:16,2)
OPTION COPY
OUTFIL REMOVECC,
HEADER2=(1:Color,11:Flower,26:Status,36:Per Pot,/,
1:7-,11:14-,26:7-,36:7-),
BUILD=(1:1,8,11:9,15,26:24,10,
36:34,2,ZD,M10,LENGTH=7)
/*
This example illustrates how you can join paired records from two files using
multiple keys. In this case, neither file has duplicates. The paired records are the
records in F1 and F2 with matching keys (for example, key1=Roses and key2=Red).
Input file1 (F1) has RECFM=FB and LRECL=80. It contains the records shown for
SORTJNF1 in the JCL shown previously in this section. Input file2 (F2) has
RECFM=FB and LRECL=80. It contains the records shown for SORTJNF2 in the
JCL shown previously in this section.
The output file will have RECFM=FB and LRECL=42. It will contain the paired
records from the two files reformatted as follows:
Color
------Lilac
Orange
Yellow
Pink
Red
Yellow
Flower
-------------Daisies
Daisies
Daisies
Roses
Roses
Roses
Status
------SoldOut
SoldOut
InStock
SoldOut
InStock
Ordered
Per Pot
------12
6
2
4
3
6
The first JOINKEYS statement defines the ddname and keys for the F1 file.
FILE=F1 tells DFSORT that the ddname for the F1 file is SORTJNF1.
473

FIELDS=(1,15,A,20,8,A) tells DFSORT that the first binary key is in positions 1-15
ascending and the second binary key is in positions 20-27 ascending. Since
SORTED is not specified, DFSORT will sort the SORTJNF1 records by the specified
binary keys.
The second JOINKEYS statement defines the ddname and keys for the F2 file.
FIELDS=(10,15,A,1,8,A) tells DFSORT that the first binary key is in positions 10-24
ascending and the second binary key is in positions 1-8 ascending. Since SORTED
is not specified, DFSORT will sort the SORTJNF2 records by the specified binary
keys.
Note that corresponding keys for the two files match in length and order.
The REFORMAT statement defines the fields to be extracted for the joined records
in the order in which they are to appear. FIELDS=(F1:20,8,1,15,F2:26,10,F1:16,2) tells
DFSORT to create the joined records as follows:
Joined Record Positions
----------------------1-8
9-23
24-33
34-35
Extracted from
-----------------F1 positions 20-27
F1 positions 1-15
F2 positions 26-35
F1 positions 16-17
Since there is no JOIN statement, only paired records are joined by default.
The OPTION COPY statement tells DFSORT to copy the joined records. The
OUTFIL statement tells DFSORT to reformat the joined records, display a header at
the top of each page and remove the carriage control characters. Note that the
BUILD operand of the OUTFIL statement must reference the positions of fields in
the joined records.
Conceptually, JOINKEYS application processing proceeds as follows:
v Subtask1 sorts the SORTJNF1 (F1 file) records as directed by its JOINKEYS
statement. As a result, it passes the following records to the main task:
Daisies
Daisies
Daisies
Roses
Roses
Roses
Roses
12
06
02
09
04
03
06
Lilac
Orange
Yellow
Blue
Pink
Red
Yellow
v Subtask2 sorts the SORTJNF2 (F2 file) records as directed by its JOINKEYS
Lilac
Orange
White
Yellow
Red
Pink
Red
Yellow
Daisies
Daisies
Daisies
Daisies
Lilies
Roses
Roses
Roses
SoldOut
SoldOut
InStock
InStock
InStock
SoldOut
InStock
Ordered
v The main task joins the records passed from subtask1 and subtask2 as directed
by the specified JOINKEYS and REFORMAT statements, resulting in the
following joined records:
474

Lilac
Daisies
Orange Daisies
Yellow Daisies
Pink
Roses
Red
Roses
Yellow Roses
SoldOut
SoldOut
InStock
SoldOut
InStock
Ordered
12
06
02
04
03
06
v Finally, the main task copies and reformats the joined records according to the
OUTFIL statement, and writes the resulting records to SORTOUT. Thus,
SORTOUT contains these records:
Color
------Lilac
Orange
Yellow
Pink
Red
Yellow
Flower
-------------Daisies
Daisies
Daisies
Roses
Roses
Roses
Status
------SoldOut
SoldOut
InStock
SoldOut
InStock
Ordered
Per Pot
------12
6
2
4
3
6
Example 2 - Paired F1/F2 records with duplicates (cartesian)

//SYSOUT
DD SYSOUT=*
//VBIN DD DSN=MY.VBFILE,DISP=SHR
//FBIN DD DSN=MY.FBFILE,DISP=SHR
//SORTOUT DD DSN=MY.FB.OUTPUT,DISP=(NEW,CATLG,DELETE),
//SYSIN
DD
*
JOINKEYS F1=VBIN,FIELDS=(18,16,A),SORTED
JOINKEYS F2=FBIN,FIELDS=(1,16,A)
REFORMAT FIELDS=(F2:22,12,F1:5,12,F2:1,16)
OPTION EQUALS
/*
This example illustrates how you can join paired records from two files, both of
which have duplicate records. The result will be a cartesian join. The paired
records are the records in F1 and F2 with matching keys (for example, key=Cats).
Input file1 has RECFM=VB and LRECL=50. It contains the following records:
Len|Data
40|Eliot
40|Lloyd-Webber
48|Hart
48|Rodgers
47|Hammerstein
47|Rodgers
Cats
Cats
Pal Joey
Pal Joey
South Pacific
South Pacific
Musical
Musical
Musical,
Musical,
Musical,
Musical,
Comedy
Comedy
Drama
Drama
The output file will have RECFM=FB and LRECL=40. It will contain the paired
cartesian product of the two files sorted as follows:
Start:
End:
Start:
End:
Start:
End:
Start:
End:
Start:
End:
Start:
End:
1982
2000
1949
1954
1940
1941
1982
2000
1940
1941
1949
1954
Eliot
Cats
Eliot
Cats
Hammerstein South Pacific
Hart
Pal Joey
Hart
Pal Joey
Lloyd-WebberCats
Lloyd-WebberCats
Rodgers
Pal Joey
Rodgers
Pal Joey
Rodgers
South Pacific
Rodgers
South Pacific
475

The first JOINKEYS statement defines the ddname and key for the F1 file.
F1=VBIN tells DFSORT that the ddname for the F1 file is VBIN. FIELDS=(18,16,A)
tells DFSORT that the key is in positions 18-33 ascending. Note that since VBIN is
a VB file, the starting position of its key must take the RDW in positions 1-4 into
account. Since SORTED is specified, indicating that the records are already in order
by the specified binary key, DFSORT will copy the VBIN records.
The second JOINKEYS statement defines the ddname and binary key for the F2
file. F2=FBIN tells DFSORT that the ddname for the F2 file is FBIN.
FIELDS=(1,16,A) tells DFSORT that the binary key is in positions 1-16 ascending.
Since SORTED is not specified, DFSORT will sort the FBIN records by the specified
binary key.
in the order in which they are to appear. FIELDS=(F2:22,12,F1:5,12,F2:1,16) tells
DFSORT to create the joined records as follows:
----------------------1-12
13-24
25-40
Extracted from
-----------------F2 positions 22-33
F1 positions 5-16
F2 positions 1-16
Note that since VBIN (F1) is a VB file, the starting position of its REFORMAT field
must take the RDW in positions 1-4 into account.
Since there is no JOIN statement, only paired records are joined by default. Since
there are duplicates in each input file, a cartesian join is performed.
The SORT FIELDS=(13,12,CH,A) statement tells DFSORT to sort the joined records
by a different key than the one used for the join of F1 and F2 records. Note that
the FIELDS operand of the SORT statement must reference the positions of fields
in the joined records.
v Subtask1 copies the VBIN (F1 file) records as directed by its JOINKEYS
Len|Data
40|Eliot
40|Lloyd-Webber
48|Hart
48|Rodgers
47|Hammerstein
47|Rodgers
Cats
Cats
Pal Joey
Pal Joey
South Pacific
South Pacific
Musical
Musical
Musical,
Musical,
Musical,
Musical,
Comedy
Comedy
Drama
Drama
v Subtask2 sorts the FBIN (F2 file) records as directed by its JOINKEYS statement.
As a result, it passes the following records to the main task:
Cats
Cats
Pal Joey
Pal Joey
South Pacific
South Pacific
Start:
End:
Start:
End:
Start:
End:
1982 50
2000
1940 22
1941
1949 13
1954
Start:
End:
Start:
End:
476
1982
2000
1982
2000
Eliot
Cats
Eliot
Cats
Lloyd-WebberCats
Lloyd-WebberCats

Start:
End:
Start:
End:
Start:
End:
Start:
End:
1940
1941
1940
1941
1949
1954
1949
1954
Hart
Hart
Rodgers
Rodgers
Hammerstein
Hammerstein
Rodgers
Rodgers
Pal Joey
Pal Joey
Pal Joey
Pal Joey
South Pacific
South Pacific
South Pacific
South Pacific
v Finally, the main task sorts the joined records according to the SORT statement,
and writes the resulting records to SORTOUT. Thus, SORTOUT contains these
records:
Start:
End:
Start:
End:
Start:
End:
Start:
End:
Start:
End:
Start:
End:
1982
2000
1949
1954
1940
1941
1982
2000
1940
1941
1949
1954
Eliot
Cats
Eliot
Cats
Hart
Pal Joey
Hart
Pal Joey
Lloyd-WebberCats
Lloyd-WebberCats
Rodgers
Pal Joey
Rodgers
Pal Joey
Rodgers
South Pacific
Rodgers
South Pacific
Example 3 - Paired F1 records

//SYSOUT
DD SYSOUT=*
//MASTER DD DSN=MASTER.FILE,DISP=SHR
//PULL DD DSN=PULL.FILE,DISP=SHR
//JNF1CNTL DD *
* Control statement for subtask1 (F1)
%=(ENDBEFR=C,),
%02=(ENDBEFR=C,,FIXLEN=10)),
OVERLAY=(36:%01,52:%02)
/*
//DFSPARM
DD
*
JOINKEYS F1=MASTER,FIELDS=(36,15,A,52,10,A)
JOINKEYS F2=PULL,FIELDS=(12,15,A,1,10,A)
REFORMAT FIELDS=(F1:1,35)
* Control statement for main task
OPTION COPY
/*
This example illustrates how you can select only paired records from one of two
files. In this case, we will select the F1 records that have a match in F2 on the
specified keys (for example, key1=Development and key2=Master). The F1 records
are in comma delimited form so we will parse them to get the binary keys we
need.
Input file1 (F1) has RECFM=FB and LRECL=35. It contains the following records:
Marketing,William,Master
Development,John,Bachelor
Manufacturing,Louis,Master
Development,Carol,Master
Research,Angela,Master
Research,Anne,Doctorate
Development,Sara,Doctorate
Marketing,Joseph,Master
477

Manufacturing,Donna,Bachelor
Development,Susan,Master
Manufacturing,Nick,Master
Research,Howard,Doctorate
Input file2 (F2) has RECFM=FB and LRECL=30. It contains the following records:
Master
Development
Master
Manufacturing
Bachelor
Development
Doctorate Research
The output file will have RECFM=FB and LRECL=35. It will contain the paired F1
records, that is, the records from F1 that have a match in F2 for the specified keys
(the first and third fields):
The first JOINKEYS statement defines the ddname and keys for the F1 file.
F1=MASTER tells DFSORT that the ddname for the F1 file is MASTER.
The control statements in JNF1CNTL will be used to process the F1 file before it is
sorted. The input records are in comma delimited form, so to use the first and
third fields as binary keys, while preserving the original data, we use an INREC
statement to parse out the fields we want and add them to the end of the record.
The parsed first key will be in positions 36-50 and the parsed second key will be in
positions 52-61. For example, the first reformatted record will look like this:
Marketing
Master
FIELDS=(36,15,A,52,10,A) in the JOINKEYS statement (referring to the reformatted

INREC positions) tells DFSORT that the first key is in positions 36-50 ascending
and the second key is in positions 52-61 ascending.
F2=PULL tells DFSORT that the ddname for the F2 file is PULL.
FIELDS=(12,15,A,1,10,A) in the JOINKEYS statement tells DFSORT that the first
key is in positions 12-26 ascending and the second key is in positions 1-10
ascending.
in the order in which they are to appear. FIELDS=(F1:1,35) tells DFSORT to create
the joined records from the original F1 input records (positions 1-35 of the F1
records). Since we only needed the added parsed fields for sorting, we don't need
to include them in the joined records. Of course, if we wanted the main task to
"see" the parsed fields in the joined records, we could include the parsed fields in
the REFORMAT FIELDS operand.
Since there is no JOIN statement, only paired records are joined by default.
The OPTION COPY statement tells DFSORT to copy the joined records.
478

v Subtask1 performs INREC processing for the MASTER (F1 file) records as
directed by the control statement in JNF1CNTL and sorts the resulting records as
directed by its JOINKEYS statement. As a result, it passes the following records
to the main task:
Development,Sara,Doctorate
Manufacturing,Donna,Bachelor
Marketing,Joseph,Master
Research,Angela,Master
Development
Development
Development
Development
Manufacturing
Manufacturing
Manufacturing
Marketing
Marketing
Research
Research
Research
Bachelor
Doctorate
Master
Master
Bachelor
Master
Master
Master
Master
Doctorate
Doctorate
Master
v Subtask2 sorts the PULL (F2 file) records as directed by its JOINKEYS statement.
As a result, it passes the following records to the main task:
Bachelor
Development
Master
Development
Master
Manufacturing
Doctorate Research
v Finally, the main task copies the joined records to SORTOUT. Thus, SORTOUT
Example 4 - Unpaired F2 records

//SYSOUT
DD SYSOUT=*
//IN1 DD DSN=FILE1.IN,DISP=SHR
//IN2 DD DSN=FILE2.IN,DISP=SHR
//SORTOUT DD DSN=FILE3.OUT,DISP=OLD
//JNF1CNTL DD *
OMIT COND=(10,5,UFF,EQ,99999)
INREC BUILD=(1,8,9:10,5,UFF,TO=ZD,LENGTH=5)
/*
//JNF2CNTL DD *
OMIT COND=(14,5,UFF,EQ,99999)
INREC BUILD=(1,4,5:14,5,UFF,TO=ZD,LENGTH=5,10:5)
/*
//SYSIN
DD
*
479

REFORMAT FIELDS=(F2:1,4,10)
* Control statement for main task
OPTION COPY
/*
This example illustrates how you can select only unpaired records from one of two
files. In this case, we will select the F2 records that do not have a match in F1 on
the specified keys (for example, key1=Molly and key2=2100). We will also omit
certain records from each input file and handle unnormalized keys.
Input file1 has RECFM=FB and LRECL=15. It contains the following records:
Molly
145
Molly
99999
Molly
2143
Jasmine 1292
Jasmine 5
Jasmine 28
Jasmine 99999
Input file2 has RECFM=VB and LRECL=35. It contains the following records:
Len|Data
30|Molly
145
31|Molly
2100
28|Molly
18
28|Jasmine 99999
28|Jasmine 5
30|Jasmine 28
29|Jasmine 103
31|Jasmine 99999
Thursday
Wednesday
Sunday
Monday
Sunday
Saturday
Tuesday
Wednesday
The output file will have RECFM=VB and LRECL=35. F1 records with 99999 in
positions 10-14 and F2 records with 99999 in positions 14-18 will be removed
before join processing. The output file will contain unpaired F2 records (that is,
records from F2 that do not have a match in F1 for the specified keys) as follows:
Len|Data
29|Jasmine 103
31|Molly
2100
28|Molly
18
Tuesday
Wednesday
Sunday
The first JOINKEYS statement defines the ddname and keys for the F1 file. F1=IN1
tells DFSORT that the ddname for the F1 file is IN1.
sorted. The OMIT statement removes records that have 99999 in positions 10-14.
We want to use the numeric field as our key. However, the numeric field is
unnormalized since it is left aligned instead of right aligned, so sorting it as a
binary key will not work. To handle this, we use the INREC statement to reformat
the numeric field as ZD values in positions 9-13 (positive ZD values with the same
sign can be sorted as binary). For example, the first reformatted FB record will look
like this:
Molly
00145
Since we don't need the F1 records for output, we don't need to keep the original
left aligned numeric value.
480

FIELDS=(1,8,A,9,5,D) in the JOINKEYS statement (referring to the reformatted
INREC positions) tells DFSORT that the first key is in positions 1-8 ascending and
the second key is in positions 9-13 descending.
F2=IN2 tells DFSORT that the ddname for the F2 file is IN2.
sorted. The OMIT statement removes records that have 99999 in positions 14-18.
Again, we need a ZD version of the left aligned numeric value to use for the
binary key. But in this case, since we want the original F2 records for output, we
need to keep the original numeric value as well. Using the INREC statement, we
add the ZD value at positions 5-9 between the RDW and the first data field. That
shifts the original data to start at position 10. For example, the first reformatted VB
record will look like this:
Len|Data
35|00145Molly
145
Thursday
In this case, since the input is a VB file, we specify the RDW (1,4), then the
converted field, and then the rest of the record (5 without a length) in the INREC
statement.
FIELDS=(10,8,A,5,5,D) in the JOINKEYS statement (referring to the reformatted
INREC positions) tells DFSORT that the first key is in positions 10-17 ascending
and the second key is in positions 5-9 descending.
Note that since IN2 is a VB file, all of its starting positions must take the RDW in
positions 1-4 into account.
The JOIN statement tells DFSORT that the joined records should be the F2 records
that do not have a match in F1 on the specified keys.
in the order in which they are to appear. We need the RDW (1,4) and the original
data which starts in position 10 of the reformatted F2 records. So we use
FIELDS=(F2:1,4,10). Since the last field (10) is a position without a length, it tells
DFSORT to create VB records. The joined records are created as follows from the
reformatted F2 records:
----------------------1-4
5 to end
Extracted from
--------------------------------RDW (not extracted)
Reformatted F2 position 10 to end

v Subtask1 performs OMIT and INREC processing for the IN1 (F1 file) records as
directed by the control statements in JNF1CNTL and sorts the resulting records
as directed by its JOINKEYS statement. As a result, it passes the following
records to the main task:
Jasmine
Jasmine
Jasmine
Molly
Molly
01292
00028
00005
02143
00145
481

v Subtask2 performs OMIT and INREC processing for the IN2 (F2 file) records as
directed by the control statements in JNF2CNTL and sorts the resulting records
as directed by its JOINKEYS statement. As a result, it passes the following
records to the main task:
Len|Data
34|00103Jasmine 103
35|00028Jasmine 28
33|00005Jasmine 5
36|02100Molly
2100
35|00145Molly
145
33|00018Molly
18
Tuesday
Saturday
Sunday
Wednesday
Thursday
Sunday
by the specified JOINKEYS, JOIN and REFORMAT statements, resulting in the
following joined records (unmatched F2 records):
Len|Data
29|Jasmine 103
31|Molly
2100
28|Molly
18
Tuesday
Wednesday
Sunday
v Finally, the main task copies the joined records, and writes them to SORTOUT.
Thus, SORTOUT contains these records:
Len|Data
29|Jasmine 103
31|Molly
2100
28|Molly
18
Tuesday
Wednesday
Sunday
Example 5 - Paired and unpaired F1/F2 records (indicator

method)
//SYSOUT
DD SYSOUT=*
//SORTJNF1 DD DSN=FIRST.FILE,DISP=SHR
//SORTJNF2 DD DSN=SECOND.FILE,DISP=SHR
//F1ONLY DD SYSOUT=*
//BOTH DD SYSOUT=*
//SYSIN
DD
*
JOIN UNPAIRED,F1,F2
REFORMAT FIELDS=(F1:1,14,F2:1,20,?)
OPTION COPY
OUTFIL FNAMES=F1ONLY,INCLUDE=(35,1,CH,EQ,C1),
BUILD=(1,14)
OUTFIL FNAMES=F2ONLY,INCLUDE=(35,1,CH,EQ,C2),
BUILD=(15,20)
OUTFIL FNAMES=BOTH,INCLUDE=(35,1,CH,EQ,CB),
BUILD=(1,14,/,15,20)
/*
This example illustrates how you can create three output files; with paired F1/F2
records, unpaired F1 records and unpaired F2 records. In this case:
v F1 records which do not have a match in F2 on the specified keys (for example,
key=David) will be written to the F1ONLY output file.
v F2 records which do not have a match in F1 on the specified keys (for example,
key=Karen) will be written to the F2ONLY output file.
v F1 and F2 records which have a match in F1 and F2 on the specified keys (for
example, key=Carrie) will be written to the BOTH output file.
482

Carrie
David
Frank
Holly
Vicky
F101
F102
F103
F104
F105
No
Yes
Yes
No
Yes
Carrie
F201
Holly
F202
Karen
F203
Sri Hari F204
Vicky
F205
The F1ONLY file will have RECFM=FB and LRECL=14. It will contain the
unpaired F1 records as follows:
David
Frank
F102
F103
The F2ONLY file will have RECFM=FB and LRECL=20. It will contain the
unpaired F2 records as follows:
Yes
No
Karen
Sri Hari
F203
F204
The BOTH file will have RECFM=FB and LRECL=20. It will contain the paired F1
and F2 records as follows:
Carrie
F101
No
Carrie
F201
Holly
F104
Yes
Holly
F202
Vicky
F105
Yes
Vicky
F205
The first JOINKEYS statement defines the ddname and key for the F1 file. FILE=F1
tells DFSORT that the ddname for the F1 file is SORTJNF1. FIELDS=(1,10,A) tells
DFSORT that the key is in positions 1-10 ascending. Since SORTED is specified,
indicating that the records are already in order by the specified binary key,
DFSORT will copy the SORTJNF1 records. Since NOSEQCK is specified, DFSORT
will not check that the records are in order by the key. (Only use NOSEQCK if you
know for sure that the records are in order by the key.)
The second JOINKEYS statement defines the ddname and key for the F2 file.
FIELDS=(7,10,A) tells DFSORT that the key is in positions 7-16 ascending. Since
SORTED is specified, indicating that the records are already in order by the
specified binary key, DFSORT will copy the SORTJNF2 records. Since NOSEQCK is
specified, DFSORT will not check that the records are in order by the key. (Only
use NOSEQCK if you know for sure that the records are in order by the key.)
The JOIN statement tells DFSORT that the joined records should include the
unpaired F1 and F2 records as well as the paired F1/F2 records.
in the order in which they are to appear, and includes an indicator in the last
position that will be set to '1' if the key is found only in the F1 file, '2' if the key is
found only in the F2 file, or 'B' if the key is found in the F1 file and in the F2 file.
FIELDS=(F1:1,14,F2:1,20,?) tells DFSORT to create the joined records as follows:
483

----------------------1-14
15-34
35
Extracted from
----------------F1 positions 1-14
F2 positions 1-20
Indicator of where key was found
OUTFIL statements use the indicator in position 35 to determine where to find the
F1 or F2 fields in the joined records and where to write the fields (F1ONLY,
F2ONLY or BOTH).
v Subtask1 copies the SORTJNF1 (F1) records as directed by the JOINKEYS
statement. As a result, it copies the unchanged SORTJNF1 records to the main
task.
task.
following joined records (paired and unpaired):
Carrie
David
Frank
Holly
Vicky
F101No
F102
F103
F104Yes
Yes
No
F105Yes
Carrie
Holly
Karen
Sri Hari
Vicky
F201B
1
1
F202B
F2032
F2042
F205B
For F1 records without a match in F2 (for example, the F102 record), the
indicator in position 35 has a '1'. For F2 records without a match in F1 (for
example, the F203 record), the indicator in position 35 has a '2'. For F1 records
with a match in F2 (for example, the F101 and F201 records), the indicator in
position 35 has a 'B'.
v The first OUTFIL statement finds records with a '1' in position 35. These are the
F1 records without a match in F2. The F1 field is in positions 1-14 of the joined
record, so those positions are written to the F1ONLY file. Thus, F1ONLY
David
Frank
F102
F103
v The second OUTFIL statement finds records with a '2' in position 35. These are
the F2 records without a match in F1. The F2 field is in positions 15-34 of the
joined record, so those positions are written to the F2ONLY file. Thus, F2ONLY
Yes
No
Karen
Sri Hari
F203
F204
v The third OUTFIL statement finds records with 'B' in position 35. These are the
F1 and F2 records with a match. The F1 field is in positions 1-14 of the joined
record and the F2 field is in positions 15-34 of the joined record, so each joined
record is split into those two records and written to the BOTH file. The shorter
F1 record is padded with blanks on the right to the length of the F2 record.
Thus, BOTH contains these records:
484

Carrie
F101
No
Carrie
Holly
F104
Yes
Holly
Vicky
F105
Yes
Vicky
F201
F202
F205
Note: If you only want one record per key in BOTH, you can have the BUILD for
FNAMES=BOTH specify the positions for just that record. For example,
BUILD=(1,14) for the F1 records or BUILD=(15,20) for the F2 records.
Example 6 - Paired and unpaired F1/F2 records (FILL method)

//SYSOUT
DD SYSOUT=*
//SORTJNF1 DD DSN=FIRST.FILE,DISP=SHR
//SORTJNF2 DD DSN=SECOND.FILE,DISP=SHR
//BOTH DD SYSOUT=*
//SYSIN
DD
*
JOIN UNPAIRED,F1,F2
REFORMAT FIELDS=(F1:1,14,F2:1,20),FILL=C$
OPTION COPY
OUTFIL FNAMES=F1ONLY,INCLUDE=(15,1,CH,EQ,C$),
BUILD=(1,14)
OUTFIL FNAMES=F2ONLY,INCLUDE=(1,1,CH,EQ,C$),
BUILD=(15,20)
OUTFIL FNAMES=BOTH,INCLUDE=(15,1,CH,NE,C$,AND,1,1,CH,NE,C$),
BUILD=(1,14,/,15,20)
/*
This example illustrates an alternate way to create three output files; with paired
F1/F2 records, unpaired F1 records and unpaired F2 records. It produces the same
output as Example 5. Whereas Example 5 uses an indicator in one position to
determine where the key was found, Example 6 uses a fill character in different
positions to determine where the key was found. Another drawback of the
Example 6 method is that you must use a FILL character that does not appear in
either input record. The Explanation for Example 6 is the same as for Example 5
up to the REFORMAT statement and then it differs as follows:
in the order in which they are to appear. FIELDS=(F1:1,14,F2:1,20) tells DFSORT to
create the joined records as follows:
----------------------1-14
15-34
Extracted from
----------------F1 positions 1-14
F2 positions 1-20
FILL=C'$' tells DFSORT to use $ as the fill character for the F1 field in an unpaired
F2 record and for the F2 field in an unpaired F1 record. We use the FILL character
to identify the unpaired records from each file; when used for this purpose, the
default of FILL=X'40' is usually not a good choice. You must select a FILL character
that will not appear in either input file.
OUTFIL statements use the presence or absence of the $ fill character in certain
485

positions to determine where to find the F1 or F2 fields in the joined records and
where to write the fields (F1ONLY, F2ONLY or BOTH).
task.
task.
following joined records (paired and unpaired):
Carrie
F101No
Carrie
F201
David
F102$$$$$$$$$$$$$$$$$$$$
Frank
F103$$$$$$$$$$$$$$$$$$$$
Holly
F104Yes
Holly
F202
$$$$$$$$$$$$$$Yes
Karen
F203
$$$$$$$$$$$$$$No
Sri Hari F204
Vicky
F105Yes
Vicky
F205
For F1 records without a match in F2 (for example, the F102 record), the F2 field
is filled with the FILL character. For F2 records without a match in F1 (for
example, the F203 record), the F1 field is filled with the FILL character. For F1
records with a match in F2 (for example, the F101 and F201 records), no FILL
characters are used.
v The first OUTFIL statement finds records with the FILL character in position 15.
These are the F1 records without a match in F2. The F1 field is in positions 1-14
of the joined record, so those positions are written to the F1ONLY file. Thus,
F1ONLY contains these records:
David
Frank
F102
F103
v The second OUTFIL statement finds records with the FILL character in position
1. These are the F2 records without a match in F1. The F2 field is in positions
15-34 of the joined record, so those positions are written to the F2ONLY file.
Thus, F2ONLY contains these records:
Yes
No
Karen
Sri Hari
F203
F204
v The third OUTFIL statement finds records without the FILL character in position
1 or position 15. These are the F1 and F2 records with a match. The F1 field is in
positions 1-14 of the joined record and the F2 field is in positions 15-34 of the
joined record, so each joined record is split into those two records and written to
the BOTH file. The shorter F1 record is padded with blanks on the right to the
length of the F2 record. Thus, BOTH contains these records:
Carrie
F101
No
Carrie
F201
Holly
F104
Yes
Holly
F202
Vicky
F105
Yes
Vicky
F205
Note: If you only want one record per key in BOTH, you can have the BUILD for
FNAMES=BOTH specify the positions for just that record. For example,
BUILD=(1,14) for the F1 records or BUILD=(15,20) for the F2 records.
486
Chapter 5. Using your own user exit routines

User exit routine overview
DFSORT can pass program control to your own routines at points in the executable
code called user exits. Your user exit routines can perform a variety of functions
including deleting, inserting, altering, and summarizing records.
If you need to perform these tasks, you should be aware that DFSORT already
provides extensive facilities for working with your data in the various DFSORT
program control statements. See the discussions of the INCLUDE, OMIT, INREC,
OUTFIL, OUTREC, and SUM program control statements in Chapter 3, Using
DFSORT program control statements, on page 81. You might decide that using a
program control statement to work with your records is more appropriate to your
needs.
Although this chapter discusses only routines written in assembler or COBOL, you
can write your exit routines in any language that can:
v Pass and accept the address into general register 1 of a:
Record
Full word of zeros
Parameter list.
v Pass a return code in register 15.
You can easily activate user exit routines at run-time with the MODS program
control statement (see MODS control statement on page 166). Alternatively,
under certain circumstances you can also activate a user exit routine by passing the
address of your exit routine in the invocation parameter list. See Chapter 6,
Invoking DFSORT from a program, on page 541 for details.
Parameters that affect the way user exit routines are handled include:
v The MODS statement, explained in MODS control statement on page 166
v The E15=COB and E35=COB PARM options of the EXEC statement, explained in
Specifying EXEC/DFSPARM PARM options on page 32
v The COBEXIT option of the OPTION statement, explained in OPTION control
statement on page 173
v The EXITCK option of the OPTION statement, explained in OPTION control
Note: To avoid ambiguity in this chapter, it is assumed that the IBM default,
EXITCK=STRONG, was selected at your site.
Certain user exit routines can be written in COBOL, using a special interface. If
you write your exit routines in PL/I, you must use the PL/I subroutine facilities.
You might need to reserve space to be used by your exits. See Use main storage
efficiently on page 807 for more information about storage.
487
DFSORT Program Phases
DFSORT program phases

A DFSORT program phase is a large DFSORT component designed to perform a
specific task such as writing the output file. Various user exits are contained in the
input and output phases and are activated at a particular time during DFSORT
processing. The input phase is used only for a sort or copy. When the output phase
is completed, DFSORT returns control to the operating system or invoking
program. Figure 10 on page 489 is a representation of DFSORT input/output logic.
488
DFSORT Program Phases

INPUT/EXIT/OUTPUT LOGIC
FOR SORT APPLICATION
FOR COPY APPLICATION
START
START
START
A
INPUT
PHASE
FOR MERGE APPLICATION
A
INPUT/
OUTPUT
PHASE
READ RECORD
FROM SORTIN
OUPUT READ RECORD

PHASE
FROM EACH
SORTINnn
READ RECORD
FROM SORTIN
A
E15
E15
CHOOSE A
RECORD
MORE INPUT
RECORDS?
YES
A
E35
E35
NO
E15
WRITE
RECORD
TO SORTOUT
AND/OR
OUTFIL
DATA SETS
RETURNS
RC=8
SORT THE
RECORDS
MORE INPUT
RECORDS?
B
OUTPUT
PHASE
NO
E15
GET SORTED
RECORD
WRITE
RECORD
TO SORTOUT
AND/OR
OUTFIL
DATA SETS
YES
MORE INPUT
RECORDS?
NO
YES
READ
RECORD
FROM
SORTINnn
E35
A
RETURNS
RC=8
E35
WRITE
RECORD
TO SORTOUT
AND/OR
OUTFIL
DATA SETS
MORE SORTED YES
RECORDS?
E35
RETURNS
RC=8
EXIT
RETURNS
RC=8
EXIT
NO
E35
RETURNS
RC=8
EXIT
Figure 10. Examples of DFSORT Input/User Exit/Output Logic
489
Functions of Routines at User Exits
Functions of routines at user exits

You can use exit routines to accomplish a variety of tasks:
v Open and initialize data sets
v Modify control fields
v Insert, delete, or alter records
v
v
v
v
v
Sum records
Handle special I/O conditions
Determine action when intermediate storage is insufficient
Close data sets
Terminate DFSORT.
Figure 10 on page 489, Table 62 on page 491, and Table 63 on page 491 summarize
the functions of user exit routines and the exits and phases with which they can be
associated.
The following restrictions apply to the applications for which specific user exits
can be used:
v E11. E16, E17, E18 and E19 can be used for sort applications, but are ignored for
copy or merge applications.
v E15 can be used for sort or copy applications, but is ignored for merge
applications.
v E32 can be used for merge applications, but is ignored for copy or sort
applications.
v E61 can be used for sort or merge applications, but is ignored for copy
applications.
v E31, E35, E37, E38 and E39 can be used for sort, copy or merge applications.
DFSORT input/user exit/output logic examples

Figure 10 on page 489 gives examples of the logic flow for sort, copy, and merge
applications as it relates to SORTINnn, E15 or E35 user exits, and SORTOUT. The
intent is to show how your E15 and E35 user exits fit into the logic of an
application. All possible paths are not covered. For simplicity, it is assumed that all
of the applicable data sets and exits are present and that records are not inserted or
deleted. (For a merge, similar logic would be used if an E32 user exit supplied the
records rather than SORTINnn data sets.)
Figure 10 on page 489 illustrates the following logic:
v E15 and E35 user exits continue to be entered until they pass back a return code
of 8. If your user exit passes a return code of 8 to DFSORT when input records
still remain to be processed, the records are processed by DFSORT without being
passed to your exit.
v During a sort, each record is read from SORTIN and passed to E15 user exit.
When all of the records have been processed in this manner, they are sorted by
DFSORT, then each sorted record is passed to E35 and written to the output data
sets.
v During a copy, each record is read from SORTIN, passed to E15 and E35 user
exits, and written to the output data sets.
v During a merge, one record is initially read from each SORTINnn data set. The
record to be output is chosen, passed to E35, and written to the output data sets.
490

The chosen record is then replaced by reading a record from the same
SORTINnn data set and the process continues.
Note: For a merge application, records deleted during an E35 user exit routine
are not sequence-checked. If you use an E35 user exit routine without an output
data set, sequence checking is not performed at the time the records are passed
to the E35 user exit; therefore, you must ensure that input records are in correct
sequence.
Table 62. Functions of Routines at Program User Exits (Sort)
Functions
Sort Input Phase
Sort Output Phase
Open/Initialize
E11, E15 user exits
E31 user exit
Modify control fields
E61 user exit
N/A
Insert, Delete/Alter
E15 user exit
E35 user exit

E35 user exit1
Sum records
Handle special I/O conditions:
QSAM/BSAM and VSAM SORTIN
QSAM/BSAM SORTOUT
VSAM SORTOUT
E18 user exit

E19 user exit2
N/A
E38 user exit2

E39 user exit3
E39 user exit3
Determine action when intermediate storage E16 user exit4

is insufficient
N/A
Close/housekeeping
E15, E17 user exits
E35, E37 user exits
Terminate DFSORT
E15 user exit
E35 user exit
Note:
1. The SUM control statement can be used instead of your own routine to sum records.
2. Applies only to a tape work data set sort.
3. E39 can be used for SORTOUT, but not for OUTFIL data sets.
4. Applies only to a tape work data set sort or a Peerage/Vale sort without work data sets.
Table 63. Functions of Routines at Program User Exits (Copy and Merge)
Functions
Copy
Merge
Open/Initialize
E15, E31 user exits
E31 user exit
Modify control fields
N/A
E61 user exit
Insert
E15, E35 user exits
E32, E35 user exits
Delete/alter
E15, E35 user exits
E35 user exit
Sum records
E35 user exit
E35 user exit1
E38 user exit

E39 user exit
E38 user exit

E39 user exit
Close/housekeeping
E35, E37 user exits
E35, E37 user exits
Terminate DFSORT
E15, E35 user exits
E32, E35 user exits
Handle special I/O conditions:

QSAM/BSAM and VSAM SORTIN(nn)
QSAM/BSAM and VSAM SORTOUT
Note:
1. The SUM control statement can be used instead of your own routine to sum records.
491
Opening and initializing data sets

You can write your own routines to open data sets and perform other forms of
initialization; you must associate these routines with the E11, E15, E31 and E35
user exits.
To check labels on input files, use the E18 and E38 user exits.
Modifying control fields

You can write a routine to alter control fields before DFSORT compares them. This
allows you, for example, to normalize floating-point control fields. It also allows
you to modify the order in which the records are finally sorted or merged, a
function for which you would usually use DFSORT's ALTSEQ program control
statement. You must associate this routine with the E61 user exit.
When an E61 user exit is used, the subsequent comparisons always arrange the
modified control fields in ascending order.
Note: Although you are altering control fields before a compare, your original
records are not altered.
Inserting, deleting, and altering records

You can write your own routines to delete, insert, or alter records. You must
associate these routines with the E15, E32, and E35 user exits.
Note: DFSORT also provides INCLUDE and OMIT statements, and OUTFIL
INCLUDE and OMIT parameters that automatically include or delete records
based on your field criteria. For more information on these control statements,
refer to Chapter 3, Using DFSORT program control statements, on page 81.
Summing records
You can sum records for output by using the E35 user exit. However, you can also
use DFSORT's SUM program control statement to accomplish this without a user
exit. See SUM control statement on page 451.
Handling special I/O

DFSORT contains four exits to handle special I/O conditions: E18 and E38 user
exits for SORTIN and SORTINnn, and E19 and E39 user exits for SORTOUT (but
not for OUTFIL data sets). They are particularly useful for a tape work data set
sort. With all disk work data set sorts, E19 and E38 user exits are ignored.
You can use these exits to incorporate your own or your site's I/O error recovery
routines into DFSORT. Your read and write error routines must reside in a
partitioned data set (library). Your library routines are brought into main storage
with their associated phases. When DFSORT encounters an uncorrectable I/O
error, it passes the same parameters as those passed by QSAM/BSAM or VSAM. If
no user routines are supplied and an uncorrectable read or write error is
encountered, DFSORT issues an error message and then terminates.
With QSAM/BSAM, the following information is passed to your synchronous error
routine:
v General registers 0 and 1 are unchanged; they contain the information passed by
QSAM/BSAM, as documented in the data management publications.
v General register 14 contains the return address of DFSORT.
492

v General register 15 contains the address of your error routine.
VSAM will go directly to any routine specified in the EXLST macro you passed to
DFSORT via the E18, E38, or E39 user exit, as appropriate. Your routine must
return to VSAM via register 14. For details, see z/OS DFSMS Macro Instructions for
Data Sets or z/OS DFSMS Using Data Sets
Routines for read errors

You must associate these routines with the E18 and E38 user exits. They must pass
certain control block information back to DFSORT to tell it whether to accept the
record as it is, skip the block, or request termination. They can also attempt to
correct the error.
Routines for write errors

You must associate these routines with the E19 and E39 user exits. These routines
can perform any necessary abnormal end-of-task operations for SORTOUT before
DFSORT is terminated.
VSAM user exit functions

There are three user exits that can be used with VSAM SORTIN, SORTINnn, and
SORTOUT data sets (but not with OUTFIL data sets), to supply passwords or a
user exit list to journal a VSAM data set. They can carry out other VSAM exit
functions except EODAD. The user exits are E18 for sort SORTIN, E38 for merge
SORTINnn or copy SORTIN, and E39 for SORTOUT.
Determining action when intermediate storage is insufficient

You can write a routine to direct DFSORT program action if DFSORT determines
that insufficient intermediate storage is available to handle the input data set. You
must associate this routine with the E16 user exit for sorts using tape work data
sets. For a sort that uses tape data sets, you can choose between sorting current
records only, trying to complete the sort, or terminating DFSORT. For more details,
see Exceeding tape work space capacity on page 861.
Closing data sets

You can write your own routines to close data sets and perform any necessary
housekeeping; you must associate these routines with the E15, E17, E35, and E37
user exits. To write SORTOUT labels, use the E19 and E39 user exits. If you have
an end-of-file routine you want to use for SORTIN, include it at the E18 user exit.
Terminating DFSORT
You can write an exit routine to terminate DFSORT before all records have been
processed. You must associate these routines with the E15, E16, E32, and E35 user
exits.
32-bit and 64-bit parameter lists

For the E15, E32 and E35 exits, DFSORT supports both a 32-bit parameter list and
a 64-bit parameter list. For all of the other exits, DFSORT supports only a 32-bit
parameter list.
If you want DFSORT to use the 64-bit parameter list for your E15, E32 or E35, you
must call DFSORT with the 64-bit invocation list, and:
v use a MODS control statement for E15 or E35 with the N64 parameter or
493
32-Bit and 64-Bit Parameter Lists

v set on bit 4 (for E15 or E32) or bit 5 (for E35) of byte 9 in the 64-bit invocation
list
Otherwise, DFSORT will use the 32-bit parameter list for your E15, E32 or E35.
See Chapter 6, Invoking DFSORT from a program, on page 541 for details of
using the 64-bit invocation parameter list.
See the sections that follow on E15, E32 and E35 for complete details on using the
32-bit and 64-bit parameter lists.
64-bit address terminology

When referring to 64-bit addresses in this chapter:
v A clean 24-bit address refers to binary zeros in the high-order 5 bytes followed
by a 24-bit address in the low-order 3 bytes, that is, X0000000000aaaaaa
by a 31-bit address in the low-order 4 bytes, that is, X00000000aaaaaaaa where
Bit 32 of the 31-bit address must be 0.
v A 64 bit address uses all 64-bits for the address.
The following summarizes the previous information:
Table 64. Summary of 64-bit Terminology for Exits
Bits 0-31
Bit 32
Bits 33-39
Bits 40-63
24-bit
Address
31-bit
64-bit
Address
Address
Addressing and residence modes for user exits

To allow user exits called by Blockset or Peerage/Vale to reside above or below
16MB virtual, and use either 24-bit, 31-bit or 64-bit addressing, DFSORT supplies
these features:
v To ensure that DFSORT enters your user exit with the correct addressing mode,
you must observe these rules:
If the user exit name is specified in a MODS control statement, the user exit is
entered with the addressing mode indicated by the linkage editor attributes of
the routine (for example, 31-bit addressing in effect if AMODE 31 is
specified).
If the address of the exit is passed to DFSORT (preloaded exit) via the 24-bit
list, the user exit is entered with 24-bit addressing in effect.
If the address of the user exit is passed to DFSORT via the extended
parameter list (preloaded exit), the user exit is entered with 24-bit addressing
in effect if bit 0 of the user exit address in the list is 0 or with 31-bit
addressing in effect if bit 0 of the user exit address in the list is 1.
If the address of the user exit is passed to DFSORT via the 64-bit parameter
list (preloaded exit), the user exit is entered with the addressing mode in
effect as specified by the appropriate flag in byte 8 or 9 of the parameter list
as follows:
494
Addressing and Residence Modes for User Exits

Table 65. Addressing Mode Flags for Exits in 64-bit Parameter List
Byte
Bit
Enter E15 or E32 in AMODE 24
Enter E35 in AMODE 24
Meaning when on
v User exits can return to DFSORT with either 24-bit, 31-bit, or 64-bit addressing
in effect. The return address that DFSORT placed in register 14 must be used.
Note: For a conventional merge or tape work data set sort application, user
exits:
must reside below 16MB virtual
must use 24-bit addressing mode
must not use a user exit address constant.
How user exit routines affect DFSORT performance

Before writing a user exit routine, consider the following factors:
v Your routines occupy main storage that would otherwise be available to
DFSORT. Because its main storage is restricted, DFSORT might need to perform
extra passes to sort the data. This, of course, increases sorting time.
v User exit routines increase the overall run-time.
Attention: Several of the user exits give your routine control once for each
record until you pass a do not return return code to DFSORT. You must
remember this when designing your routines.
v Using INCLUDE, OMIT, INREC, OUTFIL, OUTREC, and SUM instead of user
exit routines allows DFSORT to perform more efficiently.
Summary of rules for user exit routines

When preparing your routines, remember that:
v User-written routines must follow standard linkage conventions and use the
required interfaces. COBOL E15 and E35 user exits must use the special interface
provided.
v To use an E32 user exit, your invoking program must pass its address to
DFSORT in the parameter list.
v To use any other user exit, you must associate your routine with the appropriate
user exits using the MODS control statement. See MODS control statement on
page 166.
v Your invoking program can alternatively pass the address of an E15, E18, E35,
and E39 user exit to DFSORT in the parameter list.
495
Summary of Rules for User Exit Routines

v When Blockset or Peerage/Vale is used and your user exits are reenterable, the
entire DFSORT program is reenterable.
v If you are using ASCII input, remember that data presented to your user exits at
user exits are in EBCDIC format. If the E61 user exit is used to resolve ASCII
collating for special alphabetic characters, substituted characters must be in
EBCDIC, but the sequencing result depends on the byte value of the ASCII
translation for the substituted character.
Loading user exit routines

You must assemble or compile each user exit as a separate program. If your user
exit operates independently, bind or link-edit it separately into a partitioned data
set (library) with the member name to be used in the MODS statement. If your
user exit operates in conjunction with other user exits in the same phase (for
example, E11, E15, and E17 user exits all use the same DCB), you can request
DFSORT to dynamically bind or link-edit them together (see MODS statement).
Alternatively, you can bind or link-edit them together into a partitioned data set
following these rules:
1. Specify RENT as a bind or linkage editor parameter.
2. Include an ALIAS statement for each user exit using the external entry name of
the routine (for example, the CSECT name).
3.
4. Specify the appropriate ALIAS name for each user exit on the MODS statement.
DFSORT includes the names and locations of your user exits in the list of modules
to be run during each phase. No user exit is loaded more than once in a program
phase, but the same user exit can appear in different phases. For example, you can
use the same Read Error user exit in both phases, but not twice in one phase.
The length you specify for a user exit must include storage for the user exit itself
as well as any storage used by the user exit outside of the load modules such as
I/O buffers or COBOL library subroutines. If you specify a ddname for a user exit
in the MODS statement, it must match the DD statement that defines the library
containing that user exit. For example:
//MYLIB DD
DSNAME=MYRTN, etc.
.
.
.
MODS
E15=(MODNAME,500,MYLIB,N)
User exit linkage conventions

To enter a user exit, DFSORT loads the address of the DFSORT return point in
register 14 and the address of the user exit routine in register 15. A branch to the
address in register 15 is then performed.
The general registers used by DFSORT for linkage and communication of
parameters observe operating system conventions. When your routine gets control,
the general registers have the following contents:
Register
Contents
1
496
DFSORT places the address of a parameter list in this register.

13
DFSORT places the address of a standard save area in this register. The
area can be used to save contents of registers used by your user exit. The
first word of the area contains the characters SM1 in its three low-order
bytes.
14
Contains the address of DFSORT return point.
15
Contains the address of your user exit. This register can be used as a base
register for your user exit; your user exit can also use it to pass return
codes to DFSORT.
You can return control to DFSORT by performing a branch to the DFSORT return
point address in register 14 or by using a RETURN macro instruction. The
RETURN instruction can also be used to set return codes when multiple actions
are available at a user exit.
Your user exit must save all the general registers it uses. You can use the SAVE
macro instruction to do this. If you save registers, you must also restore them; you
can do this with the RETURN macro instruction.
Linkage examples
When calling your user exit, DFSORT places the return address in general register
14 and your routine's entry point address in general register 15. DFSORT has
already placed the register's save area address in general register 13. DFSORT then
makes a branch to your routine.
Your routine for the E15 user exit might incorporate the following assembler
instructions:
ENTRY E15
.
.
E15 SAVE
(5,9)
.
.
RETURN (5,9)
This coding saves and restores the contents of general registers 5 through 9. The
macro instructions are expanded into the following assembler language code:
ENTRY E15
.
.
E15 STM
5,9,40(13)
.
.
LM
5,9,40(13)
BR
14
If multiple actions are available at a user exit, your routine sets a return code in
general register 15 to inform DFSORT of the action it is to take. The following
macro instruction can be used to return to DFSORT with a return code of 12 in
register 15:
RETURN
RC=12
A full explanation of linkage conventions and the macro instructions discussed in

this section is in z/OS MVS Programming: Assembler Services Guide.
497
Dynamically binding or link-editing user exit routines

You can dynamically bind or link-edit any user exit routine written in any
language that has the ability to pass the location or address of a record or
parameter in general register 1 and a return code in register 15 (see MODS
statement). This does not include E15 and E35 user exits written in COBOL.
Dynamic binding or link-editing does not support AMODE 31 or RMODE 31 for
the option T. The user exits that are bound or link-edited together by DFSORT are
not loaded above 16MB virtual and cannot be entered in 31-bit addressing mode.
User exits bound or link-edited with the S option retain the AMODE and RMODE
attributes of the object modules and are loaded above or below 16MB virtual
depending upon the load module's RMODE; they are entered in the addressing
mode of the user exit.
Note:
1. The Blockset technique is not used for dynamic binding or link-editing.
2. Dynamic binding or link-editing cannot be used with copy.
When the option T is specified for a user exit routine, that routine must contain an
entry point whose name is that of the associated program user exit. This is to
accommodate special DFSORT dynamic binder or link-edit requirements. For
example, when the option T is specified on the MODS statement for E35, the
following assembler instructions must be included in the user exit routine
associated with the E35 user exit:
E35
ENTRY E35
.
.
or
E35
CSECT
.
.
In all other circumstances, the user exit is not required to have an entry point that
has the same name as that of the associated program user exit.
Assembler user exit routines (input phase user exits)

You can use these program user exits in the DFSORT input phase:
v E11
v E15
v
v
v
v
v
E16
E17
E18
E19
E61
These user exits are discussed in sequence. To determine whether a particular user
exit can be used for your application, refer to Table 62 on page 491 and Table 63 on
page 491.
498
Assembler User Exit Routines (Input Phase User Exits)
E11 user exit: opening data sets/initializing routines

You might use routines at this user exit to open data sets needed by your other
routines in the input phase. It can also be used to initialize your other routines.
Return codes are not used, however.
Note: To avoid special linkage editor requirements (see Summary of rules for user
exit routines on page 495), you can include these functions in your E15 user exit
rather than in a separate E11 user exit routine.
E15 user exit: passing or changing records for sort and copy
applications
If you write your E15 user exit in COBOL, see COBOL user exit routines on page
521 and COBOL E15 user exit: passing or changing records for sort on page 523.
The EXITCK option affects the way DFSORT interprets certain return codes from
user exit E15. To avoid ambiguity, this section assumes that the IBM default,
EXITCK=STRONG, was selected at your site. For complete information about E15
return codes in various situations with EXITCK=STRONG and EXITCK=WEAK,
see E15/E35 return codes and EXITCK on page 537.
DFSORT enters the E15 user exit routine each time a new record is brought into
the input phase. DFSORT continues to enter E15 (even when there are no input
records) until the user exit tells DFSORT, with a return-code of 8, not to return.
See Figure 10 on page 489 for logic flow details.
Some uses for the E15 user exit are:
v
v
v
v
Adding records to an input data set

Passing an entire input data set to DFSORT
Deleting records from an input data set
Changing records in an input data set.
Note:
1. If your E15 user exit is processing variable-length records, include a 4-byte
RDW at the beginning of each record you change or insert, before you pass it
back to DFSORT. The format of an RDW is described in z/OS DFSMS Using
Data Sets or System Programming Reference. (Alternatively, you can pad records
to the maximum length and process them as fixed-length.)
2. DFSORT uses the specified or defaulted value for L2 in the RECORD statement
to determine the length of the records your E15 user exit passes back to
DFSORT. For fixed-length records, be sure that the length of each record your
E15 user exit changes or inserts corresponds to the specified or defaulted L2
value. For variable-length records, be sure that the RDW of each record your
E15 user exit changes or inserts indicates a length that is less than or equal to
the specified or defaulted L2 value. Unwanted truncation or abends may occur
if DFSORT uses the wrong length for the records passed to it by your E15 user
exit.
For details of the L2 value, see RECORD control statement on page 438.
3. If you use the E15 user exit to pass all your records to DFSORT, you can omit
the SORTIN DD statement, in which case you must include a RECORD
statement in the program control statements.
499

4. If you invoke DFSORT from an assembler program and pass the address of
your E15 user exit in the parameter list, DFSORT ignores the SORTIN data set
and terminates if you specify E15 in a MODS statement.
5. If you omit the SORTIN DD statement, or it is ignored, all input records are
passed to DFSORT through your routine at user exit E15. The address of each
input record in turn is placed in general register 1, and you return to DFSORT
with a return code of 12. When DFSORT returns to the E15 user exit after the
last record has been passed, you return to DFSORT with a return code of 8 in
register 15, which indicates "do not return."
6. DFSORT continues to reenter your E15 user exit until a return code of 8 is
received. However, if STOPAFT is in effect, no additional records are inserted
to DFSORT after the STOPAFT count is satisfied (even if you pass back a return
code of 12).
7. An RDW must be built for variable-length VSAM records (see z/OS DFSMS
Using Data Sets).
Information DFSORT passes to your routine at E15 user exit

Your E15 user exit routine is entered each time a new record is brought into the
input phase. DFSORT passes two fields to your routine each time it is entered:
v The address of the new record. End of input is reached when there are no more
records to pass to your E15 user exit; DFSORT indicates end of input by setting
this address to zero before entering your E15 user exit. If there are no records in
the input data set (or no input data set), this address is zero the first time your
E15 is entered.
After end of input is reached, DFSORT continues to enter your user exit routine until
you pass back a return code of 8.
Your E15 user exit must not change the address of the new record.
v The user exit address constant. If you invoked DFSORT with a user exit address
constant in the parameter list, the address constant is passed to your E15 user
exit the first time it is entered. This address constant can be changed by your
E15 user exit any time it is entered; the address constant is passed along on
subsequent entries to your E15 user exit and also on the first entry to your E35
user exit. For example, you can obtain a dynamic storage area, use it in your E15
user exit, and pass its address to your E35 user exit.
Note: The user exit address constant must not be used for a tape work data set
sort application.
Which E15 parameter list will be used

For the E15 exit, DFSORT supports both a 32-bit parameter list and a 64-bit
parameter list. If you want DFSORT to use the 64-bit parameter list for your E15,
you must call DFSORT with the 64-bit invocation list, and:
v use a MODS control statement for E15 with the N64 parameter or
v set on bit 4 of byte 9 in the 64-bit invocation list.
Otherwise, DFSORT will use the 32-bit parameter list for your E15.
32-bit E15 parameter list

In 32-bit general register 1, DFSORT places the address of a parameter list that
contains the record address and the user address constant. The list is two fullwords
long and begins on a fullword boundary. The format of the parameter list is:
500

Table 66. 32-bit E15 User Exit Parameter List
Contents
Bytes 1 through 4
Address of the new record
Bytes 5 through 8
User exit address constant
DFSORT provides a 72-byte Format 0 save area pointed to by 32-bit register 13 in

which the exit can save the 32-bit registers.
Before returning control to DFSORT, you must:
v Place the address of the record in 32-bit register 1.
v Put the return code in 32-bit register 15.

contains the record address and the user address constant. The list is two
doublewords long and begins on a doubleword boundary. The format of the
parameter list is:
Value
Contents
Bytes 1 through 8
X'00000000'
Address of the new record
Bytes 9 through 16
X'00000000'

The E15 should expect and use 64-bit addresses.
v Place the address of the record in 64-bit register 1. This must be a 64-bit address,
a clean 31-bit address or a clean 24-bit address.
The DFSORT target library, SICEUSER, contains a mapping macro called ICEPL64,
which provides a separate Assembler DSECT for the E15 64-bit parameter list.
E15 return codes

Your E15 routine must pass a return code to DFSORT. Following are the return
codes for the E15 user exit:
Return Code
Description
00 (X'00')
No Action/Record Altered
04 (X'04')
Delete Record
08 (X'08')
Do Not Return
12 (X'0C')
Insert Record
501

16 (X'10')
Terminate DFSORT
0: No Action
If you want DFSORT to retain the record unchanged, place the address of the
record in general register 1 and return to DFSORT with a return code of 0
(zero).
0: Record Altered
If you want to change the record before passing it back to DFSORT, your
routine must move the record into a work area, perform whatever modification
you want, place the address of the modified record in general register 1, and
return with a return code of 0 (zero).
4: Delete Record
If you want DFSORT to delete the record from the input data set, return to
DFSORT with a return code of 4. You need not place the address of the record
in general register 1.
8: Do Not Return
DFSORT continues to return control to the user routine until it receives a
return code of 8. After that, the user exit is not used again during the DFSORT
application. You need not place an address in general register 1 when you
return with a return code of 8. Unless you are inserting records after the end of the
data set, you must pass a return code of 8 when the program indicates the end of the
data set. It does this by passing your routine a zero address in the parameter
list.
If your user exit routine passes a return code of 8 to DFSORT when input
records still remain to be processed, the remaining records are processed by
DFSORT, but are not passed to your user exit.
12: Insert Record
To add a record before the record whose address was just passed to your
routine, place the address of the record to be added in general register 1 and
return to DFSORT with a return code of 12. DFSORT keeps returning to your
routine with the same record address as before so that your routine can insert
more records at that point or alter the current record. You can make insertions
after the last record in the input data set (after DFSORT places a zero address
in the parameter list). DFSORT keeps returning to your routine until you pass a
return code of 8.
16: Terminate DFSORT
If you want to terminate DFSORT, return with a code of 16. DFSORT then
returns to its calling program or to the system with a return code of 16.
See E15/E35 return codes and EXITCK on page 537 for complete details of the
meanings of return codes in various situations.
Storage usage for E15 user exit

DFSORT obtains storage (using GETMAIN or STORAGE OBTAIN) for the
parameter list and the records it passes to your E15 user exit routine. You must not
attempt to modify or free the storage obtained by DFSORT.
If you need to obtain storage for use by your E15 user exit routine, such as to pass
altered records to DFSORT, you can use the following strategy:
1. The first time your exit is called, obtain the storage you need
2. Use the storage you obtained each time your exit is called
3. Free the storage before you pass back return code 8 to DFSORT
502

Note: When you obtain your storage you can save its address in the user exit
address constant and restore it on each subsequent call to your exit.
E16 user exit: handling intermediate storage miscalculation

For a tape work data set sort or a Peerage/Vale sort without work data sets, you
would use a routine at this user exit to decide what to do if the sort exceeds its
calculated estimate of the number of records it can handle for a given amount of
main storage and intermediate storage. This user exit is ignored for a sort with
work data sets because DFSORT uses the WRKSEC option to determine whether
secondary allocation is allowed. See SORTWKdd DD statement on page 71. See
also Exceeding tape work space capacity on page 861.
Note: When using magnetic tape, remember that the system uses an assumed tape
length of 2400 feet. If you use tapes of a different length, the Nmax figure is not
accurate; for shorter tapes, capacity can be exceeded before NMAX EXCEEDED
is indicated.
E16 return codes

Return Code
Description
00 (X'00')
Sort Current Records Only
04 (X'04')
Try to Sort Additional Records
08 (X'08')
Terminate DFSORT
0: Sort Current Records Only
If you want DFSORT to continue with only that part of the input data set it
estimates it can handle, return with a return code of 0 (zero). Message ICE054I
contains the number of records with which sort is continuing. You can sort the
remainder of the data set on one or more subsequent runs, using SKIPREC to
skip over the records already sorted. Then you can merge the sort outputs to
complete the operation.
4: Try to Sort Additional Records
If you want DFSORT to continue with all of the input data set, return with a
return code of 4. If tapes are used, enough space might be available for
DFSORT to complete processing. If enough space is not available, DFSORT
generates a message and terminates. Refer to Exceeding tape work space
capacity on page 861.
8: Terminate DFSORT
If you want DFSORT to terminate, return with a return code of 8. DFSORT
then returns to its calling program or to the system with a return code of 16.
E17 user exit: closing data sets

Your E17 user exit routine is entered once at the end of the input phase. It can be
used to close data sets used by your other routines in the phase or to perform any
housekeeping functions for your routines.
503

exit routines on page 495), you can include these functions in your E15 user exits
rather than in a separate E17 user exit routine.
E18 user exit: handling input data sets

You can use this user exit to handle special I/O conditions for QSAM/BSAM and
VSAM input data sets.
Using E18 user exit with QSAM/BSAM

Your routines at this user exit can pass DFSORT a parameter list containing the
specifications for three data control block (DCB) fields: SYNAD, EXLST, and
EROPT. Your E18 user exit routine can also pass a fourth DCB field (EODAD) to
DFSORT.
Note: If you are using a disk sorting technique, the EROPT option is ignored.
Your routines are entered at the beginning of each phase so that DFSORT can
obtain the parameter lists. The routines are entered again during processing of the
phase at the points indicated in the parameter lists. For example, if you choose the
EXLST option, DFSORT enters your E18 user exit routine early in the sort (input)
phase. DFSORT picks up the parameter list including the EXLST address. Later in
the phase, DFSORT enters your routine again at the EXLST address when the data
set is opened.
Information your routine passes to DFSORT at E18 user exit: Before returning
control to DFSORT, your routine passes the DCB fields in a parameter list by
placing the parameter list address in general register 1. The parameter list must
begin on a fullword boundary and be a whole number of fullwords long. The
high-order byte of each word must contain a character code that identifies the
parameter. One or more of the words can be omitted. A word of all zeros marks
the end of the list.
If VSAM parameters are specified, they are accepted but ignored.
The format of the list is shown as follows:
Byte 1
Byte 2
Byte 3
01
SYNAD field
02
EXLST field
03
00
04
00
00
Byte 4
EROPT code
EODAD field
00
00
00
SYNAD
Contains the location of yourread synchronous error routine. This routine is
entered only after the operating system has tried unsuccessfully to correct the
error. The routine must be assembled as part of your E18 user exit routine.
When the routine receives control, it must not store registers in the save area
pointed to by register 13.
EXLST
Contains the location of a list of pointers to routines that you want used to
check labels and accomplish other tasks not handled by data management. The
504

list, and the routines to which it points, must be included in your read error
routine. This parameter can only be used for EXLST routines associated with
opening the first SORTIN data set.
EROPT
Indicates what action DFSORT must take when it encounters an uncorrectable
read error. The three possible actions and the codes associated with them are:
X'80'
Accept the record (block) as is
X'40'
Skip the record (block)
X'20'
Terminate the program.
If you include this parameter in the DCB field list, you must place one of the
previous codes in byte 4 of the word. Bytes 2 and 3 of the word must contain
zeros.
When you use the EROPT option, the SYNAD field and the EODAD field must
contain the appropriate address in bytes 2 through 4. Or, if no routine is
available, bytes 2 and 3 must contain zeros, and byte 4 must contain X'01'. You
can use the assembler instruction DC AL3(1) to set up bytes 2 through 4.
EODAD
Contains the address of your end-of-file routine. If you specify EODAD, you
must include the end-of-file routine in your own routine.
A full description of these DCB fields is contained in z/OS DFSMS Macro
Instructions for Data Sets.
Using E18 user exit with VSAM

If input to DFSORT is a VSAM data set, you can use the E18 user exit to perform
various VSAM user exit functions and to insert passwords in VSAM input ACBs.
E18 user exit restrictions: If passwords are to be entered through a user exit and
Blockset is not selected, the data set cannot be opened during the initialization
phase. This means that MAINSIZE|SIZE=MAX must not be used because the
program cannot make the necessary calculations.
Information your routine passes to DFSORT at E18 user exit: When you return
to DFSORT, you must place the address of a parameter list in general register 1:
Bytes 2 through 4
05
Address of VSAM user exit list
06
Address of password list
00
000000
If QSAM parameters are passed instead, they are accepted but ignored.
Either address entry can be omitted; if they both are included, they can be in any
order.
E18 password list: A password list included in your routine must have the
following format:
Two bytes on a halfword boundary:
505
Number of entries in list
Followed by the 16-byte entries:
8 bytes: ddname
8 bytes: Password
The last byte of the ddname field is destroyed by DFSORT. This list must not be
altered at any time during the program. MAINSIZE|SIZE=MAX must not be used
if this function is used.
E18 user exit list: The VSAM user exit list must be built using the VSAM EXLST
macro instruction giving the addresses of your routines handling VSAM user exit
functions. VSAM branches directly to your routines which must return to VSAM
via register 14.
Any VSAM user exit function available for input data sets can be used except
EODAD. If you need to do EODAD processing, write a LERAD user exit and
check for X'04' in the FDBK field of the RPL. This will indicate input EOD. This
field must not be altered when returning to VSAM because it is also needed by
DFSORT.
For details, see z/OS DFSMS Macro Instructions for Data Sets.
Figure 11 on page 507 shows an example of code your program can use to return
control to DFSORT.
506
ENTRY
E18
PARMLST
VSAMEXL
PWDLST
USYNAD
ULERAD
SER
LST
RTN
QSAMEOD
E18
.
.
LA
1,PARMLST
RETURN
CNOP
0,4
DC
X01
DC
AL3(SER)
DC
X02
DC
AL3(LST)
DC
X03
DC
X000080
EROPT CODE
DC
A(0)
DC
X04
DC
AL3(QSAMEOD)
DC
X05
DC
AL3(VSAMEXL)
DC
X06
DC
AL3(PWDLST)
DC
A(0)
.
.
EXLST SYNAD=USYNAD,LERAD=ULERAD
DC
H1
DC
CL8SORTIN
SORTIN DDNAME
DC
CL8INPASS
SORTIN PASSWORD
...
VSAM SYNCH ERROR RTN
...
VSAM LOGIC ERROR RTN
...
QSAM ERROR RTN
DC
X85,AL3(RTN) EXLST ADDRESS LIST1
...
EXLST ROUTINE
...
QSAM END OF FILE ROUTINE
Figure 11. E18 User Exit Example

1
X'85'= X'80' plus X' 05', where:

v X'80' means this entry is the LAST ENTRY of the list.
v X'05' means this user exit is the data control block user exit.
For more information, refer to z/OS DFSMS Using Data Sets.
E19 user exit: handling output to work data sets

This user exit is used to handle write error conditions in the input phase when
DFSORT is unable to correct a write error to a work data set. It is used only for a
tape work data set sort.

Your routines at this user exit can pass DFSORT a parameter list containing the
specifications for two DCB fields (SYNAD and EXLST). Your routines are entered
first early in the input phase so that DFSORT can obtain the parameter lists. The
routines are entered again later in the phase at the points indicated by the options
in the parameter lists.
Information your routine passes to DFSORT at E19 user exit: Before returning
control to DFSORT, your routine passes the DCB fields in a parameter list by
placing the parameter list address in general register 1. The list must begin on a
fullword boundary and must be a whole number of fullwords long. The first byte
of each word must contain a character code that identifies the parameter. Either
word can be omitted. A word of all zeros indicates the end of the list.
If VSAM parameters are passed, they are accepted but ignored.
The format of the list is as follows:
507

Byte 1
Byte 2
Byte 3
01
SYNAD field
02
EXLST field
00
00
00
Byte 4
00
SYNAD
This field contains the location of your write synchronous error routine. This
routine is entered only after the operating system has unsuccessfully tried to
correct the error. It must be assembled as part of your own routine.
EXLST
The EXLST field contains the location of a list of pointers. These pointers point
to routines that are used to process labels and accomplish other tasks not
handled by data management. This list, and the routines to which it points,
must be included as part of your own routine.
A full description of these DCB fields can be found in z/OS DFSMS Macro
Instructions for Data Sets.
E61 user exit: modifying control fields

You can use a routine at this user exit to lengthen, shorten, or alter any control
field within a record. The E option for the s parameter on the SORT or MERGE
control statement must be specified for control fields changed by this routine as
described in MERGE control statement on page 162 and SORT control
statement on page 442. After your routine modifies the control field, DFSORT
collates the records in ascending order using the format(s) specified.18
Note:
1. Routine E61 will not be used with EFS fields that have a D1 format.
2. Although your E61 routine alters control fields before a compare, your original
records are not altered.
3. If locale processing is used for SORT or MERGE fields, an E61 user exit must
not be used. DFSORT's locale processing may eliminate the need for an E61
user exit. See OPTION control statement on page 173 for information related
to locale processing.
Some uses of E61 user exit

Your routine can normalize floating-point control fields or change any other type
of control field in any way that you desire. You need to be familiar with the
standard data formats used by the operating system before modifying control
fields.
If you want to modify the collating sequence of EBCDIC data, for example, to
permit the alphabetic collating of national characters, you can do so without the
need for an E61 user exit routine by using the ALTSEQ control statement (as
described in ALTSEQ control statement on page 88).
18. With a conventional merge or a tape work data set sort, control fields for which E is specified are treated as binary byte format
regardless of the actual format(s) specified.
508

DFSORT places the address of a parameter list in general register 1. The list begins
on a fullword boundary and is three fullwords long. The parameter list for the E61
user exit is as follows:
Byte 1
Byte 2
Byte 3
Byte 4
00
00
00
Control Field No.
00
Address of Control Field Image

Not Used
Control Field Length
The control field length allows you to write a more generalized modification
routine.
To alter the control field, change the control field image at the indicated address
(changing the address itself will have no effect).
The control field number is relative to all fields in the SORT or MERGE statement.
SORT FIELDS=(4,2,CH,A,8,10,CH,E,25,2,BI,E)
field numbers 2 and 3 will be passed to user exit E61.

For all fields except binary, the total number of bytes DFSORT passes to your
routine is equal to the length specified in the m parameter of the SORT or MERGE
statement.
All binary fields passed to your routine contain a whole number of bytes; all bytes
that contain any bits of the control field are passed. If the control field is longer
than 256 bytes, DFSORT splits it into fields of 256 bytes each and passes them one
at a time to your routine.
Your routine cannot physically change the length of the control field. If you must
increase the length for collating purposes, you must previously specify that length
in the m parameter of the SORT or MERGE statement. If you must shorten the
control field, you must pad it to the specified length before returning it to
DFSORT. Your routine must return the field to DFSORT with the same number of
bytes that it contained when your routine was entered.
When user exit E61 is used, records are always ordered into ascending sequence. If
you need some other sequence, you can modify the fields further; for example, if
after carrying out your planned modification for a binary control field, and before
handing back control to DFSORT, you reverse all bits, the field is, in effect, collated
in descending order as illustrated by the E61 example in Figure 17 on page 521.
Note that if E61 is used to resolve ASCII collating for special alphabetic characters,
substituted characters must be in EBCDIC, but the sequencing depends upon the
byte value of the ASCII translation for the substituted character.
509
Assembler User Exit Routines (Output Phase User Exits)
Assembler user exit routines (output phase user exits)

You can use these program user exits located in the DFSORT output phase:
v E31
v E32
v E35
v E37
v E38
v E39
The functions of these user exits are discussed in sequence.
E31 user exit: opening data sets/initializing routines

You might use routines at this user exit to open data sets needed by your other
routines in the output phase or to initialize your other routines. Return codes are
not used.
rather than in a separate E31 routine.
E32 user exit: handling input to a merge only

This user exit can be used only in a merge operation invoked from a program and
cannot be specified on the MODS statement. When an E32 user exit is activated, it
must supply all input to the merge. DFSORT ignores SORTINnn data sets when an
E32 user exit is used.
You must indicate the number of input files you want to merge using either (1) the
FILES=n option on the MERGE control statement, or (2) the X'04' entry in the
24-bit parameter list. Your E32 user exit routine must insert records for these files
as DFSORT requests them.
If input is variable-length records, you must be sure the beginning of each record
contains a 4-byte RDW before merged. The format of an RDW is described in z/OS
DFSMS Macro Instructions for Data Sets (Alternatively, you can declare the records
as fixed-length and pad them to the maximum length.)

Your E32 user exit routine is entered each time the merge program requires a new
input record. DFSORT passes three fields to your routine:
v The increment of the next file to be used for input. The file increment is
0,4,8,...,N4, where N is four times the number of input files. Thus, the
increment 0 (zero) represents the first input file, 4 the second file, 8 the third,
and so on.
v The address of the next input record. Your routine must provide a separate
input buffer for each input file used. An input buffer containing the first record
for a file must not be altered until you have passed the first record from each
file to DFSORT.
v The user exit address constant. If you invoked DFSORT with a user exit address
constant in the parameter list, the address constant is passed to your E32 user
exit the first time it is entered. This address constant can be changed by your
510

E32 user exit any time it is entered; the address constant is passed along on
subsequent entries to your E32 user exit and E35 user exit. For example, you can
obtain a dynamic storage area, use it in your E32 user exit, and pass its address
to your E35 user exit.
Note: The user exit address constant must not be used for a Conventional merge
application.

you must:
v Call DFSORT with the 64-bit invocation list
v Set on bit 4 of byte 9 in the 64-bit invocation list

contains the file increment, the record address and the user address constant. The
list is three fullwords long and begins on a fullword boundary. The format of the
parameter list is:
Contents
Bytes 1 through 4
Increment of next file to be used for input
Bytes 5 through 8
Address of next input record
Bytes 9 through 12

which the exit can save the 32-bit registers
v Place the address of the next input record from the requested input file in the
second word of the parameter list

contains the file increment, the record address and the user address constant. The
list is three doublewords long and begins on a doubleword boundary. The format
of the parameter list is:
Contents
Description
Bytes 1 through 8
X'00000000'
Increment of next file to be

used for input
Bytes 9 through 16
Address of next input record
Bytes 17 through 24
X'00000000'
511

v Place the address of the next input record from the requested input file in the
second doubleword of the parameter list.
E32 return codes

Return Code
Description
08 (X'08')
End of input for requested file
12 (X'0C')
Insert record
16 (X'10')
Terminate DFSORT
8: End of input for requested file
DFSORT continues to return control to the user routine until it receives a
return code of 8 for every input file. After that, the user exit is not used again
during the DFSORT application. You need not place an address in the second
field of the parameter list when you return with a return code of 8.
12: Insert Record
To add a record from the requested input file, place the address of the next
record in the parameter list and return to DFSORT with a return code of 12.
DFSORT keeps returning to your routine until you pass a return code of 8 for every
input file.
returns to its calling program with a return code of 16.
E35 user exit: changing records

If you write your E35 user exit in COBOL, see COBOL user exit routines on page
521 and COBOL E35 user exit: changing records on page 529.
EXITCK=STRONG, was selected at your site. For complete details of the meaning
of E35 return codes in various situations with EXITCK=STRONG and
EXITCK=WEAK, see E15/E35 return codes and EXITCK on page 537.
DFSORT enters the E35 user exit routine each time it prepares to place a record in
the output area.
512

v Adding records for output data sets
v Omitting records for output data sets
v Changing records for output data sets
Note:
1. If your E35 user exit is processing variable-length records, include a 4-byte
RDW at the beginning of each record you change or insert, before you pass it
back to DFSORT. The format of an RDW is described in z/OS DFSMS Using
Data Sets or System Programming Reference. (Alternatively, you can pad records
to the maximum length and process them as fixed-length.)
2. DFSORT uses the specified or defaulted value for L3 in the RECORD statement
to determine the length of the records your E35 user exit passes back to
DFSORT. For fixed-length records, be sure that the length of each record your
E35 user exit changes or inserts corresponds to the specified or defaulted L3
value. For variable-length records, be sure that the RDW of each record your
E35 user exit changes or inserts indicates a length that is less than or equal to
the specified or defaulted L3 value. Unwanted truncation or abends may occur
if DFSORT uses the wrong length for the records passed to it by your E35 user
exit.
3. If you use the E35 user exit to dispose of all your output records, you can omit
the SORTOUT DD statement.
4. If you invoke DFSORT from a program and you pass the address of your E35
user exit in the parameter list:
v DFSORT ignores the SORTOUT data set (but not any OUTFIL data sets).
v DFSORT terminates if you specify E35 in a MODS statement.
5. If you omit the SORTOUT DD statement or it is ignored, and you do not
specify any OUTFIL data sets, your E35 user exit routine must dispose of each
output record and return to DFSORT with a return code of 4. When DFSORT
returns to your routine after you have disposed of the last record, return to
DFSORT with a return code of 8 to indicatedo not return.
6. Remember that if input records are variable-length from a VSAM data set, they
will have been prefixed by a 4-byte RDW.
7. After records have been put into the output area, their lengths cannot be
increased.
8. For a merge application, records deleted by an E35 user exit routine are not
sequence-checked. If you use an E35 user exit routine without an output data
set, sequence checking is not performed. In this case, you must ensure that the
records are sequenced correctly.

Your E35 user exit routine is entered each time DFSORT prepares to place a record
(including the first record) in the output area. DFSORT passes three fields to your
routine:
v The address of the record leaving DFSORT, which usually follows the record in
the output area. End of input is reached when there are no more records to pass
to your E35 user exit; DFSORT indicates end of input by setting this address to
zero before entering your E35 user exit.
After end of input is reached, DFSORT continues to enter your user exit routine until a
return code of 8 is passed back.
Your E35 user exit must not change the address of the record leaving DFSORT.
513

v The address of a record in the output area is zero the first time your routine is
entered because there is no record in the output area at that time. It remains
zero provided you pass a return code of 4 (delete record) to DFSORT.
Note: If the record pointed to is variable-length, it has an RDW at this point
even if output is to a VSAM data set.
v The user exit address constant is passed to your user exit exactly as it was set
by your E15 or E32 user exit or invoking program's parameter list.
Note: The user exit address constant must not be used for a Conventional merge
or tape work data set sort application.

you must call DFSORT with the 64-bit invocation list, and:
v use a MODS control statement for E35 with the N64 parameter or
v set on bit 5 of byte 9 in the 64-bit invocation list

contains the two record addresses and the user address constant. The list is three
fullwords long and begins on a fullword boundary. The format of the parameter
list is:
Contents
Bytes 1 through 4
Address of record leaving DFSORT
Bytes 5 through 8
Address of record in output area
Bytes 9 through 12

v Place the address of the record in 32-bit register 1.

contains the two record addresses and the user address constant. The list is three
doublewords long and begins on a doubleword boundary. The format of the
parameter list is:
Bytes 1 through 8
514
Contents
Description
X'00000000'
Address of record leaving

DFSORT

Table 71. 64-bit E35 User Exit Parameter List (continued)
Contents
Description
Bytes 9 through 16
X'00000000'
Address of record in output

area
Bytes 17 through 24
X'00000000'

v Place the address of the record in 64-bit register 1. This must be a 64-bit address,
E35 return codes

Return Code
Description
00 (X'00')
No Action/Record Altered
04 (X'04')
Delete Record
08 (X'08')
Do Not Return
12 (X'0C')
Insert Record
16 (X'10')
Terminate DFSORT
0: No Action
If you want DFSORT to retain the record unchanged, load the address of the
record leaving DFSORT in general register 1 and return to DFSORT with a
return code of 0 (zero).
0: Record Altered
If you want to change the record before having it placed in the output data set,
move the record to a work area, make the change, load the address of the
modified record into general register 1, and return to DFSORT with a return
code of 0 (zero).
4: Delete Record
Your routine can delete the record leaving DFSORT by returning to DFSORT
with a return code of 4. You need not place an address in general register 1.
8: Do Not Return
DFSORT keeps returning to your routine until you pass a return code of 8.
After that, the user exit is not used again during the DFSORT application.
515

When you return with a return code of 8, you need not place an address in
general register 1. Unless you are inserting records after the end of the data set, you
must pass a return code of 8 when DFSORT indicates the end of the data set. This is
done by passing a zero as the address of the record leaving DFSORT.
If you do not have an output data set and would usually return with a return
code of 8 before EOF, you can avoid getting the ICE025A message by
specifying NOCHECK on the OPTION control statement (if installation option
CHECK=NO had not already been specified).
DFSORT, but are not passed to your user exit.
12: Insert Record
To add an output record ahead of the record leaving DFSORT, place the
address of the new record in general register 1 and return to DFSORT with a
return code of 12. DFSORT returns to your routine with the same address as
passed on the previous call to the user exit for the record leaving DFSORT.
DFSORT passes the address of the inserted record as the address of the record
in the output area. You can make more insertions at that point, or delete the
record leaving DFSORT.
DFSORT does not perform sequence checking for disk work data set sorts. For
tape work data set sorts, DFSORT does not perform sequence checking on
records that you insert unless you delete the record leaving DFSORT and insert
a record to replace it. DFSORT keeps returning to your routine until you pass a
return code of 8.
returns to its calling program or to the system with a return code of 16.
Summing records at E35 user exit: You can use the SUM control statement to
sum records. However, you can sum records for output by changing the record in
the output area and then, if you want, by deleting the record leaving DFSORT.
DFSORT returns to your routine with the address of a new record leaving
DFSORT, and the same record remains in the output area, so that you can continue
summing. If you do not delete the record leaving DFSORT, that record is added to
the output area, and its address replaces the address of the previous record in the
output area. DFSORT returns with the address of a new record leaving DFSORT.
Storage usage for E35 user exit

DFSORT obtains storage (using GETMAIN or STORAGE OBTAIN) for the
parameter list and the records it passes to your E35 user exit routine. You must not
attempt to modify or free the storage obtained by DFSORT.
If you need to obtain storage for use by your E35 user exit routine, such as to pass
altered records to DFSORT, you can use the following strategy:
1. The first time your exit is called, obtain the storage you need
2. Use the storage you obtained each time your exit is called
3. Free the storage before you pass back return code 8 to DFSORT.
Note: When you obtain your storage you can save its address in the user exit
address constant and restore it on each subsequent call to your exit.
516
E37 user exit: closing data sets

Your E37 user exit routine is entered once at the end of the output phase. It can be
used to close data sets used by your other routines in the phase or to perform any
housekeeping functions for your routines.
rather than in a separate E37 user exit.
E38 user exit: handling input data sets

The routine here is the same as for E18. If the Blockset or Peerage/Vale technique
is selected, I/O error conditions cannot be handled through the E38 user exit.

This user exit can be used during a merge or copy to insert VSAM passwords into
VSAM input ACBs and to perform various VSAM user exit functions. The
following example shows code your program can use to return control to DFSORT.
ENTRY
E38
.
.
E38
LA
1,PARMLST
RETURN
CNOP
0,4
PARMLST DS
0H
DC
X05
DC
AL3(VSAMEXL)
DC
X06
DC
AL3(PWDLST)
DC
A(0)
.
.
VSAMEXL EXLST
SYNAD=USYNAD,LERAD=ULERAD
PWDLST
DC
H2
DC
CL8SORTIN01
SORTIN01 DDNAME
DC
CL8INPASS1
SORTIN01 PASSWORD
DC
CL8SORTIN02
SORTIN02 DDNAME
DC
CL8INPASS2
SORTIN02 PASSWORD
USYNAD
...
ULERAD
...
E39 user exit: handling output data sets

Your E39 user exit routine is entered for the SORTOUT data set, but not for
OUTFIL data sets.

The technique is the same as for E19 for QSAM/BSAM. See E19 user exit:
handling output to work data sets on page 507 for details.

For VSAM, this user exit can be used to insert VSAM passwords intothe VSAM
SORTOUT ACB and to perform various VSAM user exit functions. The example
that follows shows code your program can use to return control to DFSORT.
517
Sample E15 and E35 Routines using the 64-bit Parameter Lists
ENTRY
.
.
E39
LA
RETURN
CNOP
PARMLST DS
DC
DC
DC
DC
DC
.
.
VSAMEXL EXLST
PWDLST
DC
DC
DC
USYNAD
...
ULERAD
...
E39
1,PARMLST
0,4
0H
X05
AL3(VSAMEXL)
X06
AL3(PWDLST)
A(0)
SYNAD=USYNAD,LERAD=ULERAD
H1
CL8SORTOUT
SORTOUT DDNAME
CL8OUTPASS
SORTOUT PASSWORD
Sample E15 and E35 routines using the 64-bit parameter lists
Example 15. Sort with 64-bit parameter lists, E15, E35 and OUTFIL on page 839
in Chapter 11, Examples of DFSORT job streams, on page 819 shows, in
assembler language, how to use the 64-bit parameter lists for E15 and E35 exit
routines.
Sample routines written in assembler using the 32-bit parameter lists

This section provides some sample program user exits written in assembler using
the 32-bit parameter lists.
E15 user exit: altering record length

This routine changes the variable-length input records making them all the same
length.
518
Sample Routines Written in Assembler using the 32-bit Parameter Lists
E15
CSECT
* IF A RECORD IS GREATER THAN 204 BYTES, TRUNCATE IT TO 204 BYTES.
* IF A RECORD IS LESS THAN 204 BYTES, PAD IT OUT TO 204 BYTES.
* ALL OF THE RESULTING RECORDS WILL BE 204 BYTES LONG
* (4 BYTES FOR THE RDW AND 200 BYTES OF DATA).
USING E15,12
SHOW BASE REG
STM 14,12,12(13)
SAVE ALL REGS EXCEPT 13
LA
12,0(0,15)
SET BASE REG
ST
13,SAVE15+4
SAVE BACKWARD POINTER
LA
14,SAVE15
SET FORWARD POINTER
ST
14,8(13)
IN SAVE AREA
LR
13,14
SET OUR SAVE AREA
LR
2,1
SAVE PARM LIST POINTER
L
3,0(,2)
LOAD ADDR OF RECORD
LTR 3,3
EOF
BZ
EOF
YES - DO NOT RETURN
LH
4,0(,3)
GET RDW
CH
4,CON204
IS RDW EQ 204
BE
ACCEPT
YES-ACCEPT IT
BL
PAD
LESS THAN 204-PAD
LH
4,CON204
LIMIT LENGTH TO 204
B
TRUNC
MORE THAN 204-TRUNCATE
PAD
DS
0H
PAD OR TRUNCATE
MVI DATA,X00
ZERO OUT THE BUFFER
MVC DATA+1(199),DATA
TRUNC
DS
0H
PAD OR TRUNCATE
BCTR 4,0
DECREASE RDW FOR EXECUTE
EX
4,MVPAD
MOVE RECORD INTO PAD/TRUNCATE BUFFER
MVC NEWRDW(2),CON204 SET NEW RDW TO 204
LA
3,BUFFER
POINT TO PADDED/TRUNCATED RECORD
ACCEPT DS
0H
SR
15,15
SET RC=0
LR
1,3
SET RECORD POINTER
B
GOBACK
EOF
LA
15,8
EOF - SET RC=8
GOBACK L
13,4(,13)
L
14,12(,13)
LM
2,12,28(13)
RESTORE REGS
BR
14
RETURN
MVPAD
MVC BUFFER(*-*),0(3) FOR EXECUTE
SAVE15 DS
18F
CON204 DC
H204
BUFFER DS
0H
NEWRDW DS
H
NEW RDW OF 204
DC
H0
DATA
DC
XL20000
BUFFER FOR PADDING/TRUNCATING
END
E16 user exit: sorting current records when NMAX is

exceeded
This routine tells DFSORT that, when DFSORT issues the message "NMAX
EXCEEDED", it must sort only the records already read in.
E16
CSECT
LA
BR
END
15,0
14
SET RETURN CODE
E35 user exit: altering record length

This routine changes the variable-length output records making them all the same
length.
519
Sample Routines Written in Assembler using the 32-bit Parameter Lists
E35
CSECT
* IF A RECORD IS GREATER THAN 204 BYTES, TRUNCATE IT TO 204 BYTES.
* IF A RECORD IS LESS THAN 204 BYTES, PAD IT OUT TO 204 BYTES.
* ALL OF THE RESULTING RECORDS WILL BE 204 BYTES LONG
* (4 BYTES FOR THE RDW AND 200 BYTES OF DATA).
USING E35,12
SHOW BASE REG
STM 14,12,12(13)
SAVE ALL REGS EXCEPT 13
LA
12,0(0,15)
SET BASE REG
ST
13,SAVE15+4
LA
14,SAVE15
SET FORWARD POINTER
ST
14,8(13)
IN SAVE AREA
LR
13,14
SET OUR SAVE AREA
LR
2,1
SAVE PARM LIST POINTER
L
3,0(,2)
LOAD ADDR OF RECORD
LTR 3,3
EOF
BZ
EOF
YES - DO NOT RETURN
LH
4,0(,3)
GET RDW
CH
4,CON204
IS RDW EQ 204
BE
ACCEPT
YES-ACCEPT IT
BL
PAD
LESS THAN 204-PAD
LH
4,CON204
LIMIT LENGTH TO 204
B
TRUNC
MORE THAN 204-TRUNCATE
PAD
DS
0H
PAD OR TRUNCATE
MVI DATA,X00
ZERO OUT THE BUFFER
MVC DATA+1(199),DATA
TRUNC
DS
0H
PAD OR TRUNCATE
BCTR 4,0
DECREASE RDW FOR EXECUTE
EX
4,MVPAD
MOVE RECORD INTO PAD/TRUNCATE BUFFER
MVC NEWRDW(2),CON204 SET NEW RDW TO 204
LA
3,BUFFER
POINT TO PADDED/TRUNCATED RECORD
ACCEPT DS
0H
SR
15,15
SET RC=0
LR
1,3
SET RECORD POINTER
B
GOBACK
EOF
LA
15,8
EOF - SET RC=8
GOBACK L
13,4(,13)
L
14,12(,13)
LM
2,12,28(13)
RESTORE REGS
BR
14
RETURN
MVPAD
MVC BUFFER(*-*),0(3) FOR EXECUTE
SAVE15 DS
18F
CON204 DC
H204
BUFFER DS
0H
NEWRDW DS
H
NEW RDW OF 204
DC
H0
DATA
DC
XL20000
BUFFER FOR PADDING/TRUNCATING
END
E61 user exit: altering control fields

This routine can be used to change the order of binary control fields passed to it
(that is, those for which 'E' is specified) from ascending order to descending order.
520
COBOL User Exit Routines
* E61 PARAMETER LIST DSECT

PARML
DSECT
DS
3C
PARMNUM DS
C CONTROL FIELD NUMBER
PARMPTR DS
A ADDRESS OF CONTROL FIELD
DS
2C
PARMLEN DS
H CONTROL FIELD LENGTH
*
E61REV CSECT
* CHANGE THE ORDER OF EACH CONTROL FIELD PASSED TO THIS ROUTINE
* FROM ASCENDING TO DESCENDING BY REVERSING ALL OF THE BITS.
* ASSUMES THAT ONLY BI CONTROL FIELDS ARE PASSED.
USING E61REV,12
SHOW BASE REG
STM 14,12,12(13)
SAVE ALL REGS EXCEPT R13
LA
12,0(0,15)
SET BASE REG
ST
13,SAVE61+4
LA
14,SAVE61
SET FORWARD POINTER
ST
14,8(13)
IN SAVE AREA
LR
13,14
SET OUR SAVE AREA
LR
3,1
SET PARM LIST POINTER
USING PARML,3
L
4,PARMPTR
GET POINTER TO CONTROL FIELD IMAGE
LH
5,PARMLEN
GET LENGTH OF CONTROL FIELD
BCTR 5,0
SUBTRACT 1 FOR EXECUTE
EX
5,REVCF
CHANGE FROM ASCENDING TO DESCENDING
GOBACK L
13,4(,13)
LM
14,12,12(13)
RESTORE REGS
BR
14
RETURN
REVCF
XC
0(*-*,4),REVFF
REVERSE CONTROL FIELD BITS
SAVE61 DS
18F
REVFF
DC
256XFF
LTORG
END
COBOL user exit routines

You can perform the same tasks with E15 and E35 user exit routines written in
COBOL that you can perform with E15 and E35 user exit routines written in
assembler. However, COBOL routines differ from assembler routines in the way
they pass information between themselves and DFSORT.
v COBOL routines must pass information through fields described in the
LINKAGE SECTION of the DATA DIVISION. Assembler uses general register 1
and pointers in a parameter list.
v COBOL routines must use RETURN-CODE, a COBOL special register. Assembler
uses register 15 for the return code.
v COBOL routines must use return code 20 to alter or replace a record. Assembler
uses return code 0.
v COBOL routines can use the user exit area for E15/E35 communication.
Assembler uses the user address constant.
COBOL user exit requirements

The following rules apply to COBOL user exits. Failure to observe these COBOL
user exit rules can result in termination or unpredictable results.
v User exits written in COBOL must not use STOP RUN statements. To return to
DFSORT, use the GOBACK statement.
v VS COBOL II user exits must be compiled with the RES and RENT compiler
options.
v If a user exit contains a DISPLAY statement, the DFSORT messages normally
written to SYSOUT must be directed to another data set using the MSGDDN
parameter. For DISPLAY statements, COBOL also writes to SYSOUT. The
messages to SYSOUT can, therefore, be lost because of interleaving of output.
521

An alternative is to direct the COBOL output to another data set by using the
OUTDD compiler option.
v COBOL user exits must not contain a SORT or a MERGE verb.
v When coding the MODS control statement to describe a COBOL user exit, use C
for the fourth parameter. This instructs DFSORT to build the correct parameter
list.
v If invoking DFSORT from a COBOL program, you can use a COBOL E15 if the
FASTSRT option is in effect for input and a COBOL E35 if FASTSRT is in effect
for output.
v COBOL library routines in Language Environment must be available at run time.
COBOL requirements for copy processing

For copy processing, all sort requirements apply except for the following
restrictions:
v When you directly invoke DFSORT, you can use either a separately compiled
COBOL E15 user exit or a separately compiled COBOL E35 user exit, but not
both.
v When you invoke DFSORT from a COBOL program, the following limitations
apply when FASTSRT is in effect for:
Input only: You can use a separately compiled E15 user exit, but not a
separately compiled E35 user exit.
Output only: You can use a separately compiled E35 user exit, but not a
separately compiled E15 user exit.
Input and output: You can use either a separately compiled E15 or a separately
compiled E35, but not both together.
If separately compiled E15 and E35 user exits are found together, DFSORT copy
processing terminates. Message ICE161A is issued.
COBOL storage requirements

If you are running COBOL user exits compiled with the RES compiler option,
make sure that you have enough storage available for the COBOL library
subroutines. (This does not apply if the library has been installed resident.)
Besides the minimum DFSORT main storage requirements, you need an additional
1200KB of storage in your REGION to run Language Environment.
Under certain conditions, DFSORT can use all the storage in your REGION below
16MB virtual, thus leaving no room to load the COBOL library subroutines
required during processing of your user exit.
Main storage is available above 16MB virtual unless the TMAXLIM or
SIZE/MAINSIZE options specify an extremely high value (for example, your
system limit for main storage above 16MB virtual). In that case, you can use the
ARESALL or ARESINV option to release storage.
During processing, the actual amount of storage required for the COBOL library
subroutines depends on the functions performed in the COBOL user exit. You must
add a minimum of 20KB to the size of the user exit. If the user exit does I/O,
additional storage must be reserved for the I/O buffers. Additional storage for
buffers is specified by the m parameter on the MODS statement.
When SIZE/MAINSIZE=MAX is in effect, an alternative way to release storage is
to use the RESALL or RESINV option.
522

Note: You might need to release an additional 70KB of storage when you are:
v Calling both E15 and E35 user exits
v Running with nonresident library subroutines
This can be done by adding 70KB more to one of the following:
v The m parameter of the MODS statement for the E35 user exit (m = E35 user
exit size + 20KB + 70KB)
v The RESALL option when SIZE/MAINSIZE=MAX is in effect.
COBOL user exit routines (input phase user exit)

COBOL E15 user exit: passing or changing records for sort
DFSORT enters the E15 user exit routine each time a new record is brought into
the input phase. DFSORT continues to enter E15 (even when there are no input
records) until the user exit tells DFSORT, with a return code of 8, not to return.
v Adding records to an input data set
v Passing an entire input data set to DFSORT
v Deleting records from an input data set
v Changing records in an input data set.
Note:
1. If both E15 and E35 user exits are used, they must be in the same version of
COBOL.
2. If you use the E15 user exit to pass all your records to DFSORT, you can omit
the SORTIN DD statement, in which case you must include a RECORD
statement in the program control statements.
3. If you omit the SORTIN DD statement, all input records are passed to DFSORT
through your COBOL E15 user exit. You return to DFSORT with a return code
of 12. When DFSORT returns to the E15 user exit after the last record has been
passed, you return to DFSORT with a return code of 8 in register 15, which
indicates do not return.
4. DFSORT continues to reenter your E15 user exit until a return code of 8 is
received. However, if STOPAFT is in effect, no additional records are inserted
to DFSORT after the STOPAFT count is satisfied (even if you pass back a return
code of 12).
5. You cannot use dynamic binding or link-editing with a COBOL E15 user exit.
E15 interface with COBOL

Each time the E15 user exit is called, DFSORT supplies the following fields:
v Record flags
v New record
523
COBOL User Exit Routines (Input Phase User Exit)

v Length of the new record (for variable-length records)
v Length of user exit area
v User exit area.
When E15 returns to DFSORT, the E15 user exit provides to DFSORT some or all of
the fields mentioned in the following. The first field is required; the others can be
modified as appropriate.
v RETURN-CODE (assigned by the user exit by setting the COBOL special register
RETURN-CODE)
v Return record
v Length of the return record (for VLR)
v User Exit area.
For more information on how these fields are used in a COBOL E15 user exit, see
E15 LINKAGE SECTION fields for fixed-length and variable-length records on
page 526.
Figure 18 on page 525 details the interface to COBOL for the E15 user exit.
524
Figure 18. E15 DFSORT Interface with COBOL
E15 LINKAGE SECTION examples: Figure 19 on page 526 is an example of the

LINKAGE SECTION code for a fixed-length record (FLR) data set with a logical
record length (LRECL) of 100. The example shows the layout of the fields passed
to your COBOL routine.
525
LINKAGE SECTION.
01 RECORD-FLAGS
PIC 9(8) BINARY.
88 FIRST-REC
VALUE 00.
88 MIDDLE-REC
VALUE 04.
88 END-REC
VALUE 08.
01 NEW-REC
PIC X(100).
01 RETURN-REC
PIC X(100).
01 UNUSED1
PIC 9(8) BINARY.
01 UNUSED2
PIC 9(8) BINARY.
01 UNUSED3
PIC 9(8) BINARY.
01 UNUSED4
PIC 9(8) BINARY.
01 UNUSED5
PIC 9(8) BINARY.
01 EXITAREA-LEN
PIC 9(4) BINARY.
01 EXITAREA.
05 EAREA OCCURS 1 TO 256 TIMES
DEPENDING ON EXITAREA-LEN PIC X.
Figure 19. LINKAGE SECTION Code Example for E15 (Fixed-Length Records)
Figure 20 is an example of the LINKAGE SECTION code for a variable-length

record (VLR) data set with a maximum LRECL of 200. The example shows the
layout of the fields passed to your COBOL routine.
LINKAGE SECTION.
01 RECORD-FLAGS
PIC 9(8) BINARY.
88 FIRST-REC
VALUE 00.
88 MIDDLE-REC
VALUE 04.
88 END-REC
VALUE 08.
01 NEW-REC.
05 NREC OCCURS 1 TO 200 TIMES
DEPENDING ON NEW-REC-LEN
PIC X.
01 RETURN-REC.
05 RREC OCCURS 1 TO 200 TIMES
DEPENDING ON RETURN-REC-LEN PIC X.
01 UNUSED1
PIC 9(8) BINARY.
01 UNUSED2
PIC 9(8) BINARY.
01 NEW-REC-LEN
PIC 9(8) BINARY.
01 RETURN-REC-LEN
PIC 9(8) BINARY.
01 UNUSED3
PIC 9(8) BINARY.
01 EXITAREA-LEN
PIC 9(4) BINARY.
01 EXITAREA.
Figure 20. LINKAGE SECTION Code Example for E15 (Variable-Length Record)
E15 LINKAGE SECTION fields for fixed-length and

variable-length records
The fields in the LINKAGE SECTION are used by DFSORT and your routine as
stated later in this section. For clarity, the field names from Figure 20 have been
used.
v To give your COBOL routine the status of the passed records, DFSORT uses the
record flags field (RECORD-FLAGS) in the following way:
0 (FIRST-REC)
The new record is the first passed record.
4 (MIDDLE-REC)
The new record is not the first passed record.
8 (END-REC)
All records have been passed to your routine or there were no records to
pass.
v DFSORT places the next input record in the new record field (NEW-REC). A
VLR does not contain an RDW, but DFSORT places the length of this VLR in the
526

new record length field (NEW-REC-LEN). The value in the NEW-REC-LEN field
is the length of the record only and does not include the 4 bytes for the RDW.
v When your routine places an insertion/replacement record in the return record
field (RETURN-REC), the VLR must not contain an RDW; your routine must
place the length of this record in the return record length field
(RETURN-REC-LEN). The value of the RETURN-REC-LEN field is the length of
the record only and must not include the 4 bytes for the RDW.
v Each time DFSORT calls your COBOL E15 or COBOL E35 user exit, it passes the
user exit a 256-byte user exit area field (EXITAREA). The first time the user exit
area field is passed to your COBOL E15 user exit, it contains 256 blanks, and the
user exit area length field (EXITAREA-LEN) contains 256.
Any changes you make to the user exit area field or user exit area length fields
are passed back both to your COBOL E15 user exit and your COBOL E35 user
exit.
Note:
1. Do not set the user exit area length field to more than 256 bytes.
2. If the data used for input was not created by a COBOL run, you need to
know the LRECL defined for your data set. For a VLR, the maximum length
of the record defined in your COBOL user exit is 4 bytes less than the
LRECL value, because COBOL does not include the RDW as part of the
record. (Each VLR begins with an RDW field of 4 bytes. The RDW is not
included in the record passed to your COBOL user exit.)
3. You need to code only up to the last field that your routine actually uses (for
example, up to RETURN-REC if you do not use the user exit area).
4. DFSORT uses the specified or defaulted value for L2 in the RECORD
statement to determine the length of the records your E15 user exit passes
back to DFSORT. For fixed-length records, be sure that each record your E15
user exit changes or inserts has a length that is equal to the specified or
defaulted L2 value. For variable-length records, be sure that each record your
E15 user exit changes or inserts has a length that is less than or equal to the
specified or defaulted L2 value. Unwanted truncation or abends may occur if
DFSORT uses the wrong length for the records passed to it by your E15 user
exit.
E15 return codes

Your COBOL E15 routine must pass a return code to DFSORT in the
RETURN-CODE field, a COBOL special register. Following are the return codes for
the E15 user exit:
Return Code
Description
00 (X'00')
No Action
04 (X'04')
Delete Record
08 (X'08')
Do Not Return
12 (X'0C')
Insert Record
527

16 (X'10')
Terminate DFSORT
20 (X'14')
Alter or Replace Record
0: No Action
If you want DFSORT to retain the record unchanged, return with
RETURN-CODE set to 0.
4: Delete Record
If you want DFSORT to delete the record, return with RETURN-CODE set to 4.
8: Do Not Return
DFSORT continues to enter your routine until you return with RETURN-CODE
set to 8. After that, the user exit is not used again during the DFSORT
application. Unless you are inserting records after the end of the data set, you must
set RETURN-CODE to 8 when DFSORT indicates the end of the data set, which it
does by entering your routine with the record flags field set to 8.
DFSORT but are not passed to your user exit.
12: Insert Record
If you want DFSORT to add a record before the new record in the input data
set:
v Move the insert record to the return record field
v For VLR, move the length to the return record length field (Do not include
the 4-byte RDW in this length.)
v Return with RETURN-CODE set to 12.
DFSORT returns to your routine with the same record as before in the new
record field, allowing your routine to insert more records or handle the new
record.
You can also insert records after the end of the data set. DFSORT keeps
returning to your routine as long as you pass it a RETURN-CODE of 12 and until
you return with a RETURN-CODE set to 8.
If you want to terminate DFSORT, return with RETURN-CODE set to 16.
DFSORT then returns to its calling program or to the system with a return
code of 16.
20: Alter Record
If you want to change the new record:
v Move the new record to the return record field.
v Change the record in the return record field.
v For VLR records, move the length to the return record length field.
Note: If your routine changes record size, you must indicate the new size on
the RECORD statement.
20: Replace Record
If you want to replace the new record:
v Move the replacement record to the return record field.
528

v For VLR records, move the length to the return record length field. (Do not
include the 4-byte RDW in this length.)
E15 procedure division requirements

When coding the PROCEDURE DIVISION, the following requirements must be
met:
v To return control to DFSORT, you must use the GOBACK statement.
v In the USING option of the PROCEDURE DIVISION header, you must specify
each 01-level name in the LINKAGE SECTION. You must specify each name in
order up to the last one you plan to use even when you do not use all the
01-level names preceding the header.
Examples:
For the FLR example, Figure 19 on page 526, you would code:
PROCEDURE DIVISION USING RECORD-FLAGS, NEW-REC,
RETURN-REC, UNUSED1, UNUSED2, UNUSED3,
UNUSED4, UNUSED5, EXITAREA-LEN, EXITAREA.
For the VLR example, Figure 20 on page 526, you would code:
PROCEDURE DIVISION USING RECORD-FLAGS, NEW-REC,
RETURN-REC, UNUSED1, UNUSED2,
NEW-REC-LEN, RETURN-REC-LEN,
UNUSED3, EXITAREA-LEN, EXITAREA.
COBOL user exit routines (output phase user exit)

COBOL E35 user exit: changing records
DFSORT enters the E35 user exit routine each time it prepares to place a record in
the output area.
v Adding records for output data sets
v Omitting records for output data sets
v Changing records for output data sets
When DFSORT indicates the end of the data set (record flags field set to 8), you
must set RETURN-CODE to 8 (unless you are inserting records after the end of the
data set); otherwise, DFSORT continues to enter E35.
Note:
1. If both E15 and E35 user exits are used, they must be in the same version of
COBOL.
529
COBOL User Exit Routines (Output Phase User Exit)

2. If you use the E35 user exit to dispose of all your output records, you can omit
the SORTOUT DD statement.
3. If you omit the SORTOUT DD statement and you do not specify any OUTFIL
data sets, your E35 user exit routine must dispose of each output record and
return to DFSORT with a return code of 4. When DFSORT returns to your
routine after you have disposed of the last record, return to DFSORT with a
return code of 8 to indicate do not return.
4. You cannot use dynamic binding or link-editing with a COBOL E35 user exit.
E35 interface with COBOL

Each time your E35 user exit is called, DFSORT supplies the following fields:
v Record flags
v Record leaving DFSORT
v Length of record leaving DFSORT (for variable-length records)
v User Exit area.
When your E35 user exit returns to DFSORT, the E35 user exit provides to
DFSORT some or all of the fields mentioned in the following. The first field is
required; the others can be modified as appropriate.
v RETURN-CODE (assigned by the user exit by setting the COBOL special register
RETURN-CODE)
v Return record
v Length of return record (for variable-length records)
v User exit area.
For more information on how these fields are used in a COBOL E35 user exit, see
E35 LINKAGE SECTION fields for fixed-length and variable-length records on
page 532.
Figure 21 on page 531 details the interface to COBOL for the E35 user exit.
530
Figure 21. E35 Interface with COBOL
E35 LINKAGE SECTION examples: Figure 22 on page 532 is an example of the

LINKAGE SECTION code for a fixed-length record (FLR) data set with a logical
record length (LRECL) of 100. The example shows the layout of the fields passed
to your COBOL routine.
531
LINKAGE SECTION.
01 RECORD-FLAGS
PIC 9(8) BINARY.
88 FIRST-REC
VALUE 00.
88 MIDDLE-REC
VALUE 04.
88 END-REC
VALUE 08.
01 LEAVING-REC
PIC X(100).
01 RETURN-REC
PIC X(100).
01 OUTPUT-REC
PIC X(100).
01 UNUSED1
PIC 9(8) BINARY.
01 UNUSED2
PIC 9(8) BINARY.
01 UNUSED3
PIC 9(8) BINARY.
01 UNUSED4
PIC 9(8) BINARY.
01 EXITAREA-LEN
PIC 9(4) BINARY.
01 EXITAREA.
Figure 22. LINKAGE SECTION Code Example for E35 (Fixed-Length Records)
Figure 23 is an example of the LINKAGE SECTION code for a variable-length

record (VLR) data set with a maximum LRECL of 200. The example shows the
layout of the fields passed to your COBOL routine.
LINKAGE SECTION.
01 RECORD-FLAGS
PIC 9(8) BINARY.
88 FIRST-REC
VALUE 00.
88 MIDDLE-REC
VALUE 04.
88 END-REC
VALUE 08.
01 LEAVING-REC.
05 LREC OCCURS 1 TO 200 TIMES
DEPENDING ON LEAVING-REC-LEN
PIC X.
01 RETURN-REC.
05 RREC OCCURS 1 TO 200 TIMES
01 OUTPUT-REC.
05 OREC OCCURS 1 TO 200 TIMES
DEPENDING ON OUTPUT-REC-LEN PIC X.
01 UNUSED1
PIC 9(8) BINARY.
01 LEAVING-REC-LEN
PIC 9(8) BINARY.
01 RETURN-REC-LEN
PIC 9(8) BINARY.
01 OUTPUT-REC-LEN
PIC 9(8) BINARY.
01 EXITAREA-LEN
PIC 9(4) BINARY.
01 EXITAREA.
Figure 23. LINKAGE SECTION Code Example for E35 (Variable-Length Records)
E35 LINKAGE SECTION fields for fixed-length and

variable-length records
The fields in the LINKAGE SECTION are used by DFSORT and your routine as
stated later in this section. For clarity, the field names from Figure 23 have been
used.
v To give your COBOL routine the status of the passed records, DFSORT uses the
record flags field (RECORD-FLAGS) in the following way:
0 (FIRST-REC)
The record leaving DFSORT is the first passed record.
4 (MIDDLE-REC)
The record leaving DFSORT is not the first passed record.
8 (END-REC)
There is no record leaving DFSORT to pass; all records have been passed
to your routine or there were no records to pass.
v DFSORT places the next output record, which usually follows the record in the
output area, in the record leaving field (LEAVING-REC). A VLR does not contain
an RDW; DFSORT places the length of this VLR in the record-leaving length
532

field (LEAVING-REC-LEN). The value in the LEAVING-REC-LEN field is the
length of the record only and does not include the 4 bytes for the RDW.
v When your routine places an insertion or replacement record in the return
record field (RETURN-REC), the VLR must not contain an RDW; your routine
must place the length of this record in the return record length field
(RETURN-REC-LEN). The value in the RETURN-REC-LEN field is the length of
the record only and does not include the 4 bytes for the RDW.
v DFSORT places the record already in the output area in the record in output
area field (OUTPUT-REC). A VLR does not contain an RDW. DFSORT places the
length, not including the 4 bytes for RDW, of this VLR in the record in output
area length field (OUTPUT-REC-LEN).
v DFSORT passes your COBOL E35 routine a 256-byte user exit area field
(EXITAREA) that can contain information passed by your COBOL E15 routine. If
no information is passed in the EXITAREA field by your COBOL E15 routine the
first time the field is passed to your COBOL E35 routine, EXITAREA contains
256 blanks, and the user exit area length field (EXITAREA-LEN) contains 256.
Any changes you make to the user exit area field or user exit area length field
are passed back to your COBOL E35 routine each time it is called by DFSORT.
Note:
1. Do not set the user exit area length field to more than 256 bytes.
2. VLR records have a 4-byte RDW field at the beginning of each record. The
maximum record length plus the RDW will be the length defined for the
LRECL attribute of your output data set. COBOL programs do not use the
RDW and, therefore, the maximum length defined in your COBOL user exit
is 4 bytes less than the LRECL value.
3. You need to code only up to the last field your routine actually uses (for
example, up to OUTPUT-REC-LEN if you do not use the user exit area).
4. DFSORT uses the specified or defaulted value for L3 in the RECORD
statement to determine the length of the records your E35 user exit passes
back to DFSORT. For fixed-length records, be sure that each record your E35
user exit changes or inserts has a length that is equal to the specified or
defaulted L3 value. For variable-length records, be sure that each record your
E35 user exit changes or inserts has a length that is less than or equal to the
specified or defaulted L3 value. Unwanted truncation or abends may occur if
DFSORT uses the wrong length for the records passed to it by your E35 user
exit.
E35 return codes

Your COBOL E35 routine must pass a return code to DFSORT in the
RETURN-CODE field, a COBOL special register. Following are the return codes for
the E35 exit:
Return Code
Description
00 (X'00')
No Action
04 (X'04')
Delete Record
08 (X'08')
Do Not Return
533

12 (X'0C')
Insert Record
16 (X'10')
Terminate DFSORT
20 (X'14')
Alter or Replace Record
0: No Action
If you want DFSORT to retain the record leaving DFSORT unchanged, return
with RETURN-CODE set to 0.
4: Delete Record
If you want DFSORT to delete the record leaving DFSORT, return with
RETURN-CODE set to 4.
8: Do Not Return
DFSORT keeps returning to your routine until you pass a RETURN-CODE set
to 8. After that, the user exit is not used again during the DFSORT application.
Unless you are inserting records after the end of the data set, you must set
RETURN-CODE to 8 when DFSORT indicates the end of the data set. This is done
by entering your routine with the record flags field set to 8.
DFSORT but are not passed to your user exit.
If you do not have an output data set and would usually return with a return
code of 8 before EOF, you can avoid getting the ICE025A message by
specifying NOCHECK on the OPTION control statement (if installation option
CHECK=NO had not already been specified).
12: Insert Record
If you want DFSORT to add an output record before the record leaving
DFSORT:
v Move the insert record to the return record field
v For VLR records, move the length to the return record length field
DFSORT returns to your routine with the inserted record in the record output
area field and with the same record as before in the record leaving DFSORT
field. In this way, your routine can insert more records or handle the record
leaving DFSORT.
You can also insert records after the end of the data set. DFSORT keeps
returning to your routine as long as you pass it a RETURN-CODE 12 and until you
return with RETURN-CODE set to 8.
DFSORT does not perform sequence checking for disk work data set sorts. For
tape work data set sorts, DFSORT does not perform sequence checking on
inserted records unless you delete the record leaving DFSORT and then replace
it.
If you want to terminate DFSORT, return with RETURN-CODE set to 16.
DFSORT then returns to its calling program or to the system with a return
code of 16.
20: Alter Record
If you want to change the record leaving DFSORT:
534

v
v
v
v
Move the record leaving DFSORT to the return record field

Change the record in the return record field
For VLR records, move the length to the return record length field
Return with RETURN-CODE set to 20.
Note: If your routine changes record size, you must indicate the new size on
the RECORD statement.
20: Replace Record
If you want to replace the record leaving DFSORT:
v Move the replacement record to the return record field
v For VLR records, move the length to the return record length field
E35 procedure division requirements

When coding the PROCEDURE DIVISION, the following requirements must be
met:
v To return control to DFSORT, you must use the GOBACK statement.
v In the USING option of the PROCEDURE DIVISION header, you must specify
each 01-level name in the LINKAGE SECTION. You must specify each name in
order up to the last one you plan to use even when you do not use all the
01-level names preceding the header.
Examples:
For the FLR example, Figure 22 on page 532, you would code:
PROCEDURE DIVISION USING RECORD-FLAGS, LEAVING-REC,
RETURN-REC, OUTPUT-REC, UNUSED1, UNUSED2,
UNUSED3, UNUSED4, EXITAREA-LEN, EXITAREA.
For the VLR example, Figure 23 on page 532, you would code:
RETURN-REC, OUTPUT-REC, UNUSED1,
LEAVING-REC-LEN, RETURN-REC-LEN,
OUTPUT-REC-LEN, EXITAREA-LEN, EXITAREA.
Sample routines written in COBOL

This section provides some sample program user exits written in COBOL.
COBOL E15 user exit: altering records

Figure 24 on page 536 shows an example of a COBOL E15 routine for a data set
with fixed-length records of 100 bytes. It examines the department field in the
passed record and takes the following action:
v If the department is D29, it changes it to J99.
v If the department is not D29, it accepts the record unchanged.
535
Sample Routines Written in COBOL
IDENTIFICATION DIVISION.
PROGRAM-ID.
CE15.
ENVIRONMENT DIVISION.
DATA DIVISION.
LINKAGE SECTION.
01 RECORD-FLAGS
88 FIRST-REC
88 MIDDLE-REC
88 END-REC
01 NEW-REC.
05 NFILL1
05 NEW-DEPT
05 NFILL2
01 RETURN-REC.
05 RFILL1
05 RETURN-DEPT
05 RFILL2
PIC 9(8) BINARY.

VALUE 00.
VALUE 04.
VALUE 08.
PIC X(10).
PIC X(3).
PIC X(87).
PIC X(10).
PIC X(3).
PIC X(87).
PROCEDURE DIVISION USING RECORD-FLAGS, NEW-REC, RETURN-REC.

IF END-REC
MOVE 8 TO RETURN-CODE
ELSE
IF NEW-DEPT EQUAL TO "D29"
MOVE NEW-REC TO RETURN-REC
MOVE "J99" TO RETURN-DEPT
ELSE
END-IF
END-IF
GOBACK.
Figure 24. COBOL E15 Routine Example (FLR)
COBOL E35 user exit: inserting records

Figure 25 on page 537 shows an example of a COBOL E35 routine for a data set
with variable-length records up to 200 bytes. It examines the department field in
each passed record (records are assumed to be sorted by the department field) and
takes the following action:
v It inserts a record for department K22 in the proper sequence.
v It accepts all passed records unchanged.
536
E15/E35 Return Codes and EXITCK
IDENTIFICATION DIVISION.
PROGRAM-ID.
CE35.
ENVIRONMENT DIVISION.
DATA DIVISION.
WORKING-STORAGE SECTION.
01 INSERT-DONE PIC 9(1) VALUE 0.
01 K22-REC.
05 K22-MANAGER PIC X(20) VALUE "J. DOE".
05 K22-DEPT
PIC X(3) VALUE "K22".
05 K22-FUNC
PIC X(20) VALUE "ACCOUNTING".
05 K22-LATER PIC X(30) VALUE SPACES.
01 LEAVING-VAR-LEN PIC 9(8) BINARY.
LINKAGE SECTION.
01 RECORD-FLAGS
PIC 9(8) BINARY.
88 FIRST-REC
VALUE 00.
88 MIDDLE-REC
VALUE 04.
88 END-REC
VALUE 08.
01 LEAVING-REC.
05 LREC-MANAGER PIC X(20).
05 LREC-DEPT
PIC X(3).
05 LREC-FUNC
PIC X(20).
05 LREC-LATER OCCURS 1 TO 157 TIMES
DEPENDING ON LEAVING-VAR-LEN PIC X.
01 RETURN-REC.
05 RREC
OCCURS 1 TO 200 TIMES
01 OUTPUT-REC.
05 OREC
OCCURS 1 TO 200 TIMES
DEPENDING ON OUTPUT-REC-LEN PIC X.
01 UNUSED1
PIC 9(8) BINARY.
01 LEAVING-REC-LEN
PIC 9(8) BINARY.
01 RETURN-REC-LEN
PIC 9(8) BINARY.
01 OUTPUT-REC-LEN
PIC 9(8) BINARY.
RETURN-REC, OUTPUT-REC, UNUSED1, LEAVING-REC-LEN,
RETURN-REC-LEN, OUTPUT-REC-LEN.
IF END-REC
ELSE
IF INSERT-DONE EQUAL TO 1
ELSE
SUBTRACT 43 FROM LEAVING-REC-LEN
GIVING LEAVING-VAR-LEN.
IF LREC-DEPT GREATER THAN K22-DEPT
MOVE 1 TO INSERT-DONE
MOVE 43 TO RETURN-REC-LEN
MOVE K22-REC TO RETURN-REC
ELSE
END-IF
END-IF
GOBACK.
Figure 25. COBOL E35 Routine Example (VLR)
E15/E35 return codes and EXITCK

DFSORT's interpretation of E15 and E35 return codes depends upon whether
EXITCK=STRONG or EXITCK=WEAK is in effect. See the discussion of the
EXITCK option in OPTION control statement on page 173 for more information.
The following tables show the exact meaning of each E15 and E35 return code with
EXITCK=STRONG and EXITCK=WEAK in all possible situations.
Note:
1. You can determine whether EXITCK=STRONG or EXITCK=WEAK is in effect
from message ICE132I.
537

2. Use of EXITCK=WEAK can make it difficult to detect errors in the logic of your
E15 and E35 user exit routines.
3. EXITCK=WEAK is treated like EXITCK=STRONG if tape work data sets are
specified for a sort application or if the Blockset technique is not selected for a
merge application.
Table 72. E15 Without a SORTIN Data Set
Meaning with
EXITCK=STRONG
Meaning with
EXITCK=WEAK
Invalid
Do not return
Invalid
Do not return
Do not return
Do not return
12
Insert record
Insert record
16
Terminate DFSORT
Terminate DFSORT
20 (COBOL only)
Invalid
Do not return
All others
Invalid
Invalid
Table 73. E15 With a SORTIN Data Set Before End of Input
Meaning with EXITCK=STRONG or
EXITCK=WEAK
0
No action/record altered
Delete record
Do not return
12
Insert record
16
Terminate DFSORT
20 (COBOL only)
Alter/replace record
All others
Invalid
Table 74. E15 With a SORTIN Data Set After End of Input
Meaning with
EXITCK=STRONG
Meaning with
EXITCK=WEAK
Invalid
Do not return
Invalid
Do not return
Do not return
Do not return
12
Insert record
Insert record
16
Terminate DFSORT
Terminate DFSORT
20 (COBOL only)
Invalid
Do not return
All others
Invalid
Invalid
Table 75. E35 With a SORTOUT or OUTFIL Data Set Before End of Input
EXITCK=WEAK
538
No action/record altered
Delete record
Do not return

Table 75. E35 With a SORTOUT or OUTFIL Data Set Before End of Input (continued)
EXITCK=WEAK
12
Insert record
16
Terminate DFSORT
20 (COBOL only)
Alter/replace record
All others
Invalid
Table 76. E35 Without a SORTOUT or OUTFIL Data Set Before End of Input
Meaning with
EXITCK=STRONG
Meaning with
EXITCK=WEAK
Invalid
Delete record
Delete record
Delete record
Do not return
Do not return
12
Invalid
Delete record
16
Terminate DFSORT
Terminate DFSORT
20 (COBOL only)
Invalid
Delete record
All others
Invalid
Invalid
Table 77. E35 With a SORTOUT or OUTFIL Data Set After End of Input
Meaning with
EXITCK=STRONG
Meaning with
EXITCK=WEAK
Invalid
Do not return
Invalid
Do not return
Do not return
Do not return
12
Insert record
Insert record
16
Terminate DFSORT
Terminate DFSORT
20 (COBOL only)
Invalid
Do not return
All others
Invalid
Invalid
Table 78. E35 without a SORTOUT or OUTFIL Data Set After End of Input
Meaning with
EXITCK=STRONG
Meaning with
EXITCK=WEAK
Invalid
Do not return
Invalid
Do not return
Do not return
Do not return
12
Invalid
Do not return
16
Terminate DFSORT
Terminate DFSORT
20 (COBOL only)
Invalid
Do not return
All others
Invalid
Invalid
539
540
Chapter 6. Invoking DFSORT from a program

Invoking DFSORT dynamically
DFSORT can be invoked dynamically from programs written in COBOL or PL/I.
Specific information on dynamic invocation is covered in the COBOL and PL/I
programming guides. JCL requirements are the same as those for assembler.
This section explains what you need to know to initiate DFSORT from within your
assembler program using a system macro instruction instead of an EXEC job
control statement in the input stream. Specific restrictions on invoking DFSORT
from PL/I and COBOL are listed in Restrictions for dynamic invocation on page
561.
What are system macro instructions?

System macro instructions are macro instructions provided by IBM for
communicating service requests to the control program. You can use these
instructions only when programming in assembler language. They are processed
by the assembler program using macro definitions supplied by IBM and placed in
the macro library when your control program was installed.
You can specify one of three system macro instructions to pass control to the
program: LINK, ATTACH, or XCTL.
When you issue one of these instructions, the first load module of DFSORT is
brought into main storage. The linkage relationship between your program and
DFSORT differs according to which of the instructions you have used. For a
complete description of the macro instructions and how to use them, refer to z/OS
MVS Programming: Assembler Services Guide, z/OS MVS Programming: Assembler
Services Reference ABE-HSP, and z/OS MVS Programming: Assembler Services Reference
IAR-XCT.
Using system macro instructions

To initiate DFSORT processing with a system macro instruction, you must:
v Write the required job control language (JCL) DD statements.
v Write DFSORT control statements as operands of assembler DC instructions.
(Using DFSPARM or SORTCNTL data sets to specify program control statements
can be more convenient. See Chapter 3, Using DFSORT program control
statements, on page 81 for details.)
v Write a parameter list containing information to be passed to DFSORT and a
pointer containing the address of the parameter list. DFSORT accepts three types
of parameter lists: a 24-bit parameter list, an extended (31-bit) parameter list,
and a 64-bit parameter list. Although you can choose any one of the three
parameter lists, the 64-bit parameter list can perform a superset of the functions
of the other parameter lists; therefore, it is recommended for new DFSORT
applications.
Note: DFSORT can also be called using the system's EXEC PARM parameter list,
provided that the rules for passing it are followed (for example, the parameter
541
Using System Macro Instructions

list must reside below 16MB virtual). DFSORT interprets a call using the EXEC
PARM parameter list as a direct invocation rather than a program invocation.
v If you are using the 64-bit parameter list, prepare the macro instruction
specifying one of the following as the entry point name: ICEMAN64 or SORT64.
v If you are using the 31-bit or 24-bit parameter list, prepare the macro instruction
specifying one of the following as the entry point name: ICEMAN, SORT,
IERRCO00, or IGHRCO00.
In addition, the following rule applies:
v If you are invoking DFSORT repeatedly (for example, from an E15 or E35 user
exit), you must always wait for the last invoked sort to end before you can give
control back to any of your user exits in an earlier invoked sort.
Note: The save area passed to DFSORT must begin on a fullword boundary.
Using JCL DD statements

JCL DD statements are usually required when invoking DFSORT from another
program. The statements and their necessary parameters are described in detail.
Overriding DFSORT control statements from programs

You can override the control statements generated or passed by a program (for
example, a COBOL SORT verb or PLISRTx routine) with DFSORT's DFSPARM data
set.
For example, you could use the following to override the SORT statement
generated by a COBOL SORT verb in order to use DFSORT's Year 2000 features:
//DFSPARM DD
*
OPTION Y2PAST=1956
* set fixed CW of 1956-2055
SORT FIELDS=(11,6,Y2T,A, * sort Cyymmdd using CW
31,10,CH,A) * sort other key
/*
You can also use DFSPARM to provide different DFSORT control statements for
multiple invocations of DFSORT from a program. However, the control statements
must be located in temporary or permanent data sets and FREE=CLOSE must be
used. Here's an example of using DFSPARM to override the control statements for
a COBOL program with three SORT verbs:
DP1, DP2, and DP3 can contain any DFSORT control statements you like. The
statements in DP1 will be used for the first call to DFSORT, the statements in DP2
for the second and the statements in DP3 for the third.
For complete details on DFSPARM, see DFSPARM DD statement on page 76.
542
Overriding DFSORT Control Statements from Programs

COBOL also offers additional methods for overriding DFSORT parameters and
control statements such as SORT special registers and the IGZSRTCD data set. See
your COBOL publications for details.

Entry point name
When using the 24-bit parameter list, you must use ICEMAN, SORT, IERRCO00, or
IGHRCO00 as the entry point name for the LINK, ATTACH or XCTL macro.
Providing program control statements

When using the 24-bit parameter list, you must supply the starting and ending
address of a valid image of each control statement to be used during run-time. You
must provide the image as a character string in EBCDIC format using assembler
DC instructions. The rules for preparing the program control statements are as
follows:
v At least one control statement must be specifiedgenerally SORT or MERGE. If
more than 15 control statements are specified, only the first 15 control statements
are accepted and all others are ignored. Control statements can also be specified
in SORTCNTL or DFSPARM.
v The MODS statement is required when user exits other than E15, E32, and E35
are to be used. It is also required when the E15 or E35 routine addresses are not
passed by the parameter list.
v The following control statements can be passed using the 24-bit parameter list:
SORT or MERGE, RECORD, ALTSEQ, DEBUG, MODS, SUM, INREC, OUTREC,
INCLUDE or OMIT, JOINKEYS, JOIN, REFORMAT and OUTFIL.
v At least one blank must follow the operation definer (SORT, for example). A
control statement can start and end with one or more blanks; however, no other
blanks are allowed.
v The content and format of the statements are as described in Chapter 3, Using
DFSORT program control statements, on page 81, except:
Labels are not allowed although a leading blank is optional.
Because each control statement image must be defined contiguously by one or
more assembler DC instructions, explicit and implicit continuation of
statements is neither necessary nor allowed.
v Neither comment statements, blank statements, nor remark fields are permitted.
v If you use ATTACH to initiate the program, you cannot use the
checkpoint/restart facility and must not specify CKPT in the SORT statement
image.
For full override and applicability details, see Appendix B, Specification/
CONTROL statement images example

SORTBEG
SORTEND
DC
DC
C SORT FIELDS=(10,15,CH,A)
C
This form, with a trailing blank separately defined, allows you to refer to the last
byte of the statement (SORT statement end address) by the name SORTEND.
INCLBEG
INCLEND
DC
DC
C INCLUDE COND=(5,3,CH,NE,CJ82)
C
543
Invoking DFSORT with the 24-Bit Parameter List

Note: Assembler requires two single apostrophes to represent one single
apostrophe.
Format of the 24-bit parameter list

Figure 26 on page 545 shows the format of the 24-bit parameter list and the pointer
containing its address which you must pass to DFSORT. Detailed specifications for
each of the entries in the parameter list follow.
For full override and applicability details, see Appendix B, Specification/override
544
Register 1
X '80'
Offset
(Hex) (Dec)
Byte 1
Address of parameter list

Byte 2
Byte 3 and 4
-2
-2
Unused
2
6
2
6
X'00'
X'00'
Starting address of SORT or MERGE statement image

Ending address of SORT or MERGE statement image
1,3
1,5
A
E
10
14
X'00'
X'00'
Starting address of RECORD statement image (zeros, if none)

Ending address of RECORD statement image (zeros, if none)
1,3
1,5
12
18
X'00'
Address of E15 or E32 routine (zeros, if none)
16
22
X'00'
Address of E35 routine (zeros, if none)
1A
1E
26
30
X'02'
X'00'
Starting address of MODS statement image

Ending address of MODS statement image
2,3
2,5
22
34
X'00'
Main storage value
26
38
X'01'
Reserved storage value
2A
42
X'03'
Address of 8-character message ddname
2E
46
X'04'
Number of input files (MERGE with E32)
2,4
32
36
50
54
X'05'
X'00'
Starting address of DEBUG statement image

Ending address of DEBUG statement image
2,3
2,5
3A
3E
58
62
X'06'
X'00'
Starting address of ALTSEQ statement image

Ending address of ALTSEQ statement image
2,3
2,5
42
66
X'F6'
Address of 256-byte ALTSEQ translation table
46
70
X'F7'
2
2
Unused
Length of parameter list in bytes
Notes
4A
74
X'FD'
The three bytes after X'FD' are ignored
4E
78
X'FE'
Address of a pointer to 104-byte ESTAE work area

(or zeros)
52
82
X'FF'
Message option
56
86
4-character prefix for "SORT" DD statement names
5A
5E
90
94
X'07'
X'00'
Starting address of SUM statement image

Ending address of SUM statement image
2,3
2,5
62
66
98
102
X'08'
X'00'
Starting address of INCLUDE or OMIT statement image

Ending address of INCLUDE or OMIT statement image
2,3
2,5
6A
6E
106
110
X'09'
X'00'
Starting address of OUTREC statement image

Ending address of OUTREC statement image
2,3
2,5
72
76
114
118
X'0A'
X'00'
Starting address of INREC statement image

Ending address of INREC statement image
2,3
2,5
7A
7E
122
126
X'0B'
X'00'
Starting address of OUTFIL statement image

Ending address of OUTFIL statement image
2,3
2,5
82
86
130
134
X'0E'
X'00'
Starting address of JOINKEYS statement image

Ending address of JOINKEYS statement image
2,3
2,5
8A
8E
138
142
X'0F'
X'00'
Starting address of REFORMAT statement image

Ending address of REFORMAT statement image
2,3
2,5
92
96
146
150
X'10'
X'00'
Starting address of JOIN statement image

Ending address of JOIN statement image
2,3
2,5
Figure 26. The 24-Bit Parameter List
Notes to Figure 26:
545

1.
Required entry. Must appear in the relative position shown. The offset
shown is the actual offset of this entry.
2.
Optional entry. Can appear anywhere after the required entries. The
displayed offset is for identification purposes onlythe actual offset of this
entry can vary. Optional entries must be consecutive but can appear in any
order.
3.
A specific control statement. Shown for illustrative purposes only. SORT or

MERGE, RECORD, ALTSEQ, DEBUG, MODS, SUM, INREC, OUTREC,
INCLUDE or OMIT, JOINKEYS, JOIN, REFORMAT and OUTFIL can be
passed using any of the following hex entry codes: X'00' (see Note 1), X'02',
X'05' through X'0B', X'0E' through X'11', X'16', X'18' and X'20' through X'29'.
4.
Required entry if the MERGE statement is present and input is supplied

through an E32 user exit. This entry is not required if the FILES option of
the MERGE statement is specified.
5.
Required entry. Contains the ending address for a control statement and
must immediately follow the entry containing the starting address for that
same control statement.
The specifications for each of the parameter list entries follow:

Byte
Explanation
-2 to -1
Unused.
0 to +1
The byte count. This 2-byte field contains the length in bytes of the
parameter list. This two byte field is not included when counting the
number of bytes occupied by the list.
The total length of the required entries is 24 (X'0018'). All optional entries
are four bytes long except those referring to control statement images
which are eight bytes each.
546
2-5
The starting address of the SORT or MERGE statement image. Must be in

the last three bytes of this fullword. The first byte must contain X'00'.
6-9
The ending address of the SORT or MERGE statement image. Must be in

the last three bytes. The first byte must contain X'00'.
10-13
The starting address of the RECORD statement image, if any; otherwise, all
zeros. Must be in the last three bytes. The first byte must contain X'00'.
14-17
The ending address of the RECORD statement image, if any; otherwise, all
zeros. Must be in the last three bytes. The first byte must contain X'00'.
18-21
The address of the E15 or E32 routine that your program has placed in
main storage, if any; otherwise, all zeros. Must be in the last three bytes.
The first byte must contain X'00'.
22-25
The address of the E35 routine that your program has placed in main
storage, if any; otherwise, all zeros. Must be in the last three bytes. The
first byte must contain X'00'.
26-29
The starting address of the MODS statement image. Must be in the last
three bytes. The first byte must contain X'02'.
30-33
The ending address of the MODS statement. Must be in the last three
bytes. The first byte must contain X'00'.

34-37
Main storage value. The first byte must contain X'00'. The next three bytes
contain either the characters MAX or a hexadecimal value. You can use this
option to temporarily override the SIZE installation option. For full
override and applicability details, see Appendix B, Specification/override
of DFSORT options, on page 863. For an explanation of this value, see the
discussion of the MAINSIZE parameter in OPTION control statement on
page 173.
38-41
A reserved main storage value. The first byte must contain X'01'. The next
three bytes contain a hexadecimal value that specifies a number of bytes to
be reserved, where the minimum is 4K. For an explanation of this value,
see the explanation of the RESINV parameter in OPTION control
You can use this option to temporarily override the RESINV installation
option. For full override and applicability details, see Appendix B,
42-45
Message ddname. The first byte must contain X'03'. The next three bytes
contain the address of an 8-byte DD statement name for the message data
set, padded with blanks on the right if necessary. The name can be any
valid DD statement name, but must be unique.
You can use this option to temporarily override the MSGDDN installation
option. For full override details, see Appendix B, Specification/override of
DFSORT options, on page 863. For details on the use of the message data
set, see z/OS DFSORT Messages, Codes and Diagnosis Guide.
46-49
Number of input files to a merge. This entry is needed only if the MERGE
statement is present without the FILES option and input to the merge is
supplied through the E32 user exit. The first byte must contain X'04'. The
next three bytes contain the number of files in hexadecimal. For full
override and applicability details, see Appendix B, Specification/override
50-53
The starting address of the DEBUG statement image. Must be in the last
54-57
The ending address of the DEBUG statement image. Must be in the last
58-61
The starting address of the ALTSEQ statement image. Must be in the last
62-65
The ending address of the ALTSEQ statement image. Must be in the last
66-69
The address of a 256-byte translate table supplied instead of an ALTSEQ

statement. The first byte must contain X'F6'. If this parameter is present,
the X'06' parameter is ignored. For full override and applicability details,
see Appendix B, Specification/override of DFSORT options, on page 863.
70-73
User exit address constant. These 4 bytes are passed to E15 (at offset 4 in
the E15 parameter list), to E32 (at offset 8 in the E32 parameter list) or to
E35 (at offset 8 in the E35 parameter list) after DFSORT replaces the X'F7'
with X'00'.
Note: The user exit address constant must not be used for a Conventional
merge or tape work data set sort application.
74-77
X'FD' in the first byte (the VLSHRT option) specifies that DFSORT is to
continue processing if it finds a variable-length input record too short to
547

contain all specified control fields, compare fields, or summary fields. For
full details of this option, see the discussion of the VLSHRT parameter in
OPTION control statement on page 173. You can use this option to
temporarily override the VLSHRT installation option. For full override and
applicability details, see Appendix B, Specification/override of DFSORT
78-81
If the first byte contains X'FE', you can use the next three bytes to pass an
address of a 104-byte field save area where ESTAE information is saved.
These bytes must contain zeros if the ESTAE information is not saved.
If a system or user exit abend occurs, the DFSORT ESTAE recovery routine
will copy the first 104 bytes of the SDWA into this area before returning to
any higher level ESTAE recovery routines.
For more information on the DFSORT ESTAE recovery routine, see
Appendix E, DFSORT abend processing, on page 907
82-85
The message option. The first byte must contain X'FF'. The following three
bytes contain the characters NOF, (I), or (U). You can use this option to
temporarily override the MSGPRT installation option.
NOF
Messages and control statements are not printed. Critical messages are written
to the master console.
(I)
All messages except diagnostic messages (ICE800I to ICE999I) are printed.
Critical messages are also written to the master console. Control statements are
printed only if LIST is in effect.
(U)
Only critical messages are printed. They are also written to the master console.
Control statements are not printed (NOLIST is forced).
All messages are written to the message data set. For details on use of the message
data set, see z/OS DFSORT Messages, Codes and Diagnosis Guide For full override
and applicability details, see Appendix B, Specification/override of DFSORT
For compatibility reasons, the forms (NO, (AB, (AP, (AC, (CB, (CC, (CP, (PC, (SC,
and (SP are also accepted.
The following list shows the equivalent specifications for these aliases:
Table 79. Aliases for Message Option
548
MSGPRT
MSGCON
(NO
NONE
NONE
(AB
ALL
ALL
(AP
ALL
CRITICAL
(AC
NONE
ALL
(CB
CRITICAL
CRITICAL
(CC
NONE
CRITICAL
(CP
CRITICAL
CRITICAL
(PC
ALL
ALL
(SC
ALL
CRITICAL

Table 79. Aliases for Message Option (continued)
(SP
86-89
MSGPRT
MSGCON
CRITICAL
ALL
Four characters, which replace SORT in the following ddnames: SORTIN,

SORTOUT, SORTINn, SORTINnn, SORTOFd, SORTOFdd, SORTWKd,
SORTWKdd, SORTJNF1, SORTJNF2, and SORTCNTL. You must use this
option when you dynamically invoke DFSORT more than once in a
program step.
The four characters must all be alphanumeric or national ($, #, or @)
characters. The first character must be alphabetic, and the reserved names
DIAG, BALN, OSCL, POLY, CRCX, PEER, LIST, and SYSc (where c is any
alphanumeric character) must not be used. Otherwise, the four characters
are ignored.
For example, if you use ABC# as replacement characters, DFSORT uses
statements ABC#IN, ABC#CNTL, ABC#WKdd, and ABC#OUT instead of
SORTIN, SORTCNTL, SORTWKdd, and SORTOUT.
Note: This parameter is equivalent to the SORTDD=cccc run-time option.
90-93
The starting address of the SUM statement image. Must be in the last three
94-97
The ending address of the SUM statement image. Must be in the last three
98-101
The starting address of the INCLUDE or OMIT statement image. Must be
in the last three bytes. The first byte must contain X'08'.
102-105
The ending address of the INCLUDE or OMIT statement image. Must be
in the last three bytes. The first byte must contain X'00'.
106-109
The starting address of the OUTREC statement image. Must be in the last
110-112
The ending address of the OUTREC statement image. Must be in the last
114-116
The starting address of the INREC statement image. Must be in the last
three bytes. The first byte must contain X'0A'.
118-121
The ending address of the INREC statement image. Must be in the last
122-125
The starting address of the OUTFIL statement image. Must be in the last
three bytes. The first byte must contain X'0B'.
126-129
The ending address of the OUTFIL statement image. Must be in the last
549

130-133
The starting address of the JOINKEYS statement image. Must be in the last
three bytes. The first byte must contain X'0E'.
134-137
The ending address of the JOINKEYS statement image. Must be in the last
138-141
The starting address of the REFORMAT statement image. Must be in the
last three bytes. The first byte must contain X'0F'.
142-145
The ending address of the REFORMAT statement image. Must be in the
last three bytes. The first byte must contain X'00'.
146-149
The starting address of the JOIN statement image. Must be in the last three
150-153
The ending address of the JOIN statement image. Must be in the last three
Invoking DFSORT with the extended (31-bit) parameter list

Entry point name
When using the extended parameter list, you must use ICEMAN, SORT,
IERRCO00, or IGHRCO00 as the entry point name for the LINK, ATTACH or
XCTL macro.

When using the extended parameter list, the control statements are written in a
single area to which the parameter list points. The control statement area consists
of:
v A 2-byte field containing the length (in binary) of the character string to follow.
v A character string in EBCDIC format using assembler DC instructions and
containing valid images of the control statements to be used during run-time.
The rules for preparing the program control statements are:
v The control statements must be separated by one or more blanks. A blank
preceding the first statement is optional; however, a trailing blank is required.
No labels, comment statements, or remark fields are allowed. Because each
control statement image must be defined contiguously by one or more assembler
DC instructions, explicit and implicit continuation of statements is not necessary
or allowed.
v The MODS statement is required when user exits other than E15, E18, E32, E35,
and E39 are to be used or when the E15, E18, E35, or E39 routine addresses are
not passed by the parameter list.
v All of the control statements described in Chapter 3, Using DFSORT program
control statements, on page 81 can be specified. None is required, but SORT,
MERGE, or OPTION COPY must be specified in the parameter list, SORTCNTL,
or DFSPARM.
550
Invoking DFSORT With The Extended Parameter List

checkpoint/restart facility. Do not specify CKPT on the SORT or OPTION
statement.
Format of the extended parameter list

Figure 27 shows the format of the extended parameter list and the pointer, which
you must pass to DFSORT, containing its address.
The first parameter must be specified. A 4-byte field containing X'FFFFFFFF' must
be used to indicate the end of the parameter list. It can be coded anywhere after
the first parameter.
If a parameter is specified, it must appear in the indicated position and must
contain a 31-bit address or a clean (the first 8 bits containing zeros) 24-bit address.
If a parameter is not specified, it is treated as if it were specified as zeros. For full
override and applicability details, see Appendix B, Specification/override of
DFSORT options, on page 863.
Figure 27. The Extended Parameter List
Detailed specifications for each of the entries in the parameter list follow:
Byte
Explanation
0-3
Required. The address of the area containing the DFSORT control

statements, if any; otherwise, all zeros. The high order bit must be 0 to
identify this as an extended parameter list.
Refer to the previous section for the format of the control statement area.
Attention: The area must start with a two-byte length field.
551

If you specify this parameter as zeros, you must supply all the required
control statements in DFSPARM or SORTCNTL.
4-7
Optional. The address of the E15 or E32 user exit routine that your
program has placed in main storage (for example, via LOAD), if any;
otherwise, all zeros.
f (bit 0) has the following meaning:
v 0 = Enter the user exit with 24-bit addressing in effect (AMODE 24).
Note: If the Blockset or Peerage/Vale technique is not selected, the user
exit is always entered with 24-bit addressing in effect (AMODE 24).
8-11
Optional. The address of the E35 user exit routine that your program has
placed in main storage (for example, via LOAD), if any; otherwise, all
zeros.
12-15
Optional. This field will be passed to the E15, E32 or E35 user exit routines.
16-19
Optional. The address of a 256-byte ALTSEQ translation table supplied

instead of an ALTSEQ statement, if any; otherwise, all zeros. You can use
this option to override an ALTSEQ installation option. For full override
and applicability details, see Appendix B, Specification/override of
DFSORT options, on page 863.
20-23
Optional. The address of a 4-byte field containing the address of a 112-byte

work area where ESTAE information is saved, or all zeros if the ESTAE
information is not saved.
If a system or user exit abend occurs, the DFSORT recovery routine will
copy the first 112 bytes of the software diagnostic work area (SDWA) into
this area before returning to your ESTAE recovery routine.
24-27
zeros.
Note: This parameter is ignored for a merge application and for a tape
work data set sort application.
552

28-31
zeros.
Note: This parameter is ignored for a conventional merge application and
for a tape work data set sort application.
exit is always entered with 24-bit addressing if effect (AMODE 24).
32-35
Optional. 4 characters to be used as an identifier for this call to DFSORT.

This field can be used to uniquely identify each call to DFSORT from a
program that calls DFSORT more than once. DFSORT prints message
ICE200I to display the field identifier exactly as you specify it; the field is
not checked for valid characters.
If the field identifier is specified, it must appear in the indicated position.
If the identifier field contains zeros (X'00000000'), or X'FFFFFFFF' is used to
end the parameter list before or at the field identifier, DFSORT does not
print message ICE200I.
Note: The list can be ended after any parameter. The last parameter in the list
must be followed by X'FFFFFFFF'.

Entry point name
When using the 64-bit parameter list, you must use ICEMAN64 or SORT64 as the
entry point name for the LINK, ATTACH or XCTL macro.
64-bit address terminology

When referring to 64-bit addresses:
by a 24-bit address in the low-order 3 bytes, that is, X0000000000aaaaaa
by a 31-bit address in the low-order 4 bytes, that is, X00000000aaaaaaaa where
Bit 32 of the 31-bit address must be 0.
v A 64 bit address uses all 64-bits for the address.
The following summarizes the previous information:
Table 80. Summary of 64-bit Terminology for Programs
Address
Bits 0-31
Bit 32
Bits 33-39
Bits 40-63
24-bit
Address
31-bit
64-bit
Address
Address
553
Invoking DFSORT With The 64-Bit Parameter List

When using the 64-bit parameter list, the control statements are written in a single
area to which the parameter list points. The control statement area can be above or
below the bar. The control statement area consists of:
v A 2-byte field containing the length (in binary) of the character string to follow.
v A character string in EBCDIC format using assembler DC instructions and
containing valid images of the control statements to be used during run-time.
The rules for preparing the program control statements are:
v The control statements must be separated by one or more blanks. A blank
preceding the first statement is optional; however, a trailing blank is required.
No labels, comment statements, or remark fields are allowed. Because each
control statement image must be defined contiguously by one or more assembler
DC instructions, explicit and implicit continuation of statements is not necessary
or allowed.
v The MODS statement is required when user exits other than E15, E18, E32, E35,
and E39 are to be used or when the E15, E18, E35, or E39 routine addresses are
not passed by the parameter list.
v All of the control statements described in Chapter 3, Using DFSORT program
control statements, on page 81 can be specified. None is required, but SORT,
MERGE, or OPTION COPY must be specified in the parameter list, SORTCNTL,
or DFSPARM.
checkpoint/restart facility. Do not specify CKPT on the SORT or OPTION
statement.
Format of the 64-bit parameter list

Figure 28 on page 555 shows the format of the 64-bit parameter list. As indicated,
some 64-bit fields must contain zeros in the first 32 bits as shown in Figure 28 on
page 555. You must pass the pointer to the parameter list in 64-bit register 1.
which provides a separate Assembler DSECT for the 64-bit invocation parameter
list.
The 8-character identifier 'PL64SORT' must be present at the start of the parameter
list. The list must be 136 bytes long with parameters or zeros specified in the
indicated positions. Each parameter is 64-bits wide (8 bytes) and must contain a
64-bit address or a clean 31-bit address as indicated in the description.
554
Figure 28. The 64-Bit Parameter List
Detailed specifications for each of the entries in the parameter list follow:
555

Byte (dec)
Explanation
0-7
C'PL64SORT' identifier used to indicate the 64-bit invocation parameter list

has been specified.
8
Bit 0
Indicates 24-bit addressing mode (AMODE 24) will be used when

the E15 or E32 exit is entered.
v 0 = Do not enter the E15 or E32 with 24-bit addressing in effect.
v 1 = Enter the E15 or E32 with 24-bit addressing in effect.
Bit 1

Bit 2

Note: If an E15 or E32 is used, only one of the E15 or E32 AMODE
bits described previously can be set. If more than one E15 AMODE
bit is set, DFSORT will terminate.
Bit 3

the E35 exit is entered.
v 0 = Do not enter the E35 with 24-bit addressing in effect.
v 1 = Enter the E35 with 24-bit addressing in effect.
Bit 4

Bit 5

Note: If an E35 is used, only one of the previous E35 AMODE bits
can be set. If more than one E35 AMODE bit is set, DFSORT will
terminate.
Bit 6

Bit 7

556

terminate.
9
Bit 0
Reserved. Must be set to zero.
Bit 1

v 0 = Do not enter the E39 with 24-bit addressing in effect
Bit 2

v 0 = Do not enter the E39 with 31-bit addressing in effect
terminate.
Bit 3
Reserved. Must be set to zero.
Bit 4
Indicates type of E15 or E32 exit parameter list used

v 0 = 32 bit exit parameter list used, 32-bit Register 1 returned
from exit, and exit saves 32-bit registers in save area (Format-0
72 byte save area) pointed at by 32-bit Register 13.
Bit 5
Indicates type of E35 exit parameter list used

Bit 6

v 1 = Reserved.
Bit 7

v 1 = Reserved.
Note: In the previous situations when a 32-bit Register 1 is
returned, the high half of Register 1 will be ignored by DFSORT
when the exit returns to DFSORT.
10-23
Reserved. Must be set to zeros.
557

24-31
The address of the area containing the DFSORT control statements, if any;
otherwise, all zeros. Can be a 64-bit address, a clean 31-bit address or a
clean 24-bit address.
Refer to the previous section for the format of the control statement area.
Attention: The area must start with a two-byte length field. If zeros are
specified, all the required control statements must be provided in DFSORT
DD DFSPARM or SORTCNTL.
32-39
The address of the E15 or E32 user exit routine that the program has
zeros. Must be a clean 31-bit address or a clean 24-bit address.
40-47
The address of the E35 user exit routine that the program has placed in
main storage (for example, via LOAD), if any; otherwise, all zeros. Must be
48-55
The 4-byte user exit address constant will be passed to the E15, E32 or E35
user exit routines.
56-63
The address of a 256-byte ALTSEQ translation table supplied instead of an

ALTSEQ statement, if any; otherwise, all zeros. It can be a 64-bit address, a
clean 31-bit address or a clean 24-bit address. You can use this option to
override an ALTSEQ installation option. For full override and applicability
details, see Appendix B, Specification/override of DFSORT options, on
page 863.
64-71
The address of a 112-byte work area where ESTAE information is saved, or

all zeros if the ESTAE information is not saved. Must be a clean 31-bit
address or a clean 24-bit address. If a system or user exit abend occurs, the
DFSORT recovery routine will copy the first 112 bytes of the software
diagnostic work area (SDWA) into this area before returning to the ESTAE
recovery routine.
72-79
Note: This parameter is ignored for a merge application and for a tape
work data set sort application.
80-87
Note: This parameter is ignored for a conventional merge application and
for a tape work data set sort application.
88-95
Four bytes of zeros followed by 4 characters to be used as an identifier for

this call to DFSORT. This field can be used to uniquely identify each call to
DFSORT from a program that calls DFSORT more than once. DFSORT
prints message ICE200I to display the 4-character identifier exactly as you
specify it; the field is not checked for valid characters. If the identifier field
contains zeros (X'00000000'), DFSORT does not print message ICE200I.
96-135 Reserved. Must be set to zeroes.
558
Writing The Macro Instruction
Writing the macro instruction

When writing the LINK, ATTACH, or XCTL macro instruction using the 64-bit
parameter list, you must:
v Specify SORT64 (the entry point) in the EP parameter of the instruction. (This
applies to sort, merge, and copy applications.)
v Load the address of the pointer to the parameter list into 64-bit register 1 or pass
it in the MF parameter of the instruction. The parameter list can be above or
below the bar.
When writing the LINK, ATTACH, or XCTL macro instruction using the extended
or 24-bit parameter list, you must:
v Specify SORT (the entry point) in the EP parameter of the instruction. (This
applies to sort, merge, and copy applications.)
v Load the address of the pointer to the parameter list into register 1 or pass it in
the MF parameter of the instruction. The parameter list must be below the bar.
Note: If you are using ATTACH, you might also need the ECB parameter.
If you provide an E15 user exit routine address in the parameter list, DFSORT
ignores the SORTIN data set; your E15 routine must pass all input records to
DFSORT. The same applies to a merge if you specify an E32 routine address. This
means that your routine must issue a return code of 12 (insert record) until the
input data set is complete, and then a return code of 8 (do not return).
DFSORT ignores the SORTOUT data set if you provide an E35 routine address in
the parameter list. Unless you use OUTFIL processing, your routine is then
responsible for disposing of all output records. It must issue a return code of 4
(delete record) for each record in the output data set. When the program has
deleted all the records, your routine issues a return code of 8 (do not return).
When DFSORT is done, it passes control to the routine that invoked it.
When a single task attaches two or more program applications, you must modify
the standard ddnames so that they are unique. For ways of doing this, and for the
rules of override, see Appendix B, Specification/override of DFSORT options, on
page 863.
If you ATTACH more than one DFSORT application from the same program, you
must wait for each to complete before attaching the next unless DFSORT and your
user exits are installed re-entrant.
When you initiate DFSORT via XCTL, you must give special consideration to the
area where the parameter list, address list, optional parameters, and modification
routines (if any) are stored. This information must not reside in the module that
issues the XCTL because the module is overlaid by DFSORT.
There are two ways to overcome this problem. First, the control information can
reside in a task that attaches the module that issues the XCTL. Second, the module
issuing the XCTL can first issue a GETMAIN macro instruction and place the
control information in the main storage area it obtains. This area is not overlaid
when the XCTL is issued. The address of the control information in the area must
be passed to DFSORT in general register 1.
559
Parameter list examples

24-Bit Parameter List Example
The example in Figure 29 shows, in assembler language, how to use a 24-bit
parameter list to code parameters and statement images and how to pass control to
DFSORT.
LA 1,PARLST
LOAD ADDR OF PARAM POINTER IN R1
ATTACH EP=SORT
INVOKE SORT
.
.
.
PARLST DC X80,AL3(ADLST)
POINTER FLAG/ADDRESS OF PARAM LIST
.
.
.
CNOP 2,4
ALIGN TO CORRECT BOUNDARY
ADLST DC AL2(LISTEND-LISTBEG) PARAM LIST LENGTH
LISTBEG DC A(SORTA)
BEGINNING ADDRESS OF SORT STMT
DC A(SORTZ)
END ADDRESS OF SORT STMT
DC A(RECA)
BEGINNING ADDR OF RECORD STMT
DC A(RECZ)
END ADDR OF RECORD STMT
DC A(MOD1)
ADDR OF E15 RTN
DC A(MOD2)
ADDR OF E35 RTN
DC CABC#
DDNAME CHARACTERS
DC F720000
OPTIONAL MAIN STORAGE VALUE
DC XFF
MESSAGE OPTION FLAG BYTE
DC C(U)
MESSAGE OPTION
LISTEND EQU *
SORTA DC C SORT FIELDS=(10,15,CH,A), SORT CONTROL STMT
DC CFILSZ=E4780
(CONTINUED)
SORTZ DC C
DELIMITER
RECA
DC C RECORD LENGTH=100,TYPE=F
RECORD CONTROL STMT
RECZ
DC C
DELIMITER
DS 0H
USING *,15
MOD1
(routine for E15 user exit)
.
.
USING *,15
MOD2
(routine for E35 user exit)
Figure 29. Coding a 24-Bit Parameter List
Extended Parameter List Example

The example in Figure 30 on page 561 shows, in assembler language, how to use
an extended parameter list to code parameters and statement images and how to
pass control to DFSORT.
560
.
.
.
LA
R1,PL1
ST
R2,PL4
*
*
SET ADDRESS OF PARAMETER LIST

TO BE PASSED TO SORT/MERGE
SET ADDRESS OF GETMAINED AREA
TO BE PASSED TO E15
INVOKE SORT/MERGE
LINK EP=SORT
.
.
.
PL1
DC
A(CTLST)
ADDRESS OF CONTROL STATEMENTS
PL2
DC
A(E15)
ADDRESS OF E15 ROUTINE
PL3
DC
A(0)
NO E35 ROUTINE
PL4
DS
A
USER EXIT ADDRESS CONSTANT
PL5
DC
F-1
INDICATE END OF LIST
CTLST DS
0H
CONTROL STATEMENTS AREA
DC
AL2(CTL2-CTL1)
LENGTH OF CHARACTER STRING
CTL1
DC
C SORT FIELDS=(4,5,CH,A)
DC
C OPTION
DC
CRESINV=2048,FILSZ=E25000,MSGDDN=MSGOUT
DC
C OMIT COND=(5,8,EQ,13,8),FORMAT=FI
DC
C RECORD TYPE=F,LENGTH=80
CTL2
EQU *
OUT
DCB DDNAME=SYSOUT,... MYSORT USES SYSOUT
E15
DS
0H
E15 ROUTINE
.
.
.
BR
R14
RETURN TO SORT/MERGE
* MAPPING OF PARAMETER LIST PASSED TO E15 FROM SORT/MERGE
SRTLST DS
A
ADDRESS OF RECORD
GMA
DS
A
ADDRESS OF AREA GETMAINED BY
*
MYSORT
.
.
.
Figure 30. Coding an Extended Parameter List
64-Bit Parameter List Example

Example 15. Sort with 64-bit parameter lists, E15, E35 and OUTFIL on page 839
in Chapter 11, Examples of DFSORT job streams, on page 819 shows, in
assembler language, how to use a 64-bit parameter list to code parameters and
statement images and how to pass control to DFSORT.
Restrictions for dynamic invocation

Merge restriction
Merge applications cannot be done when DFSORT is invoked from a PL/I
program.
Copy restrictions
Copy applications cannot be done when DFSORT is invoked from a PL/I program.
If you invoke DFSORT from a COBOL program, the following restrictions apply:
v The OPTION COPY statement can be placed in either the COBOL IGZSRTCD
data set or the DFSORT SORTCNTL or DFSPARM data set.
561
Restrictions for Dynamic Invocation

v If using the FASTSRT compiler option for any part or all of the COBOL SORT
statement, a copy application can be done.
v If using the COBOL MERGE statement, a copy application cannot be done.
See COBOL requirements for copy processing on page 522 for user exit
requirements.
562
Chapter 7. Using ICETOOL

Overview
This chapter describes ICETOOL, a multi-purpose DFSORT utility. ICETOOL uses
the capabilities of DFSORT to perform multiple operations on one or more data
sets in a single job step. These operations include the following:
v Creating multiple copies of sorted, merged, edited, or unedited input data sets
v Creating output data sets containing subsets of input data sets based on various
criteria for character and numeric field values, the number of times unique
values occur, header records, trailer records or relative record numbers
v Sorting records between headers and trailers
v Creating output data sets containing different field arrangements of input data
sets
v Resizing fixed length records by creating a larger record from multiple shorter
records, or multiple shorter records from a larger record
v Creating list data sets showing character and numeric fields in a variety of
simple, tailored, and sectioned report formats, allowing control of title, date,
time, page numbers, carriage control characters, headings, lines per page, field
formats, and total, maximum, minimum, average and count values for the
columns of numeric data
v Printing messages that give statistical information for selected numeric fields
such as minimum, maximum, average, total, count of values, and count of
unique values
v Printing messages that identify invalid decimal values
v Printing messages that give record counts
v Setting RC=12, RC=8, RC=4, or RC=0 based on record counts
v Creating an output data set containing an output record with text and the record
count
v Creating a list data set showing the DFSORT installation defaults in use
v Creating list data sets showing unique values for selected character and numeric
fields and the number of times each occurs, in a variety of simple and tailored
report formats
v Creating list and output data sets for records with: duplicate values,
non-duplicate values, or values that occur n times, less than n times or more
than n times
v Creating output data sets with information spliced together from two or more
input records with duplicate values. The information in the input records can
originate from different data sets, helping you to perform various file "join" and
"match" operations.
v Using three different modes (stop, continue, and scan) to control error checking
and actions after error detection for groups of operators.
You can use ICETOOL for SORT, MERGE and COPY operations.
ICETOOL/DFSORT relationship
ICETOOL is a batch front-end utility that uses the capabilities of DFSORT to
perform the operations you request.
563
Overview
ICETOOL includes 17 operators that perform sort, merge, copy, statistical, report,
selection, and splice operations. Most of the operations performed by ICETOOL
require only simple JCL and operator statements. Some ICETOOL operations
require or allow you to specify complete DFSORT control statements (such as
SORT, MERGE, INCLUDE or OMIT, INREC, and OUTFIL) to take full advantage
of DFSORT's capabilities.
ICETOOL automatically calls DFSORT with the particular DFSORT control
statements and options required for each operation (such as DYNALLOC for
sorting).
ICETOOL also produces messages and return codes describing the results of each
operation and any errors detected. Although you generally do not need to look at
the DFSORT messages produced as a result of an ICETOOL run, they are available
in a separate data set if you need them.
ICETOOL can be called directly or from a program. ICETOOL allows operator
statements (and comments) to be supplied in a data set or in a parameter list
passed by a calling program. For each operator supplied in the parameter list,
ICETOOL puts information in the parameter list pertaining to that operation, thus
allowing the calling program to use the information derived by ICETOOL.
ICETOOL JCL summary

The JCL statements used with ICETOOL are summarized in the following. See Job
control language for ICETOOL on page 571 for more detailed information. See
also JCL restrictions on page 573 and ICETOOL notes and restrictions on page
727.
//JOBLIB DD
Defines your program link library if it is not already known to the system.
//STEPLIB DD
Same as //JOBLIB DD
//TOOLMSG DD
Defines the ICETOOL message data set for all operations.
//DFSMSG DD
Defines the DFSORT message data set for all operations.
//SYMNAMES DD
symbol processing.
//SYMNOUT DD
table are to be listed.
//TOOLIN DD
Contains ICETOOL control statements.
//indd DD
Defines an input data set for a COPY, COUNT, DATASORT, DISPLAY,
MERGE, OCCUR, RANGE, RESIZE, SELECT, SORT, SPLICE, STATS,
SUBSET, UNIQUE, or VERIFY operation.
564
Overview
//outdd DD
Defines an output data set for a COPY, DATASORT, MERGE, RESIZE,
SELECT, SORT, SPLICE, or SUBSET operation.
//savedd DD
Defines an output data set for a SELECT or SUBSET operation.
//listdd DD
Defines a list data set for a DEFAULTS, DISPLAY, or OCCUR operation.
//countdd DD
Defines an output data set for a COUNT operation
//xxxxCNTL DD
Contains DFSORT control statements for a COPY, COUNT, DATASORT,
MERGE, RESIZE, SELECT, SORT, SPLICE or SUBSET operation.
ICETOOL operator summary

ICETOOL has 17 operators that are used to perform a variety of functions. The
functions of these operators are summarized later in this section. See ICETOOL
statements on page 574 for more detailed information. Additionally, information
pertaining to each operator is provided to calling programs that supply statements
to ICETOOL using a parameter list. See Parameter list interface on page 722 for
details.
COPY
Copies a data set to one or more output data sets.
COUNT
Prints a message containing the count of records in a data set. COUNT can
also be used to create an output data set containing text and the count, or to
set RC=12, RC=8, RC=4, or RC=0 based on meeting criteria for the number of
records in a data set.
DATASORT
Sorts data records between header and trailer records in a data set to an output
data set.
DEFAULTS
Prints the DFSORT installation defaults in a separate list data set.
DISPLAY
Prints the values or characters of specified numeric or character fields in a
separate list data set. Simple, tailored, or sectioned reports can be produced.
Maximums, minimums, totals, averages and counts can be produced.
MERGE
Merges one or more data sets to one or more output data sets.
MODE
Three modes are available that can be set or reset for groups of operators:
v STOP mode (the default) stops subsequent operations if an error is detected
v CONTINUE mode continues with subsequent operations if an error is
detected
v SCAN mode allows ICETOOL statement checking without actually
performing any operations.
OCCUR
Prints each unique value for specified numeric or character fields and how
many times it occurs in a separate list data set. Simple or tailored reports can
565
Overview
be produced. The values printed can be limited to those for which the value
count meets specified criteria (for example, only duplicate values or only
non-duplicate values).
RESIZE
Creates a larger record from multiple shorter records, or creates multiple
shorter records from a larger record, that is, resizes fixed length records.
RANGE
Prints a message containing the count of values in a specified range for a
specified numeric field in a data set.
SELECT
Selects records from a data set for inclusion in an output data set based on
meeting criteria for the number of times specified numeric or character field
values occur (for example, only duplicate values or only non-duplicate values).
Records that are not selected can be saved in a separate output data set.
SORT
Sorts a data set to one or more output data sets.
SPLICE
Splices together specified fields from records that have the same specified
numeric or character field values (that is, duplicate values), but different
information. Specified fields from two or more records can be combined to
create an output record. The fields to be spliced can originate from records in
different data sets, so you can use SPLICE to do various "join" and "match"
operations.
STATS
Prints messages containing the minimum, maximum, average, and total for
specified numeric fields in a data set.
SUBSET
Selects records from a data set based on keeping or removing header records
(the first n records), relative records, or trailer records (the last n records).
Records that are not selected can be saved in a separate output data set.
UNIQUE
Prints a message containing the count of unique values for a specified numeric
or character field.
VERIFY
Examines specified decimal fields in a data set and prints a message
identifying each invalid value found for each field.
Restriction: You can perform a JOINKEYS application with the COPY and SORT
operators, but not with the other operators.
Complete ICETOOL examples

ICETOOL example on page 850 contains a complete ICETOOL sample job with
all required JCL and control statements. The following example shows the JCL and
control statements for a simple ICETOOL job.
566
Overview
//EXAMP JOB A402,PROGRAMMER

//RUNIT EXEC PGM=ICETOOL,REGION=1024K
//TOOLIN DD *
* Show installation defaults
DEFAULTS LIST(SHOWDEF)
* Create three copies of a data set
COPY FROM(IN1) TO(OUT1,OUT2,OUT3)
* Print a report
DISPLAY FROM(IN2) LIST(REPORT) DATE TITLE(Monthly Report) PAGE HEADER(Location) ON(1,25,CH) HEADER(Revenue) ON(23,10,FS) HEADER(Profit) ON(45,10,FS) TOTAL(Totals) AVERAGE(Averages) BLANK
* Select all records with duplicate (non-unique) keys
SELECT FROM(IN2) TO(DUPKEYS) ON(1,25,CH) ALLDUPS * Save all records with non-duplicate (unique) keys
DISCARD (UNQKEYS)
/*
//SHOWDEF DD SYSOUT=A
//IN1 DD DSN=FLY.INPUT1,DISP=SHR
//IN2 DD DSN=FLY.INPUT2,DISP=SHR
//OUT1 DD DSN=FLY.NEW,DISP=OLD
//OUT2 DD DSN=FLY.BU1,DISP=OLD
//OUT3 DD DSN=FLY.BU2,DISP=OLD
//DUPKEYS DD DSN=FLY.DUPS,DISP=OLD
//UNQKEYS DD DSN=FLY.UNQS,DISP=OLD
//REPORT DD SYSOUT=A
Figure 31. Simple ICETOOL Job
Using symbols
You can define and use a symbol for any field or constant in the following
ICETOOL operators: COUNT, DATASORT, DISPLAY, OCCUR, RANGE, SELECT,
SPLICE, STATS, SUBSET, UNIQUE, and VERIFY. You can also use symbols in the
DFSORT control statements you specify for an ICETOOL run. This makes it easy to
create and reuse collections of symbols (that is, mappings) representing
information associated with various record layouts. You can use system symbols
(for example, &JOBNAME.) in your symbol constants. See Chapter 8, Using
symbols for fields and constants, on page 731 for complete details.
Using SET and PROC symbols in control statements

For jobs that directly invoke ICETOOL (PGM=ICETOOL), you can construct
DFSORT symbols that incorporate JCL PROC or SET symbols as well as text and
system symbols. You can then use those constructed DFSORT symbols in ICETOOL
and DFSORT control statements in the same way you use regular DFSORT
symbols. For complete information, see Chapter 8, Using symbols for fields and
constants, on page 731.
Invoking ICETOOL
ICETOOL can be invoked in the following three ways:
v Directly (that is, not from a program) using the TOOLIN Interface
v From a program using the TOOLIN Interface
v From a program using the Parameter List Interface.
567
Overview
With the TOOLIN Interface, you supply ICETOOL statements in a data set defined
by the TOOLIN DD statement. ICETOOL prints messages in the data set defined
by the TOOLMSG DD statement.
With the Parameter List Interface, your program supplies ICETOOL statements in a
parameter list. ICETOOL prints messages in the data set defined by the TOOLMSG
DD statement and also puts information in the parameter list for use by your
program.
Putting ICETOOL to use

By using various combinations of the 17 ICETOOL operators, you can easily create
applications that perform many complex tasks. The two small samples that follow
show some things you can do with ICETOOL.
Obtaining various statistics

Table 81. Obtaining Various Statistics
MODE STOP
VERIFY FROM(DATA1) ON(22,7,PD)
DISPLAY FROM(DATA1) LIST(SALARIES) TITLE(Employee Salaries) DATE TIME HEADER(Employee Name) HEADER(Salary) ON(1,20,CH)
ON(22,7,PD)
BLANK AVERAGE(Average Salary)
STATS FROM(DATA1) ON(22,7,PD)
RANGE FROM(DATA1) ON(22,7,PD) LOWER(20000)
RANGE FROM(DATA1) ON(22,7,PD) HIGHER(19999) LOWER(40000)
RANGE FROM(DATA1) ON(22,7,PD) HIGHER(40000)
OCCUR FROM(DATA1) LIST(SALARIES) TITLE(Employees Receiving Each Salary) DATE TIME HEADER(Salary) HEADER(Employee Count) ON(22,7,PD)
ON(VALCNT)
BLANK
Assume that you specify DD statements with the following ddnames for the
indicated data sets:
DATA1
A data set containing the name, salary, department, location and so on, of
each of your employees. The name field is in positions 1 through 20 in
character format and the salary field is in positions 22 through 28 in
SALARIES
A SYSOUT data set.
You can use the ICETOOL operators in Table 81 to do the following:
MODE STOP
If an error is found while processing one of the operators, subsequent
operators are not processed (that is, each operator is dependent on the
success of the previous operator).
VERIFY
Prints error messages in the TOOLMSG data set identifying any invalid
values in the packed decimal salary field.
DISPLAY
568
Overview
Prints a report with each employee's name and salary and the average for
all employee salaries in the SALARIES list data set.
STATS
Prints messages in the TOOLMSG data set showing the minimum,
maximum, average, and total of the individual salaries.
RANGE
The three RANGE operators print messages in the TOOLMSG data set
showing the number of salaries below $20,000, from $20,000 to $39,999, and
above $40,000.
OCCUR
Prints a report with each unique salary and the number of employees who
receive it in the SALARIES list data set.
Creating multiple versions/combinations of data sets

Table 82. Creating Multiple Versions/Combinations of Data Sets
* GROUP 1
MODE CONTINUE
COPY FROM(DATA1) TO(DATA2)
COPY FROM(MSTR1) TO(MSTR2)
SELECT FROM(DATA1) TO(SMALLDPT) ON(30,4,CH) LOWER(10)
UNIQUE FROM(MSTR1) ON(30,4,CH)
* GROUP 2
MODE STOP
COPY FROM(DATA1) TO(TEMP1) USING(NEW1)
SORT FROM(CONCAT) TO(FINALD,FINALP) USING(FINL)
Assume that you specify DD statements with the following ddnames for the
indicated data sets:
DATA1
A data set containing the name, salary, department, location, and so on, of each
of your employees. The department field is in positions 30 through 33 in
character format.
MSTR1
Master data set containing only the name and department of each of your
employees. The department field is in positions 30 through 33 in character
format.
DATA2, MSTR2, and SMALLDPT
Permanent data sets.
NEW1CNTL
A data set containing DFSORT control statements to INCLUDE employees in
department X100 and change the records to match the format of MSTR1.
NEW2CNTL
Same as NEW1CNTL but for department X200.
NEW3CNTL
Same as NEW1CNTL but for department X300.
569
Overview
TEMP1, TEMP2, and TEMP3
Temporary data sets.
FINLCNTL
A data set containing a DFSORT control statement to sort by department and
employee name.
CONCAT
A concatenation of the TEMP1, TEMP2, TEMP3, and MSTR1 data sets.
FINALD
A permanent data set.
FINALP
A SYSOUT data set.
You can use the ICETOOL operators in Table 82 on page 569 to do the following:
MODE CONTINUE
If an error is found while processing any of the group 1 operators,
subsequent group 1 operators are still processed; that is, group 1 operators
are not dependent on the success of the previous group 1 operators.
COPY
The two copy operators create backup copies of DATA1 and MSTR1.
SELECT
Creates a permanent output data set containing the name, salary,
department, location, and so on, of each employee in departments with
less than 10 people.
UNIQUE
Prints a message in the TOOLMSG data set showing the number of unique
departments.
MODE STOP
If an error is found while processing one of the group 2 operators,
subsequent group 2 operators are not processed; that is, each group 2
operator is dependent on the success of previous group 2 operators.
COPY
The three COPY operators create an output data set for the employees in
each department containing only name and department. Note that the
ddname requested by the USING(xxxx) operand is xxxxCNTL. For
example, USING(NEW1) requests ddname NEW1CNTL.
SORT
Sorts the three output data sets created by the COPY operators along with
the master name/department data set and creates permanent and SYSOUT
data sets containing the resulting sorted records.
You can combine both of these examples into a single ICETOOL job step.
570
Job Control Language for ICETOOL
Job control language for ICETOOL

An overview of the job control language (JCL) statements for ICETOOL is in
Table 83 followed by discussions of each ICETOOL DD statement and the use of
reserved DD statements and ddnames.
Table 83. JCL Statements for ICETOOL
//EXAMPL JOB ...

//* ICETOOL CAN BE CALLED DIRECTLY OR FROM A PROGRAM
//STEP EXEC PGM=ICETOOL (or PGM=program_name)
//* THE FOLLOWING DD STATEMENTS ARE ALWAYS REQUIRED
//TOOLMSG DD SYSOUT=A (or DSN=...)
//* THE FOLLOWING DD STATEMENTS ARE USED FOR SYMBOL PROCESSING
//* SYMNAMES DD ...
//* SYMNOUT DD SYSOUT=A (OR DSN=...)
//* THE TOOLIN DD STATEMENT IS ONLY REQUIRED IF THE TOOLIN
//* INTERFACE IS USED.
//TOOLIN DD *
ICETOOL statements
/*
//* THE FOLLOWING DD STATEMENTS ARE ONLY REQUIRED IF SPECIFIED
//* IN ICETOOL STATEMENTS.
//indd DD ...
.
.
.
//outdd DD ...
.
.
.
//listdd DD SYSOUT=A (or DSN=...)
.
.
.
//countdd DD ...
.
.
.
//xxxxCNTL DD *
DFSORT control statements
/*
.
.
.
TOOLMSG DD Statement
Defines the ICETOOL message data set for all operations. ICETOOL messages
and statements appear in this data set. ICETOOL uses RECFM=FBA,
LRECL=121 and the specified BLKSIZE for the TOOLMSG data set. If the
BLKSIZE you specify is not a multiple of 121, ICETOOL uses BLKSIZE=121. If
you do not specify the BLKSIZE, ICETOOL selects the block size as directed by
the SDBMSG installation option (see z/OS DFSORT Installation and
Customization).
The TOOLMSG DD statement must be present.
571

DFSMSG DD Statement
Defines the DFSORT message data set for all operations. The DFSORT
messages and control statements from all ICETOOL calls to DFSORT appear in
this data set. Refer to the discussion of SYSOUT in System DD statements on
Either a DFSMSG DD statement or an SSMSG DD statement must be present. If
both are present, ICETOOL uses DFSMSG as the DFSORT message data set. If
a DFSMSG DD is not present, but an SSMSG DD is present, ICETOOL uses
SSMSG as the DFSORT message data set.
Note: A SYSOUT data set should be used for the DFSORT message data set
(for example, //DFSMSG DD SYSOUT=*). If you define the DFSORT message data
set as a temporary or permanent data set, you will only see the DFSORT
messages from the last call to DFSORT, unless you allocate a new data set
using a disposition of MOD.
//SYMNAMES DD
Defines the SYMNAMES data set containing statements to be used for symbol
processing. See Chapter 8, Using symbols for fields and constants, on page
731 for complete details.
//SYMNOUT DD
Defines the data set in which SYMNAMES statements and the symbol table are
to be listed. See Chapter 8, Using symbols for fields and constants, on page
731 for complete details.
TOOLIN DD statement
Defines the ICETOOL statement data set that must have the following
attributes: RECFM=F or RECFM=FB and LRECL=80.
If the TOOLIN Interface is used, the TOOLIN DD statement must be present. If
the Parameter List Interface is used, the TOOLIN DD statement is not required
and is ignored if present.
indd DD Statement
Defines an input data set for an operation. Refer to SORTIN DD statement
on page 67 (copy or sort) or SORTINnn DD statement on page 69 (merge)
for details. ICETOOL imposes the additional restriction that the LRECL of this
data must be at least 4.
An indd DD statement must be present for each unique indd name specified in
each FROM operand.
outdd DD Statement
Defines an output data set for a COPY, DATASORT, MERGE, RESIZE, SELECT,
SORT, SPLICE or SUBSET operation. Refer to SORTOUT and OUTFIL DD
statements on page 73 for details.
An outdd DD statement must be present for each unique outdd name specified
in each TO operand.
savedd DD Statement
Defines an output data set for a SELECT or SUBSET operation. Refer to
SORTOUT and OUTFIL DD statements on page 73 for details.
A savedd DD statement must be present for each unique savedd name
specified in each DISCARD operand.
listdd DD Statement
Defines the list data set for a DEFAULTS, DISPLAY, or OCCUR operation. For
each listdd data set, ICETOOL uses RECFM=FBA (or RECFM=FB for DISPLAY
572

or OCCUR with the NOCC operand specified), LRECL=121 (for DEFAULTS) or
the LRECL specified in the WIDTH operand or calculated as needed if WIDTH
is not specified (DISPLAY and OCCUR), and the specified block size. If the
BLKSIZE you specify is not a multiple of the LRECL, ICETOOL uses
BLKSIZE=LRECL. If you do not specify BLKSIZE, ICETOOL selects the block
size as directed by the LISTSDB or LISTNOSDB option if specified, or
otherwise as directed by the SDBMSG installation option (see z/OS DFSORT
Installation and Customization).
A listdd DD statement must be present for each unique listdd name specified
in each LIST operand.
countdd DD Statement
Defines the output data set for a COUNT operation. For each countdd data set,
ICETOOL uses RECFM=FB, the LRECL specified in the WIDTH operand or
calculated if WIDTH is not specified, and the specified block size. If the
BLKSIZE you specify is not a multiple of the LRECL, ICETOOL uses
BLKSIZE=LRECL. If you do not specify BLKSIZE, ICETOOL uses the
system-determined optimum block size.
A countdd DD statement must be present for each unique countdd name
specified in each WRITE operand.
xxxxCNTL DD Statement
Defines the DFSORT control statement data set for a COPY, COUNT,
DATASORT, MERGE, RESIZE, SELECT, SORT, SPLICE or SUBSET operation.
Refer to SORTCNTL DD statement on page 75 for more details.
An xxxxCNTL DD statement must be present for each unique xxxx specified in
each USING operand.
JCL restrictions
You should avoid using ddnames reserved for ICETOOL and DFSORT in
ICETOOL operands (FROM, TO, LIST, DISCARD, WRITE). In general, you should
also avoid supplying DD statements with ddnames reserved for DFSORT when
using ICETOOL because doing so can cause unpredictable results. Specifically:
v SORTCNTL should not be used as a ddname in ICETOOL operators nor should
it be supplied as a DD statement. A xxxxCNTL DD statement should only be
supplied when you specify a USING(xxxx) operand. xxxx must be four
characters that are valid in a ddname of the form xxxxCNTL. xxxx must not be
SYSx.
v SYSIN, SORTCNTL, SORTIN, SORTOUT, SORTINnn, and xxxxINnn (where xxxx
is specified in a USING operand) should not be used as ddnames in ICETOOL
operators nor supplied as DD statements.
v TOOLMSG, DFSMSG, SSMSG, SYMNAMES, SYMNOUT, TOOLIN, SYSUDUMP,
and SYSABEND should not be used as ddnames in ICETOOL operators.
v In general, xxxxWKdd ddnames should not be used as ddnames in ICETOOL
operators nor supplied as DD statements. However, if you want to override
dynamic allocation of work data sets for ICETOOL operators OCCUR and
UNIQUE, you can use SORTWKdd DD statements for that purpose. If you want
to override dynamic allocation of work data sets for ICETOOL operators
DATASORT, RESIZE, SELECT, SORT, SPLICE, and SUBSET, you can use
xxxxWKdd DD statements for that purpose in conjunction with the USING
operand.
v DFSPARM (or the ddname specified for the PARMDDN installation option)
should not be used as a ddname in ICETOOL operators. It should only be used
573
JCL Restrictions
as a DD statement to override DFSORT options for all operators, if appropriate.
Refer to DFSPARM DD statement on page 76 for details.
v xxxxOFdd (where xxxx is specified in a USING operand) is required as the
ddname when an OUTFIL statement in the xxxxCNTL data set specifies
FILES=dd. To avoid this requirement, use the FNAMES=ddname operand rather
than the FILES=dd operand in OUTFIL statements, and include a DD statement
for the specified ddname. See OUTFIL control statements on page 223 for
details of the FNAMES operand.
You should not use different DDs for a data set to be used for output and then
input in the same step, if that data set can be extended to a second or subsequent
volume, because that will result in incorrect output. See Data set notes and
limitations on page 13 for more information.
ICETOOL statements
Each operation must be described to ICETOOL using an operator statement.
Additionally, ICETOOL allows comment statements and blank statements. An
explanation of the general rules for coding ICETOOL statements is given later in
this section followed by a detailed discussion of each operator.
General coding rules

The general format for all ICETOOL operator statements is:
v OPERATOR operand ... operand
where each operand consists of KEYWORD(parameter, parameter...) or just
KEYWORD. Any number of operators can be specified and in any order.
The following rules apply for operator statements:
v The operator and operands must be in uppercase EBCDIC.
v The operator must be specified first.
v One blank is required between the operator and the first operand.
v One blank is required between operands.
v Any number of blanks can be specified before or after the operator or any
operand, but blanks cannot be specified anywhere else except within quoted
character strings.
v Parentheses must be used where shown. Commas or semicolons must be used
where commas are shown.
v Operands can be in any order.
v Columns 1-72 are scanned; columns 73-80 are ignored.
v Continuation can be indicated by a hyphen (-) after the operator or after any
operand. The next operand must then be specified on the next line. For example:
SORT FROM(INDD) USING(ABCD) TO(OUTPUT1,OUTPUT2,OUTPUT3)
Any characters specified after the hyphen are ignored. Each operand must be
completely specified on one line.
A statement with an asterisk (*) in column 1 is treated as a comment statement. It
is printed with the other ICETOOL statements, but otherwise not processed. A
statement with blanks in columns 1 through 72 is treated as a blank statement. It is
ignored, because ICETOOL prints blank lines where appropriate.
574
COPY Operator
COPY operator
,
COPY
FROM(indd)
JKFROM
TO( outdd
USING(xxxx)
,
TO( outdd

VSAMTYPE(x)
USING(xxxx)

LOCALE(name)
LOCALE(CURRENT)
LOCALE(NONE)
SERIAL
Copies an input data set to one or more output data sets.

DFSORT is called to copy the indd data set to the outdd data sets; the DFSORT
control statements in xxxxCNTL are used if USING(xxxx) is specified. You can use
DFSORT control statements and options in the xxxxCNTL data set to copy a subset
of the input records (INCLUDE or OMIT statement; SKIPREC and STOPAFT
options; OUTFIL INCLUDE, OMIT, SAVE, STARTREC, ENDREC, SAMPLE, SPLIT,
SPLITBY, and SPLIT1R operands; user exit routines), reformat records for output
(INREC, OUTREC and OUTFIL statements, user exit routines), and so on.
If an INCLUDE or OMIT statement or an OUTFIL INCLUDE or OMIT operand is
specified in the xxxxCNTL data set, the active locale's collating rules affect
INCLUDE and OMIT processing, as explained in the Cultural Environment
Considerations discussion in INCLUDE control statement on page 96.
Operand descriptions
The operands described in this section can be specified in any order.
FROM(indd)
Specifies the ddname of the input data set to be read by DFSORT for this
operation. An indd DD statement must be present and must define an input
data set that conforms to the rules for DFSORT's SORTIN data set.
Refer to JCL restrictions on page 573 for more information regarding the
selection of ddnames.
JKFROM
Specifies you are using a JOINKEYS application with this COPY operator to
copy the joined records. You must provide a USING(xxxx) operand. In
xxxxCNTL, you must provide a JOINKEYS statement with F1=ddname1 for the
F1 file and a JOINKEYS statement with F2=ddname2 for the F2 file, as well as
JOIN and REFORMAT statements as needed.
TO(outdd,...)
Specifies the ddnames of the output data sets to be written by DFSORT for this
operation. From 1 to 10 outdd names can be specified. An outdd DD statement
must be present for each outdd name specified. If a single outdd data set is
specified, DFSORT is called once to copy the indd data set to the outdd data
set, using SORTOUT processing; the outdd data set must conform to the rules
for DFSORT's SORTOUT data set. If multiple outdd data sets are specified and
575
COPY Operator
SERIAL is not specified, DFSORT is called once to copy the indd data set to
the outdd data sets, using OUTFIL processing; the outdd data sets must
conform to the rules for DFSORT's OUTFIL data sets.
TO and USING can both be specified. If USING is not specified, TO must be
specified. If TO is not specified, USING must be specified.
A ddname specified in the FROM operand must not also be specified in the
TO operand.
USING(xxxx)
Specifies the first 4 characters of the ddname for the control statement data set
to be used by DFSORT for this operation. xxxx must be four characters that are
valid in a ddname of the form xxxxCNTL. xxxx must not be SYSx.
If USING is specified, an xxxxCNTL DD statement must be present and the
control statements in it must conform to the rules for DFSORT's SORTCNTL
data set.
TO and USING can both be specified. If USING is not specified, TO must be
specified. If TO is not specified, USING must be specified and the xxxxCNTL
data set must contain either one or more OUTFIL statements or a MODS
statement for an E35 routine that disposes of all records. Other statements are
optional.
VSAMTYPE(x)
Specifies the record type for a VSAM input data set. x must be either F for
fixed-length record processing or V for variable-length record processing.
If VSAMTYPE(x) is specified, ICETOOL will pass a RECORD TYPE=x control
statement to DFSORT. (If you specify a RECORD TYPE=x statement in the
xxxxCNTL data set, it will override the one passed by ICETOOL.)
For complete information on record type processing for VSAM input, see
RECORD control statement on page 438.
LOCALE(name)
Specifies that locale processing is to be used and designates the name of the
locale to be made active during DFSORT processing. LOCALE(name) can be
used to override the LOCALE installation option. For complete details on
LOCALE(name), see the discussion of the LOCALE operand in OPTION
LOCALE(CURRENT)
Specifies that locale processing is to be used, and the current locale active
when DFSORT is entered will remain the active locale during DFSORT
processing. LOCALE(CURRENT) can be used to override the LOCALE
installation option. For complete details on LOCALE(CURRENT), see the
discussion of the LOCALE operand in OPTION control statement on page
173.
LOCALE(NONE)
Specifies that locale processing is not to be used. DFSORT will use the binary
encoding of the code page defined for your data for collating and comparing.
LOCALE(NONE) can be used to override the LOCALE installation option. For
576
COPY Operator
complete details on LOCALE(NONE), see the discussion of the LOCALE
operand in OPTION control statement on page 173.
SERIAL
Specifies that OUTFIL processing is not to be used when multiple outdd data
sets are specified. DFSORT is called multiple times and uses SORTOUT
processing; the outdd data sets must conform to the rules for DFSORT's
SORTOUT data set. SERIAL is not recommended because the use of serial
processing (that is, multiple calls to DFSORT) instead of OUTFIL processing
can degrade performance and imposes certain restrictions as detailed later in
this section. SERIAL is ignored if a single outdd data set is specified.
DFSORT is called to copy the indd data set to the first outdd data set using the
DFSORT control statements in the xxxxCNTL data set if USING(xxxx) is
specified. If the first copy is successful, DFSORT is called as many times as
necessary to copy the first outdd data set to the second and subsequent outdd
data sets. Therefore, for maximum efficiency, use a disk data set as the first in
a list of outdd data sets on both disk and tape. If more than one outdd data set
is specified, DFSORT must be able to read the first outdd data set after it is
written in order to copy it to the other outdd data sets. Do not use a SYSOUT
or DUMMY data set as the first in a list of outdd data sets because:
v if the first data set is SYSOUT, DFSORT abends when it tries to copy the
SYSOUT data set to the second outdd data set.
v if the first data set is DUMMY, DFSORT copies the empty DUMMY data set
to the other outdd data sets, with the result that all outdd data sets are then
empty.
Copy examples
This section contains 2 copy examples.
Example 1
* Method 1
COPY FROM(MASTER) TO(PRINT,TAPE,DISK)
* Method 2
COPY FROM(MASTER) TO(DISK,TAPE,PRINT) SERIAL
This example shows two different methods for creating multiple output data sets.
Method 1 requires one call to DFSORT, one pass over the input data set, and
allows the output data sets to be specified in any order. The COPY operator copies
all records from the MASTER data set to the PRINT (SYSOUT), TAPE, and DISK
data sets, using OUTFIL processing.
Method 2 requires three calls to DFSORT, three passes over the input data set, and
imposes the restriction that the SYSOUT data set must not be the first TO data set.
The COPY operator copies all records from the MASTER data set to the DISK data
set and then copies the resulting DISK data set to the TAPE and PRINT (SYSOUT)
data sets. Because the first TO data set is processed three times (written, read,
read), placing the DISK data set first is more efficient than placing the TAPE data
set first. PRINT must not be the first in the TO list because a SYSOUT data set
cannot be read.
Example 2
* Method 1
COPY FROM(IN) TO(DEPT1) USING(DPT1)
577
COPY Operator
* Method 2
COPY FROM(IN) USING(ALL3)
This example shows two different methods for creating subsets of an input data
set. Assume that:
v The DPT1CNTL data set contains:
INCLUDE COND=(5,3,CH,EQ,CD01)


v The ALL3CNTL data set contains:

OUTFIL FNAMES=DEPT1,INCLUDE=(5,3,CH,EQ,CD01)
Method 1 requires three calls to DFSORT and three passes over the input data set:
v The first COPY operator copies the records from the IN data set that contain D01
in positions 5-7 to the DEPT1 data set.
v The second COPY operator copies the records from the IN data set that contain
D02 in positions 5-7 to the DEPT2 data set.
v The third COPY operator copies the records from the IN data set that contain
D03 in positions 5-7 to the DEPT3 data set.
Method 2 accomplishes the same result as method 1, but because it uses OUTFIL
statements instead of TO operands, requires only one call to DFSORT and one pass
over the input data set.
Example 3
COPY FROM(VSAMIN) TO(VSAMOUT) VSAMTYPE(V)
The COPY operator copies all records from the VSAMIN data set to the
VSAMOUT data set. The VSAM records are treated as variable-length.
COPY operator with JOINKEYS example

Here is an example of using multiple COPY operators for JOINKEYS applications
that preprocess different input files in different ways:
//CPYJK EXEC PGM=ICETOOL
//IN1 DD DSN=MY.IN1,DISP=SHR
//OUT1 DD SYSOUT=*
//OUT2 DD SYSOUT=*
//TOOLIN DD *
* First COPY operator with JOINKEYS application.
COPY JKFROM TO(OUT1) USING(CTL1)
* Second COPY operator with JOINKEYS application.
COPY JKFROM USING(CTL2)
/*
//CTL1CNTL DD *
* JOINKEYS application control statements for first COPY operator.
JOINKEYS F1=IN1,FIELDS=(5,12,A),TASKID=T1
JOIN UNPAIRED
REFORMAT FIELDS=(F1:4,40,F2:15,20),FILL=C$
578
COPY Operator
* Main task control statements for first COPY operator
* (operates on joined records).
INCLUDE COND=(8,1,CH,EQ,CY)
/*
//CTL2CNTL DD *
* JOINKEYS application control statements for second COPY operator.
JOINKEYS F2=IN3,FIELDS=(9,12,A),TASKID=T2,SORTED
* Main task control statements for second COPY operator
OUTFIL FNAMES=OUT2,HEADER1=(Analysis Report),REMOVECC
/*
//T1F1CNTL DD *
* Control statements for subtask1 (F1=IN1) of both COPY operators.
* Subtask1 sorts/joins on 5,12,A automatically
* per JOINKEYS statement for TASKID=T1/F1=IN1.
INCLUDE COND=(21,3,CH,EQ,CJ82)
SUM FIELDS=NONE
/*
//T1F2CNTL DD *
* Control statements for subtask2 (F2=IN2) of first COPY operator.
* Subtask1 sorts/joins on 11,12,A automatically
* per JOINKEYS statement for TASKID=T1/F2=IN2.
OPTION SKIPREC=1
/*
//T2F2CNTL DD *
* Control statements for subtask2 (F2=IN3) of second COPY operator.
* Subtask1 copies/joins on 9,12,A automatically
* per JOINKEYS statement for TASKID=T2/F2=IN3/SORTED.
/*
COUNT operator
COUNT FROM(indd)

USING(xxxx)
VSAMTYPE(x)
LOCALE(name)
LOCALE(CURRENT)
LOCALE(NONE)

RC4
RC8
RC12
EMPTY
NOTEMPTY
HIGHER(x)
LOWER(y)
EQUAL(v)
NOTEQUAL(w)
SUB(q)
ADD(r)
WRITE(countdd)

TEXT('string')
DIGITS(d)
EDCOUNT(formatting)
WIDTH(n)
Prints a message containing the count of records in a data set. Can also be used to
subtract a value from the count or add a value to the count, to create an output
data set containing text and the count, or to set RC=12, RC=8, RC=4, or RC=0
based on meeting criteria for the number of records in a data set.
579
COUNT Operator
DFSORT is called to copy the indd data set to ICETOOL's E35 user exit. The
DFSORT control statements in xxxxCNTL are used if USING(xxxx) is specified. You
can use a DFSORT INCLUDE or OMIT statement in the xxxxCNTL data set to
count a subset of the input records.
If an INCLUDE or OMIT statement is specified in the xxxxCNTL data set, the
active locale's collating rules affect INCLUDE and OMIT processing as explained in
the Cultural Environment Considerations discussion in INCLUDE control
If EMPTY, NOTEMPTY, HIGHER(x), LOWER(y), EQUAL(v) or NOTEQUAL(w) is
not specified, ICETOOL prints a message containing the record count as
determined by its E35 user exit.
If EMPTY, NOTEMPTY, HIGHER(x), LOWER(y), EQUAL(v) or NOTEQUAL(w) is
specified, ICETOOL checks the record count as determined by its E35 user exit
against the specified criteria. If the criteria is met (for example, HIGHER(20) is
specified and the record count is 21 or more), ICETOOL sets the following return
code for the COUNT operator:
v RC=12 if RC12 is specified, or by default if RC8 and RC4 are not specified
v RC=8 if RC8 is specified
v RC=4 if RC4 is specified
If the criteria is not met (for example, HIGHER(20) is specified and the record
count is 20 or less), ICETOOL sets RC=0 for the COUNT operator. ICETOOL uses
DFSORT's STOPAFT option to process the minimum number of records required to
determine whether or not the criteria is met.
Note: Be sure to check the messages in TOOLMSG when you initially set up any
COUNT operators with criteria to make sure that RC=12 is not issued because of
syntax errors.
If SUB(q) is specified, ICETOOL subtracts q from the count, but does not reduce
the count below 0. If ADD(r) is specified, ICETOOL adds r to the count. SUB(q)
and ADD(r) are especially useful for dealing with data sets that contain header and
trailer records.
If WRITE(countdd) is specified, ICETOOL writes a record containing the count in
the countdd data set. TEXT('string') can be used to insert a string before the count
in the output record. DIGITS(n) or EDCOUNT(formatting) can be used to change
the appearance of the count in the output record. WIDTH(n) can be used to change
the length of the output record.
You must not supply your own DFSORT MODS statement because it would
override the MODS statement passed by ICETOOL for this operator.
Note: The record count is also printed for the DISPLAY, OCCUR, RANGE,
SELECT, STATS, UNIQUE, and VERIFY operators.
FROM(indd)
See the discussion of this operand on the COPY statement in COPY operator
on page 575.
580
COUNT Operator
USING(xxxx)
control statements in it:
1. Must conform to the rules for DFSORT's SORTCNTL data set.
2. Should generally be used only for an INCLUDE or OMIT statement or
comments statements.
VSAMTYPE(x)
on page 575.
LOCALE(name)
on page 575.
LOCALE(CURRENT)
on page 575.
LOCALE(NONE)
on page 575.
RC4
Can be used to set RC=4 for this COUNT operator if the record count meets
the specified criteria. RC4 can be specified only if EMPTY, NOTEMPTY,
HIGHER(x), LOWER(y), EQUAL(v), or NOTEQUAL(w) is specified. RC4
overrides the default of setting RC=12 for this COUNT operator if the record
count meets the specified criteria
RC8
HIGHER(x), LOWER(y), EQUAL(v), or NOTEQUAL(w) is specified. RC8
overrides the default of setting RC=12 for this COUNT operator if the record
count meets the specified criteria.
RC12
HIGHER(x), LOWER(y), EQUAL(v), or NOTEQUAL(w) is specified. RC12 is
equivalent to the default used when RC4 or RC8 is not specified.
EMPTY
Sets RC=12 (or RC=8 if RC8 is specified, or RC=4 if RC4 is specified) for this
COUNT operator if the input data set or subset is empty, or sets RC=0 for this
COUNT operator if the input data set or subset is not empty.
EMPTY is equivalent to EQUAL(0).
NOTEMPTY
COUNT operator if the input data set or subset is not empty, or sets RC=0 for
this COUNT operator if the input data set or subset is empty.
581
COUNT Operator
EMPTY is equivalent to NOTEQUAL(0).
HIGHER(x)
COUNT operator if the record count is higher than x, or sets RC=0 for this
COUNT operator if the record count is equal to or lower than x.
x must be specified as n or +n where n can be 0 to 562949953421310.
LOWER(y)
COUNT operator if the record count is lower than y, or sets RC=0 for this
COUNT operator if the record count is equal to or higher than y.
y must be specified as n or +n where n can be 0 to 562949953421310.
EQUAL(v)
COUNT operator if the record count is equal to v, or sets RC=0 for this
COUNT operator if the record count is not equal to v.
v must be specified as n or +n where n can be 0 to 562949953421310.
NOTEQUAL(w)
COUNT operator if the record count is not equal to w, or sets RC=0 for this
COUNT operator if the record count is equal to w.
w must be specified as n or +n where n can be 0 to 562949953421310.
SUB(q)
Subtracts q from the record count, but does not reduce the count below 0. The
resulting modified record count is displayed in the count message in
TOOLMSG. If WRITE(countdd) is specified, the modified record count is used
in the count record. If EMPTY, NOTEMPTY, HIGHER(x), LOWER(y),
EQUAL(v) or NOTEQUAL(w) is specified, the modified record count is used
to determine if the criteria is satisfied.
SUB(q) is especially useful for getting the count of just the data records for a
data set that contains header, data and trailer records.
q must be specified as n or +n where n can be 1 to 999.
ADD(r)
Adds r to the record count. The resulting modified record count is displayed in
the count message in TOOLMSG. If WRITE(countdd) is specified, the modified
record count is used in the count record. If EMPTY, NOTEMPTY, HIGHER(x),
LOWER(y), EQUAL(v) or NOTEQUAL(w) is specified, the modified record
count is used to determine if the criteria is satisfied.
ADD(r) is especially useful for getting the count of header, data and trailer
records for a data set that just contains data records to which you want to add
header and trailer records.
r must be specified as n or +n where n can be 1 to 999.
WRITE(countdd)
Specifies the ddname of the count data set to be produced by ICETOOL for
this operation. A countdd DD statement must be present. ICETOOL sets the
attributes of the count data set as follows:
v RECFM is set to FB.
v LRECL is set to one of the following:
582
COUNT Operator
If WIDTH(n) is specified, LRECL is set to n. Use WIDTH(n) if your count
record length and LRECL must be set to a particular value (for example,
80), or if you want to ensure that the count record length does not exceed
a specific maximum (for example, 20 bytes).
If WIDTH(n) is not specified, LRECL is set to the calculated required
record length. If your LRECL does not need to be set to a particular
value, you can let ICETOOL determine and set the appropriate LRECL
value by not specifying WIDTH(n).
v BLKSIZE is set to one of the following:
The BLKSIZE from the DD statement, DSCB, or label, if it is a multiple of
the LRECL used.
The LRECL if the BLKSIZE from the DD statement, DSCB, or label is not
a multiple of the LRECL used.
The system determined blocksize if the BLKSIZE is not available from the
DD statement, DSCB, or label.
TEXT('string')
Specifies a string to be printed starting in the first byte of the count record. The
count follows the string. TEXT can only be specified if WRITE(countdd) is
specified.
The string (1 to 50 characters) must be enclosed in single apostrophes. To
include a single apostrophe (') in the string, specify two single apostrophes ('').
To suppress printing of a string, do not specify TEXT('string') or specify
TEXT('') using two single apostrophes.
DIGITS(d)
Specifies d digits for the count in the output record, overriding the default of
15 digits. d can be 1 to 15. The count is written as d decimal digits with
leading zeros. DIGITS can only be specified if WRITE(countdd) is specified.
If you know that your count requires less than 15 digits, you can use a lower
number of digits (d) instead by specifying DIGITS(d). For example, if
DIGITS(10) is specified, 10 digits are used instead of 15.
If you use DIGITS(d) and the count overflows the number of digits used,
ICETOOL terminates the operation. You can prevent the overflow by specifying
an appropriately higher d value for DIGITS(d). For example, if DIGITS(5)
results in overflow, you can use DIGITS(6) instead.
EDCOUNT(formatting)
Specifies formatting items that indicate how the count in the output record is
to be formatted, overriding the default of 15 digits with leading zeros.
Formatting items can be specified in any order, but each item can only be
specified once. EDCOUNT can only be specified if WRITE(countdd) is
specified.
mask
See the discussion of mask under ON(p,m,f,formatting) in "DISPLAY
Operator" on page mask on page 601.
E'pattern'
specifies an edit pattern to be applied to the count. The pattern (1 to 24
characters) must be enclosed in single apostrophes. Each 9 in the pattern
(up to 15) is replaced by a corresponding digit from the count. Characters
other than 9 in the pattern appear as specified. To include a single
apostrophe (') in the pattern, specify two single apostrophes (''). F'string' or
a mask cannot be specified with E'pattern'.
583
COUNT Operator
When E'pattern' is specified for the count:
v If the number of significant digits in the count is less than the number of
9's in the pattern, 0's are filled in on the left. For example, 1234 is shown
as 001234 with EDCOUNT(E'999999').
v If the number of significant digits in the count is greater than the
number of 9's in the pattern, digits are truncated from the left. For
example, 1234567 is shown as *4567* with EDCOUNT(E'*9999*').
L'string
See the discussion of L'string' under ON(p,m,f,formatting) in "DISPLAY
Operator" on page L'string' on page 603
F'string'
See the discussion of F'string' under ON(p,m,f,formatting) in "DISPLAY
Operator" on page F'string' on page 603
T'string'
See the discussion of T'string' under ON(p,m,f,formatting) in "DISPLAY
Operator" on page T'string' on page 604
LZ See the discussion of LZ under ON(p,m,f,formatting) in "DISPLAY
Operator" on page LZ on page 604
Udd
specifies the number of digits to be used for the count. dd specifies the
number of digits and must be a two-digit number between 01 and 15. The
default number of digits (d) for the count is 15.
If you know that your count requires less than 15 digits, you can use a
lower number of digits (dd) instead by specifying Udd. For example, if
EDCOUNT(U09) is specified, 9 digits (from U09) is used instead of 15
(default for the count).
If you use Udd and the count overflows the number of digits used,
ICETOOL terminates the operation. You can prevent the overflow by
specifying an appropriately higher dd value for Udd. For example, if
EDCOUNT(U05) results in overflow, you can use EDCOUNT(U06) instead.
If E'pattern' is specified, Udd is ignored, because d is determined from the
pattern.
WIDTH(n)
Specifies the record length and LRECL you want ICETOOL to use for the count
data set. n can be from 1 to 32760. WIDTH can only be specified if
WRITE(countdd) is specified. ICETOOL always calculates the record length
required to write the count record and uses it as follows:
v If WIDTH(n) is specified and the calculated record length is less than or
equal to n, ICETOOL sets the record length and LRECL to n. ICETOOL pads
the count record on the right with blanks to the record length.
v If WIDTH(n) is specified and the calculated record length is greater than n,
ICETOOL issues an error message and terminates the operation.
v If WIDTH(n) is not specified, ICETOOL sets the record length and LRECL to
the calculated record length.
Use WIDTH(n) if your count record length and LRECL must be set to a
particular value (for example, 80), or if you want to ensure that the count
record length does not exceed a specific maximum (for example, 20 bytes).
Otherwise, you can let ICETOOL calculate and set the appropriate record
length and LRECL by not specifying WIDTH(n).
584
COUNT Operator
COUNT examples
Example 1
For this example, assume that the CTL1CNTL data set contains a DFSORT
INCLUDE statement.
COUNT FROM(IN1)
COUNT FROM(IN2) USING(CTL1)
The first COUNT operator prints a message containing the count of records in the
IN1 data set.
The second COUNT operator prints a message containing the count of records
included from the IN2 data set.
Example 2
COUNT FROM(INPUT1) EMPTY
Sets RC=12 if INPUT1 is empty (that is, INPUT1 has no records), or sets RC=0 if
INPUT1 is not empty (that is, INPUT1 has at least one record).
Example 3
For this example, assume that the CTL2CNTL data set contains a DFSORT
INCLUDE statement.
COUNT FROM(INPUT2) HIGHER(50000) RC4 USING(CTL2)
Sets RC=4 if more than 50000 records are included from INPUT2, or sets RC=0 if
50000 or less records are included from INPUT2.
Example 4
COUNT FROM(IN2) WRITE(CT2) TEXT(Count is ) EDCOUNT(A1,U10) WIDTH(80)
Prints a message containing the count of records in the IN2 data set. Writes an
80-byte record with the specified string and an edited count to the CT2 data set. If
IN2 contains 3286721 records, the 80-byte output record in CT2 would look like
this:
Count is
3,286,721
Example 5
COUNT FROM(IN3) WRITE(CT3) DIGITS(6) SUB(2)
Subtracts 2 from the count of records in the IN3 data set. Prints a message
containing the modified count. Writes a 6-byte record with the modified count to
the CT3 data set. If IN3 contains 8125 records, the 6-byte output record in CT3
would look like this:
008123
DATASORT operator
DATASORT FROM(indd) TO(outdd) USING(xxxx)
HEADER
FIRST
HEADER(u)
FIRST(u)
585
DATASORT Operator
TRAILER
LAST
TRAILER(v)
LAST(v)

VSAMTYPE(x)
Copies one or more header records and one or more trailer records to the output
data set in their original input record order, while sorting the data records between
the header and trailer records, using the DFSORT control statements in xxxxCNTL.
By definition, the header records are the first n records in the input data set, the
trailer records are the last n records in the input data set, and the data records
(also called detail records) are the records between the header and trailer records.
Thus, the first n records (header records) and last n records (trailer records) are
kept in place and the data records between them are sorted.
You must specify one header operand (HEADER, FIRST, HEADER(u) or FIRST(u)),
one trailer operand (TRAILER, LAST, TRAILER(v) or LAST(v)), or one header
operand and one trailer operand. If you specify a header operand without a trailer
operand, only the header records will be kept in place. If you specify a trailer
operand without a header operand, only the trailer records will be kept in place. If
you specify a header operand and a trailer operand, both the header records and
trailer records will be kept in place.
DFSORT is called to copy the header and trailer records and to sort the data
records. DFSORT uses its E15 and E35 exits to process the records as needed.
ICETOOL passes the EQUALS option to DFSORT to ensure that duplicates are
kept in their original input order.
You must supply a DFSORT SORT statement in the xxxxCNTL data set to indicate
the control fields for sorting the data records.
You can use additional DFSORT control statements in xxxxCNTL providing you
observe these rules:
v A SORT statement must be present
v MODS and OUTREC statements should not be present.
v A STOPAFT operand should not be present.
v Comment statements can be present.
v Header and trailer records will only be affected by the SKIPREC option and
OUTFIL statements. SKIPREC=n will remove the first n indd records, so the first
header record will be the n+1 indd record. OUTFIL statements will process the
header and trailer records in the normal way.
v Data records will be processed by INCLUDE, OMIT, INREC, SUM, OPTION and
OUTFIL statements in the normal way.
v If you use INREC to change the length of the data records, DFSORT will
preserve the header and trailer records by setting the TO data set LRECL to the
maximum of the input or reformatted record length. For fixed-length records,
DFSORT will pad the header and trailer records, or data records, on the right
with blanks as appropriate.
v You can further process the outdd records after DATASORT processing using an
OUTFIL statement like this:
OUTFIL FNAMES=outdd,...
or multiple OUTFIL statements like this:
586
DATASORT Operator
OUTFIL FNAMES=outdd1,...
For example, with TO(OUT1) you could further modify the OUT1 records after
DATASORT processing, with a statement like this:
OUTFIL FNAMES=OUT1,REMOVECC,
TRAILER1=(Record count ,COUNT=(M11,LENGTH=5))
ICETOOL requires extra storage for DATASORT processing, over and above what
is normally needed by ICETOOL and DFSORT, in order to save your header and
trailer records. The amount of storage needed depends on the number of header
and trailer records you specify, and the LRECL of the FROM data set. In most
cases, the needed storage can be obtained (above 16MB virtual). However, if
ICETOOL cannot get the storage it needs, it issues a message and terminates the
DATASORT operation. Increasing the REGION by the amount indicated in the
message may allow ICETOOL to run successfully.
The DYNALLOC option is passed to DFSORT to ensure that work space is
available for the sort. If your installation defaults for dynamic allocation are
inappropriate for a DATASORT operator, you can take one of the following actions:
1. Override the DYNALLOC option using an OPTION control statement such as:
OPTION DYNALLOC=(,8)
in the xxxxCNTL data set.

2. Use xxxxWKdd DD statements to override the use of dynamic allocation. Refer
to SORTWKdd DD statement on page 71 for details.
Tape work data sets cannot be used with ICETOOL.
FROM(indd)
See the discussion of this operand on the COPY statement in "COPY Operator"
on page FROM(indd) on page 575.
TO(outdd)
Specifies the ddname of the output data set to be written by DFSORT for this
operation.
An outdd DD statement must be present and must define an output data set
that conforms to the rules for DFSORT's SORTOUT data set.
The ddname specified in the TO operand must not be the same as the ddname
specified in the FROM operand.
Refer to JCL restrictions on page 573 for more information.
USING(xxxx)
valid in a ddname of the form xxxxCNTL. xxxx must not be SYSx
An xxxxCNTL DD statement must be present. The xxxxCNTL data set must
contain a SORT statement. Other control statements are optional, but if
specified must conform to the rules for DFSORT's SORTCNTL data set, and
should be used as described for DATASORT operator on page 585.
HEADER or FIRST
Specifies one header record (the first record in the indd data set) is to be kept
in place.
587
DATASORT Operator
HEADER and FIRST are equivalent to HEADER(1) and FIRST(1).
HEADER(u) or FIRST(u)
Specifies u header records (the first u records in the indd data set) are to be
kept in place.
u must be specified as n or +n where n can be 1 to 1000000.
TRAILER or LAST
Specifies one trailer record (the last record in the indd data set) is to be kept in
place.
TRAILER and LAST are equivalent to TRAILER(1) and LAST(1).
TRAILER(v) or LAST(v)
Specifies v trailer records (the last v records in the indd data set) are to be kept
in place.
VSAMTYPE(x)
See the discussion of this operand on the COPY statement in "COPY Operator"
on page VSAMTYPE(x) on page 576.
DATASORT examples
Although the DATASORT operators in the examples in this section could all be
contained in a single ICETOOL job step, they are shown and discussed separately
for clarity.
Example 1
DATASORT FROM(INPUT) TO(OUTPUT) HEADER TRAILER USING(CTL1)
This example shows how you can sort the data records between a header record
(first record) and a trailer record (last record).
The CTL1CNTL data set contains:
The input records look like this:

MM9999900510100823DDDDD FFFFF 004200806128
AAR
FIRST C 1134341444441
XXXXXXXXX
ATX
SECOND
7777777770111
XXXXXXXXX
ATX
THIRD L 6297132201111
XXXXXXXXX
ATX
FOURTH
2830012906356
XXXXXXXXX
MM9999900510100823DDDDD FFFFF 004
We want to keep the MM records in place and sort the other records by the CH
field in positions 16-28. We use HEADER and TRAILER to indicate the first and
last records should not be sorted. We use the SORT statement to SORT ascending
on positions 16-28.
The output records would look like this:
MM9999900510100823DDDDD FFFFF 004200806128
AAR
FIRST C 1134341444441
XXXXXXXXX
ATX
FOURTH
2830012906356
XXXXXXXXX
ATX
THIRD L 6297132201111
XXXXXXXXX
ATX
SECOND
7777777770111
XXXXXXXXX
MM9999900510100823DDDDD FFFFF 004
Note that we could use FIRST and LAST instead of HEADER and TRAILER.
588
DATASORT Operator
Example 2
DATASORT FROM(IN) TO(OUT) HEADER(2) TRAILER(3) USING(CTL2)
This example illustrates how you can sort the data records between header records
(first records) and trailer records (last records), and modify just the data records or
the header, data and trailer records.
INREC IFTHEN=(WHEN=(24,2,CH,EQ,C23),
OVERLAY=(30:COld))
OUTFIL FNAMES=OUT,
IFTHEN=(WHEN=(24,2,CH,EQ,C23),
OVERLAY=(35:CFirst))

Header 1
Header 2
Geometry
Algebra
Trigonometry
Calculus
Geography
History
Trailer 1
Trailer 2
Trailer 3
2008/04/23
2008/04/23
2008/04/24
2008/04/23
2008/04/24
2008/04/25
2008/04/25
2008/04/23
2008/04/23
2008/04/23
2008/04/23
We want to keep the two Header records and the three Trailer records in place and
sort the other records by the CH field in positions 1-14. We want to put 'Old' in
positions 30-32 of each data record (but not the Header or Trailer records) that has
'23' in positions 24-25. We want to put 'First' in positions 35-39 of each record
(Header, data and Trailer) that has '23' in positions 24-25.
We use HEADER(2) and TRAILER(3) to indicate the first two records and last three
records should not be sorted. We use the INREC statement to add 'Old' to data
records that have '23' in positions 24-25. INREC applies to the data records, but not
to the Header and Trailer records. We use the SORT statement to sort ascending on
positions 1-14. We use the OUTFIL statement to add 'First' to Header, data and
Trailer records that have '23' in positions 24-25. OUTFIL applies to all of the
records.
Header 1
Header 2
Algebra
Calculus
Geography
Geometry
History
Trigonometry
Trailer 1
Trailer 2
Trailer 3
2008/04/23
2008/04/23
2008/04/23
2008/04/25
2008/04/25
2008/04/24
2008/04/23
2008/04/24
2008/04/23
2008/04/23
2008/04/23
Old
First
First
First
Old
First
First
First
First
Note that we could use FIRST(2) and LAST(3) instead of HEADER(2) and
TRAILER(3).
589
DEFAULTS Operator
DEFAULTS operator
DEFAULTS LIST(listdd)

LISTSDB
LISTNOSDB
Prints the DFSORT installation defaults report in a separate list data set.
DFSORT enables you to maintain separate sets of installation defaults for eight
different installation environments as follows:
v JCL (ICEAM1) - batch JCL directly invoked installation environment
v INV (ICEAM2) - batch program invoked installation environment
v
v
v
v
v
v
TSO (ICEAM3) - TSO directly invoked installation environment

TSOINV (ICEAM4) - TSO program invoked installation environment
TD1 (ICETD1) - first time-of-day installation environment
TD2 (ICETD2) - second time-of-day installation environment
TD3 (ICETD3) - third time-of-day installation environment
TD4 (ICETD4) - fourth time-of-day installation environment
Each installation default has two or more possible values; DFSORT is shipped with
a set of IBM-supplied defaults that can be modified using ICEPRMxx members of
PARMLIB or the ICEMAC macro. The DEFAULTS operator provides an easy way
to determine the installation default values to be used at run-time. See z/OS
DFSORT Installation and Customization for a complete discussion of ICEPRMxx
members in PARMLIB, the ICEMAC macro, the eight installation environments,
and the installation default values.
DEFAULTS produces a three-part report showing:
1. The merged PARMLIB/ICEMAC installation default values for ICEAM1-4 and
ICETD1-4 that will be used at run-time.
2. The specified PARMLIB ICEPRMxx member option values for ICEAM1-4 and
ICETD1-4 (for reference).
3. The ICEMAC installation default values for ICEAM1-4 and ICETD1-4 (for
reference).
The format of the report produced by DEFAULTS varies depending on the defaults
selected, but the first part of the report might look like this conceptually:
590
DEFAULTS Operator
Z/OS DFSORT V2R1
MERGED PARMLIB/ICEMAC DEFAULTS
- p -
* IBM-SUPPLIED DEFAULT (ONLY SHOWN IF DIFFERENT FROM THE SPECIFIED DEFAULT)

ITEM
---------item
.
.
.
item
item
JCL (ICEAM1) VALUE

-------------------value
INV (ICEAM2) VALUE

-------------------value
TSO (ICEAM3) VALUE

-------------------value
TSOINV (ICEAM4) VALUE

------------------value
value
value
value
value
* IBM_value
value
value
value
value
.
.
.
Z/OS DFSORT V2R1
- p -

ITEM
---------item
.
.
.
item
item
.
.
.
TD1 (ICETD1) VALUE

-------------------value
TD2 (ICETD2) VALUE

-------------------value
TD3 (ICETD3) VALUE

-------------------value
TD4 (ICETD4) VALUE

------------------value
value
* IBM_value
value
value
* IBM_value
value
value
value
value
value
The merged PARMLIB/ICEMAC default value for each item is shown as it is set
for each of the eight installation environments. For any value that is different from
the IBM-supplied value, the IBM-supplied value is shown below it.
The control character occupies the first byte of each record. The title and headings
are always printed; p is the page number. The item name column occupies 10
bytes, each of the item value columns occupies 20 bytes, and 5 blanks appear
between columns.
LIST(listdd)
Specifies the ddname of the list data set to be produced by ICETOOL for this
operation. A listdd DD statement must be present. ICETOOL uses
RECFM=FBA, LRECL=121 and the specified BLKSIZE for the list data set. If
the BLKSIZE you specify is not a multiple of 121, ICETOOL uses
BLKSIZE=121. If you do not specify the BLKSIZE, ICETOOL selects the block
size as directed by LISTSDB or LISTNOSDB, if specified, or otherwise as
directed by installation option SDBMSG from ICEAM2 or ICEAM4 (see z/OS
DFSORT Installation and Customization).
LISTSDB or LISTNOSDB
Can be used to override the SDBMSG value for this LIST data set. LISTSDB
directs ICETOOL to select the system-determined optimum block size for the
LIST data set in the same way as for installation option SDBMSG=YES.
LISTNOSDB directs ICETOOL to select the block size for the LIST data set in
the same way as for installation option SDBMSG=NO. See the discussion of the
591
DEFAULTS Operator
LIST(listdd) operand previously in this section for more information on how
LISTSDB or LISTNOSDB affects the LIST data set block size.
Attention: LISTSDB has no effect for SYSOUT data sets (for example, //RPT1
DD SYSOUT=*), because the system-determined optimum block size is not
used for spool or dummy data sets.
DEFAULTS example
DEFAULTS LIST(OPTIONS)
Prints, in the OPTIONS data set, the DFSORT installation defaults report. The
OPTIONS output starts on a new page and might look as follows. The first few
items are shown with illustrative values for ICEAM1-4 and ICETD1-4 for each of
the three parts of the report.
592
DEFAULTS Operator
Z/OS DFSORT V2R1
- 1 -

ITEM
---------ENABLE
JCL (ICEAM1) VALUE

-------------------NONE
INV (ICEAM2) VALUE

-------------------TD1
TSO (ICEAM3) VALUE

-------------------NONE
TSOINV (ICEAM4) VALUE

--------------------NONE
ABCODE
MSG
MSG
ALTSEQ
ARESALL
ARESINV
CFW
CHALT
SEE BELOW
0
NOT APPLICABLE
YES
YES
* NO
YES
YES
99
* MSG
SEE BELOW
0
0
YES
YES
* NO
YES
YES
SEE BELOW
0
NOT APPLICABLE
YES
NO
99
* MSG
SEE BELOW
0
0
YES
NO
YES
YES
YES
YES
CHECK
CINV
.
.
.
Z/OS DFSORT V2R1
- 4 -

ITEM
---------SUN
MON
TUE
WED
THU
FRI
SAT
ABCODE
ALTSEQ
ARESALL
ARESINV
CFW
CHALT
TD1 (ICETD1) VALUE

-------------------0600-2000
* NONE
NONE
NONE
NONE
NONE
NONE
0600-2000
* NONE
TD2 (ICETD2) VALUE

-------------------NONE
TD3 (ICETD3) VALUE

-------------------NONE
TD4 (ICETD4) VALUE

-----------------NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
NONE
99
* MSG
SEE BELOW
0
0
YES
YES
* NO
YES
YES
MSG
MSG
MSG
SEE BELOW
0
0
YES
NO
SEE BELOW
0
0
YES
NO
SEE BELOW
0
0
YES
NO
YES
YES
YES
YES
YES
YES
CHECK
CINV
.
.
.
Z/OS DFSORT V2R1
PARMLIB ICEPRMXX MEMBER OPTIONS
- 7 -
ICEOPT COMMAND IN EFFECT: START ICEOPT,ICEPRM=(03)

RELEASE:
MODULE:
APAR LEVEL:
COMPILED:
AREA:
2.01
ICEPRML
BASE
05/14/11
1
JCL (ICEAM1) OPTIONS

ITEM
---------RESALL
CHALT
VALUE
---------------------------------------40000
YES
MEMBER
-------ICEPRM03
ICEPRM03
INV (ICEAM2) OPTIONS

ITEM
---------RESINV
CHALT
VALUE
---------------------------------------40000
YES
MEMBER
-------ICEPRM03
ICEPRM03
TSO (ICEAM3) OPTIONS - NONE

TSOINV (ICEAM4) OPTIONS - NONE
TD1 (ICETD1) OPTIONS
ITEM
---------CHALT
VALUE
---------------------------------------YES
TD2 (ICETD2) OPTIONS - NONE

MEMBER
-------ICEPRM03
593
DEFAULTS Operator
The title and appropriate heading lines appear at the top of each page. The
specified and IBM-supplied ALTSEQ tables are printed separately after the other
items.
DISPLAY operator
DISPLAY FROM(indd)
ON(p,m,f)
ON(p,m,f,formatting)
ON(p,m,HEX)
ON(VLEN)
ON(VLEN,formatting)
ON(NUM)
ON(NUM,formatting)
LIST(listdd)

TITLE('string')
TITLE('string1','string2')
TITLE('string1','string2','string3')
TLEFT
TFIRST
PAGE

DATE
DATE(abcd)
DATENS(abc)
YDDD(abc)
YDDDNS(ab)
TIME
TIME(abc)
TIMENS(ab)
BLANK
PLUS
NOCC

HEADER('string1')
HEADER('string1','string2')
HEADER('string1','string2','string3')
HEADER(NONE)
NOHEADER
LINES(n)
INDENT(n)

BETWEEN(n)
TBETWEEN(n)
TOTAL('string')
MAXIMUM('string')

MINIMUM('string')
AVERAGE('string')
COUNT('string')

EDCOUNT(formatting)
LIMIT(n)
VSAMTYPE(x)
WIDTH(n)

BREAK(p,m,f)
BREAK(p,m,f,formatting)
594
LONGLINE
BTITLE('string')
BTOTAL('string')
DISPLAY Operator

BMAXIMUM('string')
BMINIMUM('string')
BAVERAGE('string')

BCOUNT('string')
EDBCOUNT(formatting)
STATLEFT
UZERO
LISTSDB
Prints the values or characters of specified numeric fields (including SMF, TOD,
and ETOD date and time) or character fields in a separate list data set. Simple,
tailored, and sectioned reports can be produced. From 1 to 50 fields can be
specified, but the resulting list data set line length must not exceed the limit
specified by the WIDTH operand or 8192 bytes if LONGLINE is specified and
WIDTH is not specified, or 2048 bytes if LONGLINE and WIDTH are not specified.
The record number can be printed as a special field.
DFSORT is called to copy the indd data set to ICETOOL's E35 user exit. ICETOOL
uses its E35 user exit to print appropriate titles, headings and data in the list data
set.
You must not supply your own DFSORT MODS, INREC, or OUTREC statement,
because they would override the DFSORT statements passed by ICETOOL for this
operator.
Specifying formatting items or the PLUS or BLANK operand, which can
compress the columns of output data, can enable you to include more fields in
your report, up to a maximum of 50, if your line length is limited by the
character width your printer or display supports.
Simple report
You can produce a simple report by specifying just the required operands. For
example, if you specify FROM and LIST operands, and ON operands for 10-byte
character and 7-byte zoned decimal fields, the output in the list data set can be
represented as follows:
(p,m,f)
characters
.
.
.
(p,m,f)
sddddddddddddddd
.
.
.
A control character occupies the first byte of each list data set record. Left-justified
standard headings are printed at the top of each page to indicate the contents of
each column, followed by a line for each record showing the characters and
numbers in the fields of that record.
The fields are printed in columns in the same order in which they are specified in
the DISPLAY statement. All fields are left-justified. For numeric fields, leading
zeros are printed, a - is used for the minus sign, and a + is used for the plus sign.
By default, the first column of data starts immediately after the control character,
and three blanks appear between columns. The NOCC operand can be used to
suppress the control character. The INDENT operand can be used to change the
595
DISPLAY Operator
number of blanks before the first column of data. The BETWEEN operand can be
used to change the number of blanks between columns.
The standard column widths are as follows:
v Character data: the length of the character field, or 20 bytes if the field length is
less than 21 bytes
v Numeric data: 16 bytes, or 32 bytes if the numeric field is BI or FI with a length
greater than 4, PD with a length greater than 8, ZD, CSF, FS, UFF or SFF with a
length greater than 15, or FL.
v Record number: 15 bytes
HEADER operands can be used to change or suppress the headings. Formatting
items or the PLUS or BLANK operand can be used to change the appearance of
numeric fields in the report. PLUS, BLANK, and HEADER operands can be used
to change the width of the columns for numeric and character fields and the
justification of headings and fields.
The NOHEADER operand can be used to create list data sets containing only data
records. Data sets created in this way can be processed further by other operators
(for example, STATS or UNIQUE) using CH format for character values or FS
format for numeric values.
The TOTAL, MAXIMUM, MINIMUM, and AVERAGE operands can be used to
print statistics for numeric fields after the columns of data. Formatting items can
be used to suppress the statistics for selected numeric fields. The COUNT operand
can be used to print the record count after the columns of data.
Tailored report
You can tailor the output in the list data set using various operands that control
title strings, date, time, page number, headings, lines per page, field formats, total,
maximum, minimum, and average values for the columns of numeric data, and the
record count. The optional operands can be used in many different combinations to
produce a wide variety of report formats. For example, if you specify FROM, LIST,
BLANK, TITLE, PAGE, DATE, TIME, HEADER and AVERAGE operands, and ON
operands for 10-byte character and 7-byte zoned decimal fields, the output in the
list data set can be represented as follows:
title
- p -
header
---------characters
.
.
.
header
-------sd
.
.
.
average
mm/dd/yy
hh:mm:ss
sd
By default, a control character occupies the first byte of each list data set record.
The NOCC operand can be used to suppress the control characters. The title lines
(up to 3) are printed at the top of each page of the list data set. The first title line
contains the elements you specify (title strings, page number, date and time) in the
order in which you specify them. The second and third title lines contain the title
strings you specify. By default, eight blanks appear between title elements, the title
strings are centered with respect to each other, and the title lines appear on every
page. The TBETWEEN(n) operand can be used to change the number of blanks
596
DISPLAY Operator
between title elements. The TLEFT operand can be used to left-justify the title
strings with respect to each other. The TFIRST operand can be used to only print
the title lines on the first page. A blank line is printed after each title line.
Your specified headings (underlined) are printed after the title line on each page to
indicate the contents of each column, followed by a line for each record showing
the characters and numbers in the fields of that record. Your specified headings
can be one, two or three lines. Headings for character fields are left-justified and
headings for numeric fields are right-justified.
Your specified statistical lines (total, maximum, minimum, average, and count, and
their associated strings) are printed for selected numeric fields and the record
count, after the columns of data. The associated strings can be printed in the first
column or to the left of it.
the DISPLAY statement. Character fields are left-justified and numeric fields are
right-justified. For numeric fields, leading zeros are suppressed, a - is used for the
minus sign, and a blank is used for the plus sign (you can specify PLUS rather
than BLANK if you want a + to be used for the plus sign).
Formatting items can be used to change the appearance of individual numeric
fields in the report with respect to separators, number of digits, decimal point,
decimal places, signs, leading zeros, division by 10, 100, 1000, 10000, 100000,
1000000, 1000000000, 1024, 1048576 (1024*1024), or 1073741824 (1024*1024*1024),
leading strings, floating strings, and trailing strings. Formatting items can also be
used to insert leading or trailing strings for character fields.
The column widths are dynamically adjusted according to the length of the
headings and the maximum number of bytes needed for the character or numeric
data.
Sectioned report
You can produce a sectioned report (simple or tailored) by including a BREAK
operand to indicate the break field used to divide the report into sections. Each set
of sequential input records (which you have previously sorted on the break field
and other fields, as appropriate), with the same value for the specified break field,
results in a corresponding set of data lines that is treated as a section in the report.
The break field is printed at the beginning of each section. Formatting items can be
used to change the appearance of numeric break fields, and to insert a string
before or after character or numeric break fields.
Optional break operands can be used to modify the break title for each section (the
break value is always printed as part of the break title) and to print statistics for
selected numeric fields and the count, in each section. For example, if you add
BTITLE, BREAK, BCOUNT, BMAXIMUM, and BMINIMUM to the operands for
the tailored report discussed previously, each section of the output in the list data
set starts on a new page and can be represented as follows:
597
DISPLAY Operator
title
btitle
- p -
mm/dd/yy
hh:mm:ss
bvalue
header
---------characters
.
.
.
bcount
header
-------sd
.
.
.
d
bmaximum
sd
bminimum
sd
The final page showing the overall statistics starts on a new page and can be
represented as follows:
title
- p -
header
----------
header
--------
average
mm/dd/yy
hh:mm:ss
sd
FROM(indd)
Specifies the ddname of the input data set to be read by DFSORT for this
operation. An indd DD statement must be present and must define an input
data set that conforms to the rules for DFSORT's SORTIN data set. In addition,
the LRECL of the data set must be at least 4.
ON(p,m,f)
Specifies the position, length, and format of a numeric or character field to be
used for this operation. '(p,m,f)' is used for the standard column heading (see
HEADER('string1'), HEADER('string1','string2'),
HEADER('string1','string2','string3'), HEADER(NONE) and NOHEADER for
alternative heading options).
By default, three blanks appear between columns. You can change the space
between columns with BETWEEN(n).
p specifies the first byte of the field relative to the beginning of the input record. p
is 1 for the first data byte of a fixed-length record and 5 for the first data byte of a
variable-length record as illustrated in the following (RRRR represents the 4-byte
record descriptor word):
Fixed-length record
|
Variable-length record
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3 4
| p=
1
2
3
4
5
6
7
8
m specifies the length of the field in bytes. A field must not extend beyond
position 32752 or beyond the end of a record. The maximum length for a field
depends on its format.
f specifies the format of the field as shown in the following.
598
DISPLAY Operator
Format Code
Length
Description
BI
1 to 8 bytes
Unsigned binary
FI
1 to 8 bytes
Signed fixed-point
PD
1 to 16 bytes
ZD
1 to 31 bytes
FL
4 or 8 bytes
Signed hexadecimal
floating-point converted to
signed integer
CH
1 to 4000 bytes
Character
CSF or FS

UFF
SFF
DT1
4 bytes

Z'yyyymmdd'
DT2
4 bytes

Z'yyyymm'
DT3
4 bytes

Z'yyyyddd'
DC1
8 bytes

Z'yyyymmdd'
DC2
8 bytes

Z'yyyymm'
DC3
8 bytes

Z'yyyyddd'
DE1
8 bytes

Z'yyyymmdd'
DE2
8 bytes

Z'yyyymm'
DE3
8 bytes

Z'yyyyddd'
TM1
4 bytes
SMF time interpreted as

Z'hhmmss'
TM2
4 bytes

Z'hhmm'
TM3
4 bytes
TM4
4 bytes

Z'hhmmssxx'
TC1
8 bytes

Z'hhmmss'
TC2
8 bytes

Z'hhmm'
TC3
8 bytes

Z'hh'
TC4
8 bytes

Z'hhmmssxx'
TE1
8 bytes

Z'hhmmss'
599
DISPLAY Operator
Format Code
Length
Description
TE2
8 bytes

Z'hhmm'
TE3
8 bytes

Z'hh'
TE4
8 bytes

Z'hhmmssxx'
Note: See Appendix C, Data

format descriptions, on page
891 for detailed format
descriptions.
For a CSF, FS, UFF, or SFF format field:

v A maximum of 31 digits is allowed. If a value with more than 31 digits is found,
v If a decimal value contains an invalid digit (A-F), ICETOOL identifies the bad
value in a message and prints asterisks for that value, and for the total,
maximum, minimum and average (if specified) for that field, in the list data set.
If the number of bad values reaches the LIMIT for invalid decimal values,
ICETOOL terminates the operation. If the LIMIT operand is not specified, a
default of 200 is used for the invalid decimal value limit.
For an FL format field:
v The normalized or unnormalized FL (hexadecimal floating-point) value is
converted to a signed integer in the range -9223372036854775808 to
9223372036854775807. The fractional part of the FL value is lost, and in some
cases the signed integer may be one of a number of possible signed integers for
the FL value depending on its precision. Converted values less than
-9223372036854775808 are set to -9223372036854775808. Converted values greater
than 9223372036854775807 are set to 9223372036854775807.
v If you are not running in z/Architecture mode, specifying an FL format field
results in an error message and termination.
For a DT1, DT2 or DT3 format field:
v An invalid SMF date can result in a data exception (0C7 ABEND) or an incorrect
ZD date.
For a TM1, TM2, TM3 or TM4 format field:
600
DISPLAY Operator
used for this operation and how the data for this field is to be formatted for
printing. The BLANK operand is automatically in effect.
See ON(p,m,f) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
NOST
Ndd
Udd
/x
specifies formatting items that indicate how the data for this field is to be
formatted for printing. Formatting items can be specified in any order, but each
item can only be specified once. Any formatting item can be specified for a
numeric field, but only L'string' and T'string' can be specified for a character
field.
The column width is dynamically adjusted to accommodate the maximum
bytes to be inserted as a result of all formatting items specified.
mask
specifies an edit mask to be applied to the numeric data for this field.
Thirty-nine pre-defined edit masks are available, encompassing many of the
numeric notations throughout the world with respect to separators, decimal
point, decimal places, signs, and so forth. ICETOOL edits the data according to
the selected mask. If other formatting items are specified but mask is not, the
default mask of A0 is applied to the data.
E'pattern' cannot be specified with a mask.
The attributes of each group of masks is shown in Table 84.
Table 84. Attributes of Edit Masks
Separators
Decimal Places
Positive Sign
Negative Sign
A0
No
blank
A1-A5
Yes
blank
B1-B6
Yes
blank
C1-C6
Yes
blank
D1-D6
Yes
blank
E1-E4
Yes
blank
()
F1-F5
Yes
blank
()
G1-G6
Yes
blank
Table 85 on page 602 describes the available masks and shows how the values
12345678 and -1234567 would be printed for each mask. In the pattern:
601
DISPLAY Operator
v d is used to represent a decimal digit (0-9)
v w is used to represent a leading sign that will be blank for a positive value
or - for a negative value
v x is used to represent a trailing sign that will be blank for a positive value or
- for a negative value
v y is used to represent a leading sign that will be blank for a positive value
or ( for a negative value
v z is used to represent a trailing sign that will be blank for a positive value or
) for a negative value
Table 85. Edit Mask Patterns
Pattern
12345678
-1234567
A0
wddddddddddddddddddddddddddddddd
12345678
-1234567
A1
wd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd
12,345,678
-1,234,567
A2
wd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd
12.345.678
-1.234.567
A3
wd ddd ddd ddd ddd ddd ddd ddd ddd ddd ddd
12 345 678
-1 234 567
A4
wd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd
12'345'678
-1'234'567
A5
d ddd ddd ddd ddd ddd ddd ddd ddd ddd dddx
B1
wddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd.d
1,234,567.8
-123,456.7
B2
wddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd,d
1.234.567,8
-123.456,7
B3
wddd ddd ddd ddd ddd ddd ddd ddd ddd ddd,d
1 234 567,8
-123 456,7
B4
wddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd.d
1'234'567.8
-123'456.7
B5
wddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd,d
1'234'567,8
-123'456,7
B6
ddd ddd ddd ddd ddd ddd ddd ddd ddd ddd,dx
C1
wdd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd.dd
123,456.78
-12,345.67
C2
wdd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd,dd
123.456,78
-12.345,67
C3
wdd ddd ddd ddd ddd ddd ddd ddd ddd ddd,dd
123 456,78
-12 345,67
C4
wdd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd.dd
123'456.78
-12'345.67
C5
wdd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd,dd
123'456,78
-12'345,67
C6
dd ddd ddd ddd ddd d ddd ddd ddd ddd,ddx
D1
wd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd.ddd
12,345.678
-1,234.567
D2
wd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd,ddd
12.345,678
-1.234,567
D3
wd ddd ddd ddd ddd ddd ddd ddd ddd ddd,ddd
12 345,678
-1 234,567
D4
wd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd.ddd
12'345.678
-1'234.567
D5
wd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd,ddd
12'345,678
-1'234,567
D6
d ddd ddd ddd ddd ddd ddd ddd ddd ddd,dddx
12 345,678
1 234,567-
E1
yd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,dddz
12,345,678
(1,234,567)
E2
yd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.dddz
12.345.678
(1.234.567)
E3
yd ddd ddd ddd ddd ddd ddd ddd ddd ddd dddz
12 345 678
(1 234 567)
E4
yd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'dddz
12'345'678
(1'234'567)
F1
ydd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd.ddz
123,456.78
(12,345.67)
F2
ydd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd,ddz
123.456,78
(12.345,67)
F3
ydd ddd ddd ddd ddd ddd ddd ddd ddd ddd,ddz
123 456,78
(12 345,67)
F4
ydd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd.ddz
123'456.78
(12'345.67)
602
12 345 678
1 234 567,8
123 456,78
1 234 567-
123 456,7-
12 345,67-
DISPLAY Operator
Table 85. Edit Mask Patterns (continued)
Pattern
12345678
123'456,78
-1234567
F5
ydd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd,ddz
G1
wddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd,ddd.dddd
1,234.5678
-123.4567
G2
wddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd.ddd,dddd
1.234,5678
-123,4567
G3
wddd ddd ddd ddd ddd ddd ddd ddd ddd,dddd
1 234,5678
-123,4567
G4
wddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd.dddd
1'234.5678
-123.4567
G5
wddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd'ddd,dddd
1'234,5678
-123,4567
G6
ddd ddd ddd ddd ddd ddd ddd ddd ddd,ddddx
1 234,5678
(12'345,67)
123,4567-
If LZ is specified, leading zeros are printed. for example, +1 is shown as

0,000.01 with ON(21,6,FS,C1,LZ).
If LZ is not specified, leading zeros are suppressed except when inappropriate.
For example, +1 is shown as 1 with ON(21,6,FS,A1) and as 0.01 with
ON(21,6,FS,C1).
The leading sign (blank for a positive value or - for a negative value) appears
to the left of the first non-suppressed digit of the formatted value. For
example, -1 is shown as -1 with ON(21,6,FS,A2), as -000.001 with
ON(21,6,FS,A2,LZ) and as -0.01 with ON(21,6,FS,C2).
E'pattern'
specifies an edit pattern to be applied to the numeric data for this field.
E'pattern' is useful for formatting unsigned numeric data such as telephone
numbers, dates, time-of-day, social security numbers, and so on. For example,
0123456789 is shown as (012)-345-6789 with ON(21,10,ZD,E'(999)-999-9999').
The pattern (1 to 44 characters) must be enclosed in single apostrophes. Each 9
in the pattern (up to 31) is replaced by a corresponding digit from the numeric
value. Characters other than 9 in the pattern appear as specified. To include a
single apostrophe (') in the pattern, specify two single apostrophes ('').
F'string' or a mask cannot be specified with E'pattern'.
When E'pattern' is specified for a field:
v Values are shown unsigned. For example, +120622 and -120622 are both
shown as 12:06:22 with ON(12,7,FS,E'99:99:99').
v If the number of significant digits in a value is less than the number of 9's in
the pattern, 0's are filled in on the left. For example, 1234 is shown as
0012-34 with ON(12,6,FS,E'9999-99').
v If the number of significant digits in a value is greater than the number of
9's in the pattern, digits are truncated from the left. For example, 1234567 is
shown as *45:67* with ON(9,4,PD,E'*99:99*').
L'string'
specifies a leading string to appear at the beginning of the character or
numeric data column for this field. For example, 'DFSORT ' is shown as
'**DFSORT ' with ON(1,8,CH,L'**').
F'string'
specifies a floating string to appear to the left of the first non-blank character
of the formatted numeric data for this field. For example, 0001234 is shown as
$12.34 with ON(9,7,ZD,C1,F'$').
603
DISPLAY Operator
E'pattern' cannot be specified with F'string'.
T'string'
specifies a trailing string to appear at the end of the character or numeric data
column for this field. For example, 'DFSORT ' is shown as '**DFSORT ***' with
ON(1,8,CH,L'**',T'***').
LZ specifies that leading zeros are to be printed when the specified edit mask is
applied to the numeric data for this field, overriding the default of suppressing
leading zeros. For example, +123 is shown as 123 with ON(21,6,FS,A0), but as
000123 with ON(21,6,FS,A0,LZ).
LZ is useful for formatting numeric data, such as account numbers, for which
leading zeros must be printed.
Leading zeros are always printed for E'pattern' regardless of whether or not LZ
is specified.
NOST
specifies that requested statistics (TOTAL, MAXIMUM, MINIMUM, AVERAGE,
BTOTAL, BMAXIMUM, BMINIMUM, BAVERAGE) are not to be printed for
this numeric field.
Ndd or Udd
specifies the number of digits to be used for the numeric field. Ndd or Udd
can be used to change the column width for numeric fields, and to prevent
overflow for totals. dd specifies the number of digits and must be a two-digit
number between 01 and 31.
The default number of digits (d) for a numeric field is the maximum number
of digits for that field. For example, d is 5 for ON(1,5,ZD). If you know that
your numeric field requires less than d digits, you can use a lower number of
digits (dd) instead by specifying Udd, thus reducing the column width if it is
determined by d. For example, ON(1,5,ZD,U03) reduces d from 5 to 3. If you
want your numeric field to be displayed with more than d digits, you can use
a higher number of digits (dd) instead by specifying Ndd or Udd, thus
increasing the column width if it is determined by d. For example,
ON(1,5,ZD,U10) increases d from 5 to 10.
The default number of digits (d) for a total is 15 if the numeric field is BI or FI
with a length up to 4, PD with a length up to 8, or ZD, CSF, FS, UFF or SFF
with a length up to 15. The default number of digits (d) for a total is 31 if the
numeric field is BI or FI with a length greater than 4, PD with a length greater
than 8, or ZD, CSF, FS, UFF or SFF with a length greater than 15.
If you know that your total requires less than d digits, you can use a lower
number of digits (dd) instead by specifying Ndd or Udd for the ON field, thus
reducing the column width if it is determined by d. For example,
ON(1,18,ZD,U18) with TOTAL reduces d from 31 to 18. If you know that your
total can overflow d digits, you can use a higher number of digits (dd) instead
by specifying Ndd or Udd, thus preventing overflow and increasing the
column width if it is determined by d. For example, ON(1,15,ZD,U17) with
TOTAL increases d from 15 to 17.
604
DISPLAY Operator
Either Ndd or Udd can be used to set d greater than the maximum for a
numeric field, but only Udd can be used to set d less than the maximum for a
numeric field.
For Udd:
dd is used for d. For example:
If
v
v
v
TOTAL and BTOTAL are not used:

If ON(1,5,ZD) is specified, 5 digits (default for 5,ZD) are used.
If ON(1,5,ZD,U10) is specified, 10 digits (from U10) are used..
If ON(1,5,ZD,U03) is specified, 3 digits (from U03) are used.
v If ON(1,16,FS) is specified, 16 digits (default for 16,FS) are used.

v If ON(1,16,FS,U16) is specified, 16 digits (from U16) are used..
v If ON(1,16,FS,U15) is specified, 15 digits (from U15) are used.
If TOTAL or BTOTAL is used:
v If ON(1,5,ZD) is specified, 15 digits (default for TOTAL or BTOTAL and
5,ZD) are used.
v If ON(1,5,ZD,U10) is specified, 10 digits (from U10) are used.
v If ON(1,16,FS) is specified, 31 digits (default for TOTAL or BTOTAL and
16,FS) are used.
If you use Udd and a numeric value or total overflows dd digits, ICETOOL
prints asterisks for that numeric value or total and terminates the operation.
You can prevent the overflow by specifying an appropriately higher dd value
for Udd. For example, if ON(1,12,ZD,U09) results in overflow, you can use
ON(1,12,ZD,U10) instead.
If E'pattern' is specified, Udd is ignored, because d is determined from the
pattern.
For Ndd:
If dd is greater than or equal d, dd is used. If dd is less than d, d is used. For
example:
If TOTAL and BTOTAL are not used:
v If ON(1,5,ZD) is specified, 5 digits (default for 5,ZD) are used.
v If ON(1,5,ZD,N10) is specified, 10 digits (from N10) are used.
v If ON(1,5,ZD,N03) is specified, 5 digits (from 5,ZD) are used
If TOTAL or BTOTAL is used:
v If ON(1,5,ZD) is specified, 15 digits (default for TOTAL or BTOTAL and
5,ZD) are used.
v If ON(1,5,ZD,N03) is specified, 5 digits (from 5,ZD) are used.
If you use Ndd and a total overflows dd digits, ICETOOL prints asterisks for
the total and terminates the operation. You can prevent the overflow by
specifying an appropriately higher dd value for Ndd. For example, if
ON(1,17,ZD,N17) with TOTAL results in overflow, you can use
ON(1,17,ZD,N18) instead
605
DISPLAY Operator
If E'pattern' is specified, Ndd is ignored, because d is determined from the
pattern.
/x specifies division of the numeric data for this field before formatting. x
indicates the division factor to be used as described later in this section. The
resulting values are rounded down to the nearest integer. Statistics (TOTAL,
MAXIMUM, MINIMUM, AVERAGE, BTOTAL, BMAXIMUM, BMINIMUM,
BAVERAGE) and column widths reflect the divided numbers.
/D
specifies division by 10 before formatting. For example, -1234 is shown

as -123 with ON(11,2,FI,/D).
/C
specifies division by 100 before formatting. For example, 12345 is

shown as 12.3 with ON(11,2,BI,/C,B1).
/K
specifies division by 1000 before formatting. For example, -1234567890

is shown as (1 234 567) with ON(1,11,FS,/K,E3).
/DK
specifies division by 10000 (10*1000) before formatting. For example,

6213849653 is shown as 0-6213-84 with ON(31,10,FS,/DK,E'9-9999-99').
/CK
specifies division by 100000 (100*1000) before formatting. For example,

1234567890123456789012345 is shown as 1,234,567,890,123,456.7890 with
ON(21,25,ZD,G1,/CK).
/M
specifies division by 1000000 (1000*1000) before formatting. For

example, -123456789 is shown as -1.23 with ON(31,10,FS,/M,C4).
/G
specifies division by 1000000000 (1000*1000*1000) before formatting.

For example, 1234567898765 is shown as 1'234 with
ON(15,13,ZD,A4,/G).
/KB
specifies division by 1024 before formatting. For example, 1234567890

is shown as 1 205 632 with ON(45,10,ZD,/KB,A3).
/MB
specifies division by 1048576 (1024*1024) before formatting. For

example, 123456789 is shown as 117 with ON(60,9,FS,/MB).
/GB
specifies division by 1073741824 (1024*1024*1024) before formatting.

For example, 1234567898765 is shown as 1,149 with
ON(15,13,ZD,/GB,A1).
ON(p,m,HEX)
Specifies the position and length of a character field to be used for this
operation and printed in hexadecimal format (00-FF for each byte). '(p,m,HEX)'
is used for the standard column heading. See HEADER('string1'),
HEADER('string1','string2'), HEADER('string1','string2','string3'),
HEADER(NONE), and NOHEADER for alternative heading options.
See ON(p,m,f) for a discussion of p.
position 32752 or beyond the end of a record. A field can be 1 to 2000 bytes.
ON(VLEN)
Equivalent to specifying ON(1,2,BI); a two-byte binary field starting at position
1. For variable-length records, ON(VLEN) represents the record-length for each
record. ' RECORD LENGTH' is used for the standard column heading. See
HEADER('string1','string2','string3'), HEADER(NONE), and NOHEADER for
alternative heading options.
ON(VLEN,formatting)
Equivalent to specifying ON(1,2,BI,formatting); a two-byte binary field starting
606
DISPLAY Operator
at position 1, and how the data for this field is to be formatted for printing.
The BLANK operand is automatically in effect.
See ON(VLEN) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
NOST
Ndd
Udd
/x
item can only be specified once.
See ON(p,m,f,formatting) for a discussion of formatting.
ON(NUM)
Specifies that the record number is to be printed. The record number starts at 1
and is incremented by 1 for each record printed in the list data set. 'RECORD
NUMBER' is used for the standard column heading. See HEADER('string1'),
HEADER(NONE), and NOHEADER for alternative heading options.
ON(NUM,formatting)
Specifies that the record number is to be printed, and how the record number
is to be formatted for printing. The BLANK operand is automatically in effect.
See ON(NUM) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
Ndd
Udd
specifies formatting items that indicate how the record number is to be

607
DISPLAY Operator
mask
See ON(p,m,f,formatting) for a discussion of mask.
E'pattern'
specifies an edit pattern to be applied to the record number. The pattern (1
to 24 characters) must be enclosed in single apostrophes. Each 9 in the
pattern (up to 15) is replaced by a corresponding digit from the numeric
value. Characters other than 9 in the pattern appear as specified. To
include a single apostrophe (') in the pattern, specify two single
apostrophes ('').
When E'pattern' is specified for the record number:
v If the number of significant digits in a record number is less than the
number of 9's in the pattern, 0's are filled in on the left. For example,
1234 is shown as 001234 with ON(NUM,E'999999').
v If the number of significant digits in a record number is greater than the
example, 1234567 is shown as *4567* with ON(NUM,E'*9999*').
L'string'
See ON(p,m,f,formatting) for a discussion of L'string'.
F'string'
See ON(p,m,f,formatting) for a discussion of F'string'.
T'string'
See ON(p,m,f,formatting) for a discussion of T'string'.
LZ See ON(p,m,f,formatting) for a discussion of LZ.
Ndd or Udd
Specifies the number of digits to be used for the record number when
determining the column width. dd specifies the number of digits and must
be a two-digit number between 01 and 15.
The default number of digits (d) for the record number is 15. If you know
that your record numbers require less than 15 digits, you can use a lower
number of digits (dd) instead by specifying Ndd or Udd thus reducing the
column width if it is determined by d. For example, if ON(NUM,N09) or
ON(NUM,U09) is specified, 9 digits (from N09 or U09) is used instead of
15 (default for record number).
If you use Ndd or Udd and the number of records overflows the number
of digits used, ICETOOL terminates the operation. You can prevent the
overflow by specifying an ppropriately higher dd value for Ndd or Udd.
For example, if ON(NUM,N05) results in overflow, you can use
ON(NUM,N06) instead.
If E'pattern' is specified, Ndd or Udd is ignored, because d is determined
from the pattern.
LIST(listdd)
Specifies the ddname of the list data set to be produced by ICETOOL for
this operation. A listdd DD statement must be present. ICETOOL sets the
attributes of the list data set as follows:
v If NOCC is not specified, RECFM is set to FBA. If NOCC is specified,
RECFM is set to FB.
v LRECL is set to one of the following:
608
DISPLAY Operator
If WIDTH(n) is specified, LRECL is set to n. Use WIDTH(n) if your
LRECL must be set to a particular value (for example, if you use
DISP=MOD to place several reports in the same data set).
If WIDTH(n) is not specified and NOCC is not specified, LRECL is set
to 121 or to the calculated required line length if it is greater than 121
characters.
If WIDTH(n) is not specified and NOCC is specified, LRECL is set to
120 or to the calculated required line length if it is greater than 120
characters.
If your LRECL does not need to be set to a particular value, you can let
ICETOOL determine and set the appropriate LRECL value by not
specifying WIDTH(n).
v BLKSIZE is set to one of the following:
The BLKSIZE from the DD statement, DSCB, or label, if it is a
multiple of the LRECL used.
The LRECL if the BLKSIZE from the DD statement, DSCB, or label is
not a multiple of the LRECL used.
The block size as directed by LISTSDB or LISTNOSDB if specified, or
otherwise as directed by the SDBMSG installation option from
ICEAM2 or ICEAM4 (see z/OS DFSORT Installation and Customization),
if the BLKSIZE is not available from the DD statement, DSCB, or
label.
TITLE('string')
Specifies printing of a title string in a title line. The title string is of the
form:
string
Up to three TITLE operands can be specified.

The first TITLE operand, if specified, supplies a string for the first title line.
Other title elements (PAGE for page number, DATE, DATE(abcd),
DATENS(abc), YDDD(abc) or YDDDNS(ab) for the date, and TIME,
TIME(abc) or TIMENS(ab) for the time) can also be specified for the first
title line with or without a title string.
The second TITLE operand, if specified, specifies a string for the second
title line. The third TITLE operand, if specified, specifies a string for the
third title line.
If you specify any title elements (title string, page number, date or time),
the first title line is printed at the top of each page of the list data set. It
contains the title elements you specify in the order in which you specify
them. By default, eight blanks appear between title elements on the first
title line. You can change the space between title elements on the first title
line with TBETWEEN(n). A blank line is printed after the first title line.
If you specify a second TITLE operand, the second title line containing the
specified title string is printed after the first title line. A blank line is
printed after the second title line. If you specify a third TITLE operand, the
third title line containing the specified title string is printed after the
second title line. A blank line is printed after the third title line.
609
DISPLAY Operator
By default, the title strings in the title lines are centered with respect to
each other. TLEFT can be used to left-justify the title string in the title lines
with respect to each other.
By default, the title lines are printed on every page. TFIRST can be used to
only print the title lines on the first page.
include a single apostrophe (') in the string, specify two single apostrophes
(''). Blanks at the start of the string move the text to the right. Blanks at the
end of the string increase the spacing between the string and the next title
element.
Specifies printing of a two part title string in a title line. The title string is
of the form:
string1string2
Note: string1 and string2 can be any combination of inline constants,

DFSORT symbol constants, and DFSORT symbol constants for system
information (for example, S'&Sysname').
See TITLE('string') for additional informaton on title lines.
Each string (1 to 50 characters) must be enclosed in single apostrophes. The
total combined length of string1 and string2 can be up to 50 characters. To
include a single apostrophe (') in a string, specify two single apostrophes
(''). Blanks at the start of a string move the text to the right. Blanks at the
end of string1 increase the spacing between string1 and string2. Blanks at
the end of string2 increase the spacing between string2 and the next title
element.
Specifies printing of a three part title string in a title line. The title string is
of the form:
string1string2string3
Note: string1, string2 and string3 can be any combination of inline

constants, DFSORT symbol constants, and DFSORT symbol constants for
system information (for example, S'&Sysname').
See TITLE('string') for additional informaton on title lines.
Each string (1 to 50 characters) must be enclosed in single apostrophes. The
total combined length of string1, string2 and string3 can be up to 50
characters. To include a single apostrophe (') in a string, specify two single
apostrophes (''). Blanks at the start of a string move the text to the right.
Blanks at the end of string1 increase the spacing between string1 and
string2. Blanks at the end of string2 increase the spacing between string2
and string3. Blanks at the end of string3 increase the spacing between
string3 and the next title element.
TLEFT
Specifies that the title strings in each title line are to be left justified with
respect to each other (overriding the default of centering the title strings
with respect to each other). See TITLE('string') for additional informaton on
title lines.
TFIRST
Specifies that the title lines are only to appear on the first page of the
610
DISPLAY Operator
report (overriding the default of having the title lines appear on every
page of the report). See TITLE('string') for additional informaton on title
lines.
PAGE
Specifies printing of the page number in the first title line. The page
number is printed in the form - p - where p is in decimal with no leading
zeros. The page number is 1 for the first page and is incremented by 1 for
each subsequent page. See TITLE('string') for additional information on the
first title line.
DATE
Specifies printing of the date in the first title line. The date is printed in the
form mm/dd/yy where mm is the month, dd is the day, and yy is the
year. DATE is equivalent to specifying DATE(MDY/). See TITLE('string')
for additional information on the first title line.
DATE(abcd)
form 'adbdc' according to the specified values for abc and d. For example,
on March 29, 2002, DATE(4MD-) would produce '2002-03-29' and
DATE(MDY.) would produce '03.29.02'. See TITLE('string') for additional
information on the first title line.
abc can be any combination of M, D, and Y or 4 (each specified once)
where M represents the month (01-12), D represents the day (01-31), Y
represents the last two digits of the year (for example, 02), and 4 represents
the four digits of the year (for example, 2002).
d can be any character and is used to separate the month, day, and year.
DATENS(abc)
form 'abc' according to the specified values for abc. For example, on March
29, 2002, DATENS(4MD) would produce '20020329' and DATENS(MDY)
would produce '032902'. See TITLE('string') for additional information on
the first title line.
abc can be any combination of M, D, and Y or 4 (each specified once)
where M represents the month (01-12), D represents the day (01-31), Y
represents the last two digits of the year (for example, 02), and 4 represents
the four digits of the year (for example, 2002).
YDDD(abc)
specifies printing of the date in the first title line. The date is printed in the
form 'acb' according to the specified values for ab and c. For example, on
April 7, 2004, YDDD(DY-) would produce '098-04' and YDDD(4D/) would
produce '2004/098'. See TITLE('string') for additional information on the
first title line.
ab can be any combination of D, and Y or 4 (each specified once) where D
represents the day of the year (001-366), Y represents the last two digits of
the year (for example, 04), and 4 represents the four digits of the year (for
example, 2004).
c can be any character and is used to separate the month, day and year.
YDDDNS(ab)
specifies printing of the date in the first title line. The date is printed in the
form 'ab' according to the specified values for ab. For example, on April 7,
611
DISPLAY Operator
2004, YDDDNS(DY) would produce '09804' and YDDD(4D) would produce
'2004098'. See TITLE('string') for additional information on the first title
line.
ab can be any combination of D, and Y or 4 (each specified once) where D
represents the day of the year (001-366), Y represents the last two digits of
the year (for example, 04), and 4 represents the four digits of the year (for
example, 2004).
TIME
Specifies printing of the time in the first title line. The time is printed in
the form hh:mm:ss where hh is hours, mm is minutes and ss is seconds.
TIME is equivalent to specifying TIME(24:). See TITLE('string') for
additional information on the first title line.
TIME(abc)
the form 'hhcmmcss xx' according to the specified value for ab and c. For
example, at 08:25:13 pm, TIME=(24:) would produce '20:25:13' and
TIME=(12.) would produce '08.25.13 pm'. See TITLE('string') for additional
information on the first title line.
ab can be:
v 12 to indicate 12-hour time. hh (hours) is 01-12, mm (minutes) is 00-59,
ss (seconds) is 00-59 and xx is am or pm.
ss (seconds) is 00-59 and xx is not included.
c can be any character and is used to separate the hours, minutes, and
seconds.
TIMENS(ab)
the form 'hhmmss xx' according to the specified value for ab. For example,
at 08:25:13 pm, TIMENS=(24) would produce '202513' and TIMENS=(12)
would produce '082513 pm'. See TITLE('string') for additional information
on the first title line.
ab can be:
ss (seconds) is 00-59 and xx is am or pm.
ss (seconds) is 00-59 and xx is not included.
BLANK
Specifies an alternate format for printing character and numeric data as
follows:
v Numeric values for which formatting is not specified are printed with
blank for plus sign, - for minus sign and no leading zeros (overriding
the default of + for plus sign and leading zeros).
Numeric values are thus displayed as:
d...d for positive values (blank sign immediately to the left of the
digits and no leading zeros)
-d...d for negative values (- sign immediately to the left of the digits
and no leading zeros)
v Column widths are dynamically adjusted according to the length of the
headings and the maximum number of bytes needed for the character or
numeric data
612
DISPLAY Operator
v Headings and data for numeric fields are right-justified (overriding the
default of left-justified headings and data for numeric fields)
PLUS
Specifies an alternate format for printing character and numeric data as
follows:
v Numeric values for which formatting is not specified are printed with +
for plus sign, - for minus sign and no leading zeros (overriding the
default of leading zeros).
Numeric values are thus displayed as:
+d...d for positive values (- sign immediately to the left of the digits
-d...d for negative values (- sign immediately to the left of the digits
v Column widths are dynamically adjusted according to the length of the
headings and the maximum number of bytes needed for the character or
numeric data
v Headings and data for numeric fields are right-justified (overriding the
default of left-justified headings and data for numeric fields)
For ON(NUM), PLUS is treated as BLANK.
NOCC
Specifies that carriage control characters are not to be included in the lines
of the list data set (overriding the default of using a carriage control
character as the first byte of each line). A blank line is used instead of a
page eject control character to separate elements of the report (such as
sections).
The RECFM of the list data set is set to FB.
The LRECL of the list data set will not include a byte for the carriage
control character. If the line length is less than or equal to 120 bytes, the
LRECL will be set to 120. If the line length is greater than 120 bytes, the
LRECL will be set to the line length. The maximum line length is 2047
bytes if LONGLINE is not specified, or 8191 bytes if LONGLINE is
specified.
If WIDTH(n) is specified, n can be 121 to 8191.
HEADER('string1')
Specifies a heading to be printed for the corresponding ON field. The
specified string is used instead of the standard column heading for the
corresponding ON field. (ON fields and HEADER operands correspond
one-for-one according to the order in which they are specified; that is, the
first HEADER operand corresponds to the first ON field, the second
HEADER operand corresponds to the second ON field, and so on.)
(''). If the string length is greater than the column width for the
corresponding ON field, the column width is increased to the string length.
The heading is left-justified for character fields or right-justified for
numeric fields and is underlined with hyphens for the entire column width
(overriding the default of left-justified, non-underlined headings).
Character values are left-justified and numeric values are right-justified
(overriding the default of left-justified field values).
A null string ('') or blank string (' ') may be used to set string1 to blanks.
613
DISPLAY Operator
Blanks at the start or end of a heading string may alter the justification of
the heading or the width of the column.
If HEADER('string1') is used for any ON field, HEADER('string1'),
HEADER('string1','string2'), HEADER('string1','string2','string3'), or
HEADER(NONE) must be used for each ON field.
Specifies a heading to be printed for the corresponding ON field. A
two-line heading is used with string1 on line1 and string2 on line2. A null
string ('') or blank string (' ') may be used to set string1 or string2 to
blanks. A comma (,) may also be used to set string1 to blanks. For
example, HEADER(,'string1') results in blanks for this heading in line1 and
string1 for this heading in line2.
If HEADER('string1','string2') is used for any ON field, HEADER('string1'),
HEADER('string1','string2'), HEADER('string1','string2','string3') or
HEADER(NONE) must be used for each ON field.
If a HEADER('string1','string2') operand is specified, a two-line heading is
also used for any HEADER('string1') operands you specify, with string1 for
that heading on line1 and blanks for that heading on line2.
HEADER(,'string1') can be used to put blanks for that heading on line1
and string1 for that heading on line2.
See HEADER('string1') for more details on the HEADER operand.
Specifies a heading to be printed for the corresponding ON field. A
three-line heading is used with string1 on line1, string2 on line2 and
string3 on line3. A null string ('') or blank string (' ') may be used to set
string1, string2 or string3 to blanks. A comma (,) may also be used to set
string1 or string2 to blanks. For example, HEADER(,,'string1') results in
blanks for this heading in line1 and line2 and string1 for this heading in
line3.
If HEADER('string1','string2','string3') is used for any ON field,
HEADER('string1','string2','string3') or HEADER(NONE) must be used for
each ON field.
If a HEADER('string1','string2','string3') operand is specified:
v a three-line heading is also used for any HEADER('string1') operands
you specify, with string1 for that heading on line1 and blanks for that
heading on line2 and line3. HEADER(,,'string1') can be used to put
blanks for that heading on line1 and line2 and string1 for that heading
on line3.
v a three-line heading is also used for any HEADER('string1','string2')
operands you specify, with string1 for that heading on line1, string2 for
that heading on line2 and blanks for that heading on line3.
HEADER(,'string1','string2') can be used to put blanks for that heading
on line1, string1 for that heading on line2 and string2 for that heading
on line3.
See HEADER('string1') for more details on the HEADER operand.
HEADER(NONE)
Specifies that a heading is not to be printed for the corresponding ON
field. The standard column heading for the corresponding ON field is
suppressed.
614
DISPLAY Operator
If HEADER(NONE) is used for any ON field, HEADER('string1'),
HEADER('string1','string2'), HEADER('string1','string2','string3'), or
HEADER(NONE) must be used for each ON field. Specifying
HEADER(NONE) for every ON field is equivalent to specifying
NOHEADER.
NOHEADER
Specifies that headings for ON fields are not to be printed (overriding the
default of printing standard headings for ON fields).
If NOHEADER is used, it must be specified only once and
HEADER('string1','string2','string3'), and HEADER(NONE) must not be
used.
If NOHEADER is specified without any TITLE, DATE, TIME, or PAGE
operands, the resulting list data set contains only data records. Data sets
created in this way can be processed further by other operators (for
example, STATS or UNIQUE) using CH for character values or FS for
numeric values.
LINES(n)
Specifies the number of lines per page for the list data set (overriding the
default of 58). n must be greater than 9, but less than 1000.
INDENT(n)
Specifies the number of blanks to be used to indent the report (overriding
the default of 0). n can be from 0 to 50. For example, if INDENT(n) is not
specified, the report starts in column 2 (after the control character),
whereas if INDENT(10) is specified, the report starts in column 12 (after
the control character and 10 blanks).
BETWEEN(n)
Specifies the number of blanks to be used between the columns of data
(overriding the default of 3). n can be from 0 to 50. For example, if
BETWEEN(n) is not specified, three blanks appear between columns,
whereas if BETWEEN(7) is specified, seven blanks appear between
columns.
TBETWEEN(n)
Specifies the number of blanks to be used between title elements (title
strings, page number, date and time) in the first title line (overriding the
default of 8). n can be from 0 to 50. For example, if TBETWEEN(n) is not
specified, eight blanks appear between the title strings and page number in
the first title line, whereas if TBETWEEN(4) is specified, four blanks appear
between the title strings and page number in the first title line.
TOTAL('string')
Specifies an overall TOTAL line is to be printed after the rows of data for
the report. The specified string is printed starting at the indent column of
the overall TOTAL line, followed by the overall total for each numeric data
column. If STATLEFT is specified, the string is printed to the left of the
first column of data with the totals on the same line as the string. If
STATLEFT is not specified, the string is printed in the first column of data
with the totals on the same line as the string, or on the next line, as
appropriate. A blank line is printed before the overall TOTAL line.
(''). To suppress printing of a string, specify TOTAL('') using two single
apostrophes.
615
DISPLAY Operator
The overall total for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The total for a
specific numeric field is suppressed if the NOST formatting item is
specified for that field. Totals are printed for ON(VLEN) fields, but not for
ON(NUM) fields.
The default number of digits (d) for a total is 15 if the ON field is BI or FI
with a length up to 4, PD with a length up to 8, or ZD, CSF, FS, UFF or
SFF with a length up to 15. The default number of digits (d) for a total is
31 if the ON field is BI or FI with a length greater than 4, PD with a length
greater than 8, ZD, CSF, FS, UFF or SFF with a length greater than 15, or
FL. By default, column widths are adjusted to allow for a maximum of a
sign and d digits for the totals. If the overall total for an ON field
overflows d digits, ICETOOL prints asterisks for the overall total for that
field and terminates the operation.
You can use the Ndd or Udd formatting item to decrease or increase the
number of digits used for a total. If you use Ndd or Udd and the overall
total for an ON field overflows dd digits, ICETOOL prints asterisks for the
overall total for that field and terminates the operation.
You can prevent overflow by specifying an appropriate dd value for Ndd
or Udd. For example, if ON(1,15,ZD) with TOTAL overflows the default of
15 digits, you can specify ON(1,15,ZD,U16) to prevent overflow. If
ON(1,15,ZD,U16) still results in overflow, you can specify
ON(1,15,ZD,U17), and so on.
Either Ndd or Udd can be used to set the number of digits greater than the
maximum for a numeric field, but only Udd can be used to set the number
of digits less than the maximum for a numeric field.
See the discussion of Ndd or Udd under ON(p,m,f,formatting) for more
details on using Ndd or Udd with TOTAL.
The TOTAL, MAXIMUM, MINIMUM, AVERAGE, and COUNT lines are
printed in the order in which you specify them.
MAXIMUM('string')
Specifies an overall MAXIMUM line is to be printed after the rows of data
for the report. The specified string is printed starting at the indent column
of the overall MAXIMUM line, followed by the overall maximum for each
numeric data column. If STATLEFT is specified, the string is printed to the
left of the first column of data with the maximums on the same line as the
string. If STATLEFT is not specified, the string is printed in the first
column of data with the maximums on the same line as the string, or on
the next line, as appropriate. A blank line is printed before the overall
MAXIMUM line.
(''). To suppress printing of a string, specify MAXIMUM('') using two
single apostrophes.
The overall maximum for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The maximum for a
specified for that field. Maximums are printed for ON(VLEN) fields, but
not for ON(NUM) fields.
616
DISPLAY Operator
MINIMUM('string')
Specifies an overall MINIMUM line is to be printed after the rows of data
of the overall MINIMUM line, followed by the overall minimum for each
left of the first column of data with the minimums on the same line as the
column of data with the minimums on the same line as the string, or on
the next line, as appropriate. A blank line is printed before the overall
MINIMUM line.
(''). To suppress printing of a string, specify MINIMUM('') using two single
apostrophes.
The overall minimum for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The minimum for a
specified for that field. Minimums are printed for ON(VLEN) fields, but
AVERAGE('string')
Specifies an overall AVERAGE line is to be printed after the rows of data
of the overall AVERAGE line, followed by the overall average for each
left of the first column of data with the averages on the same line as the
column of data with the averages on the same line as the string, or on the
next line, as appropriate. A blank line is printed before the overall
AVERAGE line.
The overall average (or mean) is calculated by dividing the overall total by
the number of values in the report and rounding down to the nearest
integer (examples: 23 / 5 = 4, -23 / 5 = -4).
('). To suppress printing of a string, specify AVERAGE('') using two single
apostrophes.
The overall average for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The average for a
specified for that field. Averages are printed for ON(VLEN) fields, but not
for ON(NUM) fields.
number of digits used for a total. If the overall total for an ON field
overflows d digits, ICETOOL prints asterisks for the overall average for
that field and terminates the operation. You can prevent overflow by
specifying an appropriate dd value for Ndd or Udd. For example, if
ON(1,15,ZD) with AVERAGE overflows the default of 15 digits for the
total, you can specify ON(1,15,ZD,U16) to prevent overflow.
617
DISPLAY Operator
details on using Ndd or Udd.
COUNT('string')
Specifies an overall COUNT line is to be printed after the rows of data for
the report. The specified string is printed starting at the indent column of
the overall COUNT line, followed by the overall count of data records. If
STATLEFT is specified, the string is printed to the left of the first column
of data. If STATLEFT is not specified, the string is printed in the first
column of data. The count is printed on the same line as the string. A
blank line is printed before the overall COUNT line.
(''). To suppress printing of a string, specify COUNT('') using two single
apostrophes.
The count is printed in the format (PLUS, BLANK, or standard) you
specify. EDCOUNT(formatting) can be used to apply formatting items to
the count. The default number of digits (d) for the count is 15.
EDCOUNT(formatting)
Specifies how the overall count is to be formatted for printing. Formatting
items can be specified in any order, but each item can only be specified
once. EDCOUNT can only be specified if COUNT('string') is specified.
mask
E'pattern'
specifies an edit pattern to be applied to the count. The pattern (1 to 24
characters) must be enclosed in single apostrophes. Each 9 in the
pattern (up to 15) is replaced by a corresponding digit from the count.
Characters other than 9 in the pattern appear as specified. To include a
When E'pattern' is specified for the count:
v If the number of significant digits in the count is less than the
1234 is shown as 001234 with EDCOUNT(E'999999').
v If the number of significant digits in the count is greater than the
example, 1234567 is shown as *4567* with EDCOUNT(E'*9999*').
L'string'
F'string'
T'string'
618
DISPLAY Operator
Udd
specifies the number of digits to be used for the count. dd specifies the
number of digits and must be a two-digit number between 01 and 15.
The default number of digits (d) for the count is 15.
If you know that your count requires less than 15 digits, you can use a
lower number of digits (dd) instead by specifying Udd. For example, if
EDCOUNT(U09) is specified, 9 digits (from U09) is used instead of 15
(default for the count).
If you use Udd and the count overflows the number of digits used,
ICETOOL terminates the operation. You can prevent the overflow by
specifying an appropriately higher dd value for Udd. For example, if
EDCOUNT(U05) results in overflow, you can use EDCOUNT(U06)
instead.
If E'pattern' is specified, Udd is ignored, because d is determined from
the pattern.
LIMIT(n)
Specifies a limit for the number of invalid decimal values (overriding the
default of 200). If n invalid decimal values are found, ICETOOL terminates
the operation. n can be 1 to 15 decimal digits, but must be greater than 0.
VSAMTYPE(x)
See the discussion of this operand on the COPY statement in COPY
operator on page 575.
WIDTH(n)
Specifies the line length and LRECL you want ICETOOL to use for your
list data set. If NOCC is not specified, n can be from 121 to 8192. If NOCC
is specified, n can be from 121 to 8191.
ICETOOL always calculates the line length required to print all titles,
headings, data, and statistics and uses it as follows:
v If WIDTH(n) is specified and the calculated line length is less than or
equal to n, ICETOOL sets the line length and LRECL to n.
v If WIDTH(n) is specified and the calculated line length is greater than n,
v If WIDTH(n) is not specified, NOCC is not specified, and the calculated
line length is less than or equal to 121, ICETOOL sets the line length and
LRECL to 121.
v If WIDTH(n) is not specified, NOCC is specified, and the calculated line
length is less than or equal to 120, ICETOOL sets the line length and
LRECL to 120.
length is between 121 and 2047, or between 121 and 8191 if LONGLINE
is specified, ICETOOL sets the line length and LRECL to the calculated
line length.
line length is between 122 and 2048, or between 122 and 8192 if
LONGLINE is specified, ICETOOL sets the line length and LRECL to the
calculated line length.
line length is greater than 2048, or greater than 8192 if LONGLINE is
specified, ICETOOL issues an error message and terminates the
operation.
619
DISPLAY Operator
length is greater than 2047, or greater than 8191 if LONGLINE is
specified, ICETOOL issues an error message and terminates the
operation.
Use WIDTH(n) if your LRECL must be set to a particular value (for
example, if you use DISP=MOD to place several reports in the same data
set) or if you want to ensure that the line length for your report does not
exceed a specific maximum (for example, 133 bytes). Otherwise, you can
let ICETOOL calculate and set the appropriate line length and LRECL by
not specifying WIDTH(n).
LONGLINE
Specifies the calculated line length can be a maximum of 8191 bytes if
NOCC is specified, or a maximum of 8192 bytes if NOCC is not specified
(overriding the defaults of 2047 bytes if NOCC is specified or 2048 bytes if
NOCC is specified). Use LONGLINE if you do not specify WIDTH(n) and
your calculated line length is larger than the default.
BREAK(p,m,f)
Specifies a numeric or character break field to be used to divide the report
into sections. Each set of sequential input records, with the same value for
the specified break field, results in a corresponding set of data lines that is
treated as a section in the report. The DISPLAY operator should be
preceded by a SORT operator (or another application) that sorts the break
field and any other appropriate fields in the desired sequence for the
report.
Each section starts on a new page. Each page of a section includes a break
title line showing the break value for the section. Numeric break values are
printed with blank for plus sign, - for minus sign, and no leading zeros.
BTITLE can be used to specify a string to appear in the break title line. The
break value and break title string appear in the order in which you specify
BREAK and BTITLE. Two blanks appear between break title elements. A
blank line is printed after the break title line.
BTOTAL, BMAXIMUM, BMINIMUM, BAVERAGE, and BCOUNT can be
used to produce break statistics for each numeric ON field and for the
break count-for example, the maximum of the values in the section for
ON(5,3,ZD) and the maximum of the values in the section for ON(22,2,BI).
The break statistics for each section are printed at the end of the section
(on one or more pages that include the break title). TOTAL, MAXIMUM,
MINIMUM, AVERAGE, and COUNT can be used to produce overall
statistics for each numeric ON field and for the overall count-for example,
the maximum of the values in the report for ON(5,3,ZD) and the maximum
of the values in the report for ON(22,2,BI). The overall statistics for each
section are printed at the end of the report (on a separate page that does
not include the break title).
See ON(p,m,f) for a discussion of p and m.
f specifies the format of the field as shown for ON(p,m,f).
Note: An FL (hexadecimal floating-point) field can be specified for ON,
but not for BREAK.
For a CSF, FS, UFF, or SFF format break field:
v A maximum of 31 digits is allowed. If a value with more than 31 digits
is found, ICETOOL issues an error message and terminates the
operation.
620
DISPLAY Operator
For a ZD or PD format break field:
v If a decimal value with an invalid digit (A-F) is found, ICETOOL issues
an error message and terminates the operation.
v An invalid SMF date can result in a data exception (0C7 ABEND) or an
incorrect ZD date.
BREAK(p,m,f,formatting)
Specifies a numeric or character break field to be used to divide the report
into sections, and how the data for this field is to be formatted for
printing.
See BREAK(p,m,f) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
Udd
specifies formatting items that indicate how the record number is to be

formatted for printing. Formatting items can be specified in any order, but
each item can only be specified once. Any formatting item can be specified
for a numeric break field, but only L'string' and T'string' can be specified
for a character break field.
mask
E'pattern'
See ON(p,m,f,formatting) for a discussion of E'pattern'.
L'string'
F'string'
T'string'
621
DISPLAY Operator
Udd
specifies the number of digits to be used for a numeric break field.
Udd can be used to change the column width for numeric break fields.
dd specifies the number of digits and must be a two-digit number
between 01 and 31.
The default number of digits (d) for a numeric break field is the
maximum number of digits for that field. For example, d is 8 for
BREAK(1,8,ZD). If you know that your break field requires less than d
digits, you can use a lower number of digits (dd) instead by specifying
Udd, thus reducing the break field width. For example,
BREAK(1,8,ZD,U06) reduces d from 8 to 6. If you want your break
field to be displayed with more than d digits, you can use a higher
number of digits (dd) instead by specifying Udd, thus increasing the
field width. For example, BREAK(1,8,ZD,U11) increases d from 8 to 11.
BTITLE('string')
Specifies a string to appear in the break title line printed for each page of a
section. BTITLE can only be specified if BREAK is specified. The break
value and break title string appear in the order in which you specify
BREAK and BTITLE. Two blanks appear between break title elements. A
blank line is printed after the break title line.
(''). Blanks at the start of the string move the text to the right. Blanks at the
end of the string increase the spacing between the string and the break
value if BTITLE is specified before BREAK.
BTOTAL('string')
Specifies a break TOTAL (BTOTAL) line is to be printed after the rows of
data for each section. BTOTAL can only be specified if BREAK is specified.
The specified string is printed starting at the indent column of the break
TOTAL line, followed by the break total for each numeric data column.If
of data with the totals on the same line as the string. If STATLEFT is not
specified, the string is printed in the first column of data with the totals on
the same line as the string, or on the next line, as appropriate. A blank line
is printed before the break TOTAL line.
(''). To suppress printing of a string, specify BTOTAL('') using two single
apostrophes.
The break total for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The total for a
specified for that field. Totals are printed for ON(VLEN) fields, but not for
ON(NUM) fields.
The default number of digits (d) for a break total is 15 if the ON field is BI
or FI with a length up to 4, PD with a length up to 8, or ZD, CSF, FS, UFF
or SFF with a length up to 15. The default number of digits (d) for a break
total is 31 if the ON field is BI or FI with a length greater than 4, PD with
a length greater than 8, or ZD, CSF, FS, UFF or SFF with a length greater
than 15. By default, column widths are adjusted to allow for a maximum
622
DISPLAY Operator
of a sign and d digits for the totals. If the break total for an ON field
overflows d digits, ICETOOL prints asterisks for the break total for that
field and terminates the operation.
number of digits used for a break total. If you use Ndd or Udd and the
break total for an ON field overflows dd digits, ICETOOL prints asterisks
for the break total for that field and terminates the operation.
You can prevent overflow by specifying an appropriate dd value for Ndd
or Udd. For example, if ON(1,15,ZD) with BTOTAL overflows the default
of 15 digits, you can specify ON(1,15,ZD,U16) to prevent overflow. If
ON(1,15,ZD,U16) still results in overflow, you can specify
ON(1,15,ZD,U17), and so on.
Either Ndd or Udd can be used to set the number of digits greater than the
maximum for a numeric field, but only Udd can be used to set the number
of digits less than the maximum for a numeric field.
details on using Ndd or Udd with BTOTAL.
The BTOTAL, BMAXIMUM, BMINIMUM, BAVERAGE, and BCOUNT lines
are printed in the order in which you specify them.
BMAXIMUM('string')
Specifies a break MAXIMUM line is to be printed after the rows of data for
each section. BMAXIMUM can only be specified if BREAK is specified. The
specified string is printed starting at the indent column of the break
MAXIMUM line, followed by the break maximum for each numeric data
first column of data with the maximums on the same line as the string. If
with the maximums on the same line as the string, or on the next line, as
appropriate. A blank line is printed before the break MAXIMUM line.
(''). To suppress printing of a string, specify BMAXIMUM('') using two
single apostrophes.
The break maximum for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The maximum for a
specified for that field. Maximums are printed for ON(VLEN) fields, but
BMINIMUM('string')
Specifies a break MINIMUM line is to be printed after the rows of data for
each section. BMINIMUM can only be specified if BREAK is specified. The
MINIMUM line, followed by the break minimum for each numeric data
first column of data with the minimums on the same line as the string. If
with the minimums on the same line as the string, or on the next line, as
appropriate. A blank line is printed before the break MINIMUM line.
623
DISPLAY Operator
(''). To suppress printing of a string, specify BMINIMUM('') using two
single apostrophes.
The break minimum for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The minimum for a
specified for that field. Minimums are printed for ON(VLEN) fields, but
BAVERAGE('string')
Specifies a break AVERAGE line is to be printed after the rows of data for
each section. BAVERAGE can only be specified if BREAK is specified. The
AVERAGE line, followed by the break average for each numeric data
first column of data with the averages on the same line as the string. If
with the averages on the same line as the string, or on the next line, as
appropriate. A blank line is printed before the break AVERAGE line.
The break average (or mean) is calculated by dividing the break total by
the number of values in the section and rounding down to the nearest
integer (examples: 23 / 5 = 4, -23 / 5 = -4).
(''). To suppress printing of a string, specify BAVERAGE('') using two
single apostrophes.
The break average for each numeric ON field is printed in the format
(formatting, PLUS, BLANK, or standard) you specify. The average for a
specified for that field. Averages are printed for ON(VLEN) fields, but not
for ON(NUM) fields.
number of digits used for a break total. If the break total for an ON field
overflows d digits, ICETOOL prints asterisks for the break average for that
field and terminates the operation. You can prevent overflow by specifying
an appropriate dd value for Ndd or Udd. For example, if ON(1,15,ZD)
with BAVERAGE overflows the default of 15 digits for the total, you can
specify ON(1,15,ZD,U16) to prevent overflow.
details on using Ndd or Udd.
BCOUNT('string')
Specifies a break COUNT line is to be printed after the rows of data for
each section. BCOUNT can only be specified if BREAK is specified. The
COUNT line, followed by the break count of data records in the section. If
of data. If STATLEFT is not specified, the string is printed in the first
624
DISPLAY Operator
column of data. The count is printed on the same line as the string. A
blank line is printed before the break COUNT line.
(''). To suppress printing of a string, specify BCOUNT('') using two single
apostrophes.
The count is printed in the format (PLUS, BLANK, or standard) you
specify. EDBCOUNT(formatting) can be used to apply formatting items to
the count. The default number of digits (d) for the count is 15.
EDBCOUNT(formatting)
Specifies how the break count is to be formatted for printing. Formatting
items can be specified in any order, but each item can only be specified
once. EDBCOUNT can only be specified if BCOUNT('string') is specified.
mask
E'pattern'
See ON(p,m,f,formatting) for a discussion of E'pattern'.
L'string'
F'string'
T'string'
Udd
See EDCOUNT(formatting) for a discussion of Udd.
STATLEFT
Specifies that the strings for statistics (TOTAL, MAXIMUM, MINIMUM,
AVERAGE, COUNT, BTOTAL, BMAXIMUM, BMINIMUM, BAVERAGE,
BCOUNT) are to be placed to the left of the first column of data
(overriding the default of placing the strings in the first column).
STATLEFT ensures that each statistic appears on the same line as its string
while making the statistics lines stand out from the columns of data.
UZERO
Specifies that -0 and +0 are to be treated as unsigned zero values, that is,
as the same value. With UZERO, -0 and +0 are treated as positive for ON,
MINIMUM, MAXIMUM, BREAK, BMINIMUM and BMAXIMUM
processing.
UZERO overrides the default of treating -0 and +0 as signed zero values,
that is, as different values. Without UZERO, -0 is treated as negative and
+0 is treated as positive for ON, MINIMUM, MAXIMUM, BREAK,
BMINIMUM and BMAXIMUM processing.
LISTSDB OR LISTNOSDB
Can be used to override the SDBMSG value for this LIST data set.
LISTSDB directs ICETOOL to select the system-determined optimum block
size for the LIST data set in the same way as for installation option
625
DISPLAY Operator
SDBMSG=YES. LISTNOSDB directs ICETOOL to select the block size for
the LIST data set in the same way as for installation option SDBMSG=NO.
See the discussion of the LIST(listdd) operand previously in this section for
more information on how LISTSDB or LISTNOSDB affects the LIST data
set block size.
Attention: LISTSDB has no effect for SYSOUT data sets (for example,
//RPT1 DD SYSOUT=*), because the system-determined optimum block
size is not used for spool or dummy data sets.
DISPLAY examples
Although the DISPLAY operators in the examples in this section could all be
for clarity. See OCCUR operator on page 646 for additional examples of tailoring
the report format.
Example 1:
DISPLAY FROM(SOURCE) LIST(FIELDS) ON(NUM) ON(40,12,CH) ON(20,8,PD)
Prints, in the FIELDS data set:

v A heading line containing the standard headings
v Data lines in the standard format containing:
The record number in the standard format
The characters from positions 40-51 of the SOURCE data set
The packed decimal values from positions 20-27 of the SOURCE data set in
the standard format
The FIELDS output starts on a new page and looks as follows (the first 2 records
are shown with illustrative values):
RECORD NUMBER
000000000000001
000000000000002
.
.
.
(40,12,CH)
SAN JOSE
MORGAN HILL
.
.
.
(20,8,PD)
000000000003745
000000000016502
.
.
.
The heading line appears at the top of each page.

Example 2:
DISPLAY FROM(IN) LIST(LIST1) TITLE(National Accounting Report) PAGE DATE TIME HEADER(Division) HEADER(Revenue) HEADER(Profit/Loss) ON(1,25,CH)
ON(45,10,ZD)
ON(35,10,ZD) BLANK TOTAL(Company Totals) AVERAGE(Company Averages)
Prints, in the LIST1 data set:

v A title line containing the specified title, the page number, the date and the time
v A heading line containing the specified underlined headings
v Data lines in the BLANK format containing:
The characters from positions 1-25 of the IN data set
The zoned decimal values from positions 45-54 of the IN data set
626
DISPLAY Operator
The zoned decimal values from positions 35-44 of the IN data set
v A TOTAL line containing the specified string and the total for each of the two
zoned decimal fields in the BLANK format
v An AVERAGE line containing the specified string and the average for each of
the two zoned decimal fields in the BLANK format.
The LIST1 output starts on a new page and looks as follows (the first 2 records are
shown with illustrative values):
National Accounting Report
- 1 -
Division
------------------------Research and Development
Manufacturing
.
.
.
Revenue
---------------54323456
159257631
.
.
.
Profit/Loss
----------------823325
1372610
.
.
.
612867321
5277836
76608415
659729
Company Totals
Company Averages
02/21/05
18:52:44
The title line and underlined heading line appear at the top of each page.
Example 3:
DISPLAY FROM(DATA) LIST(JUSTDATA) NOHEADER ON(17,5,PD) ON(1,2,FI)
Prints, in the JUSTDATA data set:

The packed decimal values from positions 17-21 of the DATA data set in the
standard format
The fixed-point values from positions 1-2 of the DATA data set in the
standard format
The JUSTDATA output contains no page ejects or heading lines and looks as
follows (the first 2 records are shown with illustrative values):
-0000000000273216
+0000000000993112
.
.
.
+0000000000000027
+0000000000000321
.
.
.
Example 4:
COPY FROM(INPUT) TO(TEMP) USING(TREG)
DISPLAY FROM(TEMP) LIST(REGULAR) PAGE TITLE(Report on Regular Tools) TBETWEEN(12)HEADER(NONE) ON(1,18,CH) HEADER(,Item) ON(35,5,CH) HEADER(Percent,Change) ON(28,4,FS,B1) LINES(66)
COPY FROM(INPUT) TO(TEMP) USING(TPOW)
DISPLAY FROM(TEMP) LIST(POWER) PAGE TITLE(Report on Power Tools ) TBETWEEN(12)-
627
DISPLAY Operator
HEADER(NONE) ON(1,18,CH) HEADER(,Item) ON(35,5,CH) HEADER(Percent,Change) ON(28,4,FS,B1) LINES(66)
This example shows how reports for different subsets of data can be produced.
Assume that:
v The TREGCNTL data set contains:
INCLUDE COND=(44,8,CH,EQ,CRegular)
v The TPOWCNTL data set contains:

INCLUDE COND=(44,8,CH,EQ,CPower)
The first COPY operator copies the records from the INPUT data set that contain
'Regular ' in positions 44-51 to the TEMP (temporary) data set
The first DISPLAY operator uses the first subset of records in the TEMP data set to
print, in the REGULAR data set:
v A title line containing the page number and specified title, with twelve blanks
between these report elements.
v A two-line heading containing the specified underlined strings (with no heading
for the first ON field)). Note the comma in HEADER(,'Item') to place 'Item' on
line2 of the heading.
v Data lines for the first subset of records containing:
The characters from positions 1-18
The floating sign values from positions 28-31 formatted with one decimal
place and a period as the decimal point
The second COPY operator copies the records from the INPUT data set that
contain 'Power ' in positions 44-51 to the TEMP (temporary) data set
The second DISPLAY operator uses the second subset of records in the TEMP data
set to print, in the POWER data set:
v A title line containing the page number and specified title, with twelve blanks
between these report elements.
v A two-line heading containing the specified underlined strings (with no heading
for the first ON field)). Note the comma in HEADER(,'Item') to place 'Item' on
line2 of the heading.
v Data lines for the second subset of records containing:
The floating sign values from positions 28-31 formatted with one decimal
place and a period as the decimal point
The REGULAR output starts on a new page and looks as follows (the first 2
records are shown with illustrative values):
- 1 -
Hammers
Wrenches
628
Report on Regular Tools
Item
----10325
00273
Percent
Change
-------7.3
15.8
DISPLAY Operator
The title line and underlined heading lines appear at the top of each page. The
number of lines per page is 66, overriding the default of 58.
The POWER output starts on a new page and looks as follows (the first 2 records
- 1 -
Report on Power Tools
Item
----10325
00273
Hammers
Wrenches
Percent
Change
------9.8
123.0
The title line and underlined heading lines appear at the top of each page. The
number of lines per page is 66, overriding the default of 58.
Example 5:
DISPLAY FROM(INV) LIST(RDWLIST1) TITLE(No Frills RDW Report) ON(NUM) ON(VLEN) ON(1,4,HEX) MINIMUM(Smallest) MAXIMUM(Largest)
Prints, in the RDWLIST1 data set:

v A title line containing the specified title
The record number
The record length
The record descriptor word (RDW) in hexadecimal
v A MINIMUM line containing the specified string and the minimum record
length in the standard format
v A MAXIMUM line containing the specified string and the maximum record
length in the standard format.
The RDWLIST1 output starts on a new page and looks as follows (the first 2
records are shown with illustrative values):
No Frills RDW Report
RECORD NUMBER
000000000000001
000000000000002
.
.
.
RECORD LENGTH
(1,4,HEX)
+000000000000075 004B0000
+000000000000071 00470000
. .
. .
. .
Smallest
+000000000000058
Largest
+000000000000078
The title line and heading line appear at the top of each page.
629
DISPLAY Operator
Example 6:
DISPLAY FROM(INV) LIST(RDWLIST2) DATE(DMY.) TFIRST TBETWEEN(3) TITLE(Fancy RDW Report with) TITLE(length in decimal and hex) TITLE(and max and min length) TIME(12:) HEADER(Relative Record) ON(NUM) HEADER(
RDW (length)) ON(VLEN) HEADER(RDW (Hex)) ON(1,4,HEX) BLANK MINIMUM(Smallest Record:) MAXIMUM(Largest Record:) COUNT(Number of Records: ) EDCOUNT(U04)
Prints, in the RDWLIST2 data set:

v Title lines containing the date, the specified title strings and the time
The record number
The record length
The record descriptor word (RDW) in hexadecimal
v A MINIMUM line containing the specified string and the minimum record
length in the BLANK format
v A MAXIMUM line containing the specified string and the maximum record
length in the BLANK format
v A COUNT line containing the specified string and the edited overall record
count.
RDWLIST2 output starts on a new page and looks as follows (the first 2 records
22.07.08
Fancy RDW Report with
05:37:43 pm
length in decimal and hex

and max and min length
Relative Record
--------------1
2
3
4
5
...
RDW (length)
---------------84
47
31
31
31
RDW (Hex)
--------00540000
002F0000
001F0000
001F0000
001F0000
Relative Record
--------------51
52
53
54
55
56
RDW (length)
---------------31
31
31
31
31
31
RDW (Hex)
--------001F0000
001F0000
001F0000
001F0000
001F0000
001F0000
Smallest Record:
31
Largest Record:
84
Number of Records:
630
56
DISPLAY Operator
The title lines only appear at the top of the first page as specified by the TFIRST
operand. The underlined heading line appears at the top of each page.
Example 7:
SORT FROM(PARTS) TO(TEMP) USING(SRT1)
DISPLAY FROM(TEMP) LIST(USA) TITLE(Parts Completion Report for USA) DATE HEADER(Part) HEADER(Completed) HEADER(Value ($)) ON(15,6,CH)
ON(3,4,ZD,A1)
ON(38,8,ZD,C1) TOTAL(Total:)
DISPLAY FROM(TEMP) LIST(FRANCE) TITLE(Parts Completion Report for France) DATE(DM4/) HEADER(Part) HEADER(Completed) HEADER(Value (F)) ON(15,6,CH)
ON(3,4,ZD,A3)
DISPLAY FROM(TEMP) LIST(DENMARK) TITLE(Parts Completion Report for Denmark) DATE(DMY-) HEADER(Part) HEADER(Completed) HEADER(Value (kr)) ON(15,6,CH)
ON(3,4,ZD,A2)
This example shows how reports for three different countries can be produced. The
reports differ only in the way that date and numeric values are displayed.
Assume that the SRT1CNTL data set contains:
The SORT operator sorts the PARTS data set to the TEMP data set using the SORT
statement in SRT1CNTL.
The first DISPLAY operator uses the sorted records in the TEMP data set to print,
in the USA data set:
v A title line containing the specified title and the date in the format commonly
used in the United States
v Data lines containing:
The zoned decimal values from positions 3-6 formatted with the separators
commonly used in the United States
The zoned decimal values from positions 38-45 formatted with two decimal
places and the separators and decimal point commonly used in the United
States.
zoned decimal fields formatted in the same way as the data values.
The second DISPLAY operator uses the sorted records in the TEMP data set to
print, in the FRANCE data set:
used in France
commonly used in France
631
DISPLAY Operator
places and the separators and decimal point commonly used in France.
The third DISPLAY operator uses the sorted records in the TEMP data set to print,
in the DENMARK data set:
used in Denmark
commonly used in Denmark
places and the separators and decimal point commonly used in Denmark.
The USA output starts on a new page and looks as follows (several records are
Parts Completion Report for USA
01/14/05
Part
-----000310
001184
029633
192199
821356
Completed
-------------------562
1,234
35
3,150
233
Value ($)
--------------------8,317.53
23,456.78
642.10
121,934.65
2,212.34
Total:
5,214
156,563.40
The FRANCE output starts on a new page and looks as follows (several record are
Parts Completion Report for France
14/01/2005
Part
-----000310
001184
029633
192199
821356
Completed
-------------------562
1 234
35
3 150
233
Value (F)
--------------------8 317,53
23 456,78
642,10
121 934,65
2 212,34
Total:
5 214
156 563,40
The DENMARK output starts on a new page and looks as follows (several records
632
DISPLAY Operator
Parts Completion Report for Denmark
14-01-05
Part
-----000310
001184
029633
192199
821356
Completed
-------------------562
1.234
35
3.150
233
Value (kr)
--------------------8.317,53
23.456,78
642,10
121.934,65
2.212,34
Total:
5.214
156.563,40
Example 8:
SORT FROM(DATA) TO(TEMP) USING(SRTX)
DISPLAY FROM(TEMP) LIST(WEST) DATE TITLE(Western Region Profit/Loss Report) PAGE BTITLE(Division:) BREAK(3,10,CH) HEADER(Branch Office) ON(16,13,CH) HEADER(Profit/Loss (K)) ON(41,4,PD,/K,E1) BMINIMUM(Lowest Profit/Loss in this Division:) BMAXIMUM(Highest Profit/Loss in this Division:) BAVERAGE(Average Profit/Loss for this Division:) MINIMUM(Lowest Profit/Loss for all Divisions:) MAXIMUM(Highest Profit/Loss for all Divisions:) AVERAGE(Average Profit/Loss for all Divisions:)
This example shows how a report with sections can be produced.

Assume that the SRTXCNTL data set contains:
The SORT operator sorts the DATA data set to the TEMP data set using the SORT
statement in SRTXCNTL.
The DISPLAY operator uses the sorted records in the TEMP data set to print, in the
WEST data set, sections with:
v A title line containing the date, the specified title string, and the page number
v A break title containing the specified break title string, and the break field
characters from positions 3-12
The packed decimal values from positions 41-44 divided by 1000 and
formatted with separators and signs as specified.
v Break MINIMUM, MAXIMUM, and AVERAGE lines containing the specified
strings and statistics for the packed decimal field values in this section,
formatted in the same way as the data values.
The last page of the report contains:
v A title line containing the date, the specified title string, and the page number
v Overall MINIMUM, MAXIMUM, and AVERAGE lines containing the specified
strings and statistics for the packed decimal field values in the report, formatted
in the same way as the data values.
633
DISPLAY Operator
The first section of the WEST output starts on a new page and looks as follows
(several records are shown with illustrative values):
01/14/05
Division:
Western Region Profit/Loss Report
- 1 -
Chips
Branch Office
------------Gilroy
Los Angeles
Morgan Hill
Oakland
San Francisco
San Jose
San Martin
Profit/Loss (K)
--------------3,293
(141)
213
1,067
(31)
92
1,535
Lowest Profit/Loss in this Division:

(141)
Highest Profit/Loss in this Division:
3,293
Average Profit/Loss for this Division:
861
The title line, break title line, and underlined heading line appear at the top of
each page of the section.
The second section of the WEST output starts on a new page and looks as follows
(several records are shown with illustrative values):
01/14/05
Division:
- 2 -
Ice Cream
Branch Office
------------Marin
Napa
San Francisco
San Jose
San Martin
Profit/Loss (K)
--------------673
95
(321)
2,318
21
Lowest Profit/Loss in this Division:

(321)
Highest Profit/Loss in this Division:
2,318
Average Profit/Loss for this Division:
557
The title line, break title line, and underlined heading line appear at the top of
each page of the section.
The last page of the WEST output starts on a new page and looks as follows:
634
DISPLAY Operator
01/14/05
Branch Office
-------------
Profit/Loss (K)
---------------
- 3 -
Lowest Profit/Loss for all Divisions:

(321)
Highest Profit/Loss for all Divisions:
3,293
Average Profit/Loss for all Divisions:
734
Example 9:
MODE CONTINUE
VERIFY FROM(CHECK) ON(2,3,PD) LIMIT(500)
DISPLAY FROM(CHECK) LIST(PDREPORT) BLANK LIMIT(500) HEADER(Relative Record) ON(NUM) HEADER(Numeric) ON(2,3,PD) HEADER(Hexadecimal) ON(2,3,HEX) HEADER(Associated Field) ON(21,20,CH)
This example shows how each record containing an invalid decimal value can be
identified either by its relative record number or an associated field in the record.
The MODE operator ensures that the DISPLAY operator is processed if the VERIFY
operator identifies an invalid decimal value.
The VERIFY operator checks for invalid digits (A-F) and invalid signs (0-9) in the
packed decimal values from positions 2-4 of the CHECK data set. Messages
ICE618A and ICE649A are printed in the TOOLMSG data set for each value (if
any) that contains an invalid digit or sign. If 500 invalid values are found, the
operation is terminated.
The DISPLAY operator checks for invalid digits (A-F) in the packed decimal values
from positions 2-4 of the CHECK data set. Messages ICE618A and ICE649A are
printed in the TOOLMSG data set for each value (if any) that contains an invalid
digit. If 500 invalid values are found, the operation is terminated. If a check for
invalid signs is required, the VERIFY operator must be used, because the DISPLAY
operator only checks for invalid digits. The VERIFY operator is not required if
signs need not be checked.
The DISPLAY operator also prints, in the PDREPORT data set:
The relative record number. This number can be matched against the
RECORD numbers printed in the ICE618A messages to find the records with
invalid signs.
The numeric representation of the packed decimal value in positions 2-4.
Asterisks are shown for values with invalid digits, making them easy to
identify. Asterisks are not shown for values with invalid signs; these must be
identified by matching the relative record number against the RECORD
number in ICE618A.
The hexadecimal representation of the packed decimal value in positions 2-4
(also shown in ICE649A). This makes it easy to find the specific hexadecimal
digits or signs that are invalid.
635
DISPLAY Operator
The characters in positions 21-40. An associated field such as this can be used
to make identification of the records with invalid values easier.
The ICE618A and ICE649A messages in TOOLMSG for the VERIFY operator are:
ICE618A
ICE649A
ICE618A
ICE649A
ICE618A
ICE649A
0 INVALID (2,3,PD)
0
HEX VALUE: 53A54C
0 INVALID (2,3,PD)
0
HEX VALUE: 621540
0 INVALID (2,3,PD)
0
HEX VALUE: 400F3C
VALUE - RECORD:
000000000000003
VALUE - RECORD:
000000000000012
VALUE - RECORD:
000000000000019
The ICE618A and ICE649A messages in TOOLMSG for the DISPLAY operator are:
ICE618A
ICE649A
ICE618A
ICE649A
0 INVALID (2,3,PD)
0
HEX VALUE: 53A54C
0 INVALID (2,3,PD)
0
HEX VALUE: 400F3C
VALUE - RECORD:
000000000000003
VALUE - RECORD:
000000000000019
The PDREPORT output looks as follows:

Relative Record
--------------1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Numeric
------18600
-93
******
86399
24215
8351
19003
-31285
88316
1860
-29285
62154
-328
-11010
1363
92132
-48500
-55
******
33218
96031
Hexadecimal
----------18600C
00093B
53A54C
86399C
24215F
08351C
19003C
31285D
88316C
01860C
29285D
621540
00328D
11010D
01363F
92132C
48500D
00055D
400F3C
33218C
96031C
Associated Field
-------------------Wagar
Gellai
Giulianelli
Mehta
Johnson
Packer
Childers
Burg
Monkman
Vezinaw
Mead
Wu
Madrid
Warren
Burt
Mao
Shen
Yamamoto-Smith
Yaeger
Leung
Kaspar
PDREPORT can be used in conjunction with the ICE618A and ICE649A messages
to identify that:
v Record 3 has an invalid digit of A and an associated field of Giulianelli
v Record 12 has an invalid sign of 0 and an associated field of Wu
v Record 19 has an invalid digit of F and an associated field of Yaeger.
Example 10:
COPY FROM(IN) USING(OUTF)
DISPLAY FROM(TEMP) LIST(EMPCT) BLANK TITLE(Employees by Function) DATE HEADER(Function) HEADER(Employees) ON(1,25,CH)
ON(30,4,ZD)
This example shows how the OUTFIL table lookup feature can be used to
substitute meaningful phrases for cryptic values in ICETOOL reports. Assume that:
v The OUTFCNTL data set contains:
636
DISPLAY Operator
OUTFIL FNAMES=TEMP,
OUTREC=(1:9,2,CHANGE=(25,
CMN,CManufacturing,
CRD,CResearch and Development,
CFN,CFinance,
CMR,CMarketing,
CIS,CInformation Systems),
30:4,4)
The COPY operator uses the OUTFIL statement in OUTFCNTL to reformat the IN
data set records to the TEMP (temporary) data set. Two fields are extracted for use
by the DISPLAY operator:
v The 2-character department code in positions 9-10 is changed to a 25-character
name in positions 1-25 using the table lookup feature.
v The zoned decimal value in positions 4-7 is moved to positions 30-33.
The DISPLAY operator uses the reformatted fields in the TEMP data set to print, in
the EMPCT data set:
v A title line containing the specified title and the date
The names from positions 1-25 that were substituted for the department codes
The zoned decimal values from positions 30-33.
The EMPCT output starts on a new page and looks as follows:
Employees by Function
Function
------------------------Manufacturing
Marketing
Research and Development
Information Systems
Finance
02/14/05
Employees
--------486
21
55
123
33
Example 11:
DISPLAY FROM(ACCTS) LIST(PLAIN) TITLE(Accounts Report for First Quarter) DATE(MD4/) BLANK HEADER(Amount) ON(12,6,ZD) HEADER(Id) ON(NUM) HEADER(Acct#) ON(31,3,PD) HEADER(Date) ON(1,4,ZD) TOTAL(Total for Q1) AVERAGE(Average for Q1)
DISPLAY FROM(ACCTS) LIST(FANCY) TITLE(Accounts Report for First Quarter) DATE(MD4/) BLANK HEADER(Amount) ON(12,6,ZD,C1,N08) HEADER(Id) ON(NUM,N02) HEADER(Acct#) ON(31,3,PD,NOST,LZ) HEADER(Date) ON(1,4,ZD,E99/99,NOST) INDENT(2) BETWEEN(5) STATLEFT TOTAL(Total for Q1) AVERAGE(Average for Q1)
637
DISPLAY Operator
This example shows some options you can use to improve the appearance of a
DISPLAY report. The first DISPLAY operator produces a "plain" report, and the
second DISPLAY operator uses the options shown in bold to produce a "fancy"
report.
The PLAIN output starts on a new page and looks as follows:
Accounts Report for First Quarter
Amount
--------------93271
137622
83147
183261
76389
920013
Id
--------------1
2
3
4
5
6
05/04/2001
Acct#
------------------15932
187
15932
2158
187
15932
Date
-------------------106
128
212
217
305
319
Total for Q1
1493703
50328
1287
Average for Q1
248950
8388
214
The FANCY output starts on a new page and looks as follows:

Accounts Report for First Quarter
Amount
-------932.71
1,376.22
831.47
1,832.61
763.89
9,200.13
Total for Q1
Average for Q1
05/04/2001
Id
--1
2
3
4
5
6
Acct#
-----15932
00187
15932
02158
00187
15932
Date
----01/06
01/28
02/12
02/17
03/05
03/19
14,937.03
2,489.50
Here is an explanation of the extra options used for the "fancy" report:
v First ON field: In the PLAIN report, BLANK causes ICETOOL to print the 6-byte
ZD values as unedited digits with leading zeros suppressed. But for this
example, we know the digits really represent dollars and cents. So in the FANCY
report, we use the C1 formatting item (one of thirty-three available masks) to
print the values with a comma (,) as the thousands separator and a period (.) as
the decimal point.
In the PLAIN report, TOTAL causes ICETOOL to allow 15 digits for the values
because it does not know how many digits are needed. But for this example, we
know the total amount will not exceed 8 digits. So in the FANCY report, we use
the N08 formatting item to set the number of digits to 8. This decreases the
column width for the field.
v Second ON field: In the PLAIN report, NUM causes ICETOOL to allow 15 digits
for the record number because it does not know how many digits are needed.
But for this example, we know the number of records will not exceed 99. So in
the FANCY report, we use the N02 formatting item to set the number of digits
to 2. This decreases the column width for the record number.
v Third ON field: In the PLAIN report, TOTAL and AVERAGE cause ICETOOL to
print the total and average for this 3-byte PD field. But for this example, we
638
DISPLAY Operator
know we do not want statistics for the field because it is an account number. So
in the FANCY report, we use the NOST formatting item to suppress the
statistics for this field.
In the PLAIN report, the default mask of A0 causes ICETOOL to suppress
leading zeros for this 3-byte PD field. But for this example, we know that we
want to show leading zeros for the field because it is an account number. So in
the FANCY report, we use the LZ formatting item to print leading zeros for this
field.
v Fourth ON field: In the PLAIN report, BLANK causes ICETOOL to print the
4-byte ZD values as unedited digits with leading zeros suppressed. But for this
example, we know the digits represent a date (month and day). So in the
FANCY report, we use the E'99/99' formatting item to print the values with
leading zeros and a slash (/) between the month and day.
In the PLAIN report, TOTAL and AVERAGE cause ICETOOL to print the total
and average for this 4-byte ZD field. But for this example, we know we do not
want the total or average for this field because it is a date. So in the FANCY
report, we use the NOST formatting item to suppress the statistics for this field.
Note: In some applications, we might want the minimum and maximum for a
date displayed with E'pattern', so we would not specify NOST for the date field.
v INDENT: In the PLAIN report, ICETOOL starts the report in column 2 (after the
control character), by default. But for this example, we want to indent the report
a bit. So in the FANCY report, we use the INDENT(2) operand to indent the
report by 2 blanks so it starts in column 4.
v BETWEEN: In the PLAIN report, ICETOOL uses 3 blanks between the columns
of data, by default. But for this example, we want more space between the
columns. So in the FANCY report, we use the BETWEEN(5) operand to insert 5
blanks between the columns.
v STATLEFT: In the PLAIN report, ICETOOL prints the strings for TOTAL and
AVERAGE under the first column of data, by default, and uses two lines for
each statistic to avoid having the string overlay the value. But for this example,
we would like to have the TOTAL and AVERAGE strings stand out in the report
and also have each string on the same line as its value. So in the FANCY report,
we use the STATLEFT operand to print the TOTAL and AVERAGE strings to the
left of the first column of data.
Example 12:
SORT FROM(RAWSMF) TO(SMF14) USING(SMFI)
DISPLAY FROM(SMF14) LIST(SMF14RPT) TITLE(SMF Type-14 Records) DATE(4MD/) HEADER(Date) ON(11,4,DT1,E9999/99/99) HEADER(Time) ON(7,4,TM1,E99:99:99) HEADER(Sys) ON(15,4,CH) HEADER(Jobname) ON(19,8,CH) HEADER(Datasetname) ON(69,44,CH)
This example shows how SMF date and time values can be displayed in a
meaningful way in a report on SMF type-14 records.
The SORT operator selects the type-14 records from the RAWSMF data set and
sorts them by date and time to the SMF14 data set. It uses the following control
statements in SMFICNTL:
INCLUDE COND=(6,1,BI,EQ,14)
SORT FIELDS=(11,4,PD,A,7,4,BI,A)
639
DISPLAY Operator
The DISPLAY operator uses the selected type-14 records in SMF14 to print, in the
SMF14RPT data set:
v A title line containing the specified title and the date
The SMF date values in positions 11-14 displayed as C'yyyy/mm/dd'
The SMF time values in positions 7-10 displayed as C'hh:mm:ss'
The character values in positions 15-18
The SMF14RPT output starts on a new page and looks as follows:
SMF Type-14 Records
Date
---------2001/04/20
2001/04/20
2001/04/21
2001/04/21
2001/04/24
2001/04/24
Time
-------06:03:15
10:03:22
14:05:37
22:11:00
00:00:08
Sys
---ID03
ID02
ID03
ID03
ID03
Jobname
-------JOB00003
JOB00002
JOB00004
JOB00005
JOB00006
Datasetname
----------- ...
SYS1.QRS
SYS1.XYZ
SYS1.MNO
SYS1.MNO
SYS1.MNO
Note: When you use SMF date formats (DTn) or SMF time formats (TMn), the
SMF values are treated as numeric. This allows you to use numeric formatting
items such as masks and patterns to edit the SMF values. By default, DTn and
TMn headings, like other numeric headings, appear right-aligned as shown in the
SMF14RPT output example shown previously in this section. If you want to
center-align or left-align headings for numeric values, you can add an appropriate
number of blanks at the end of HEADER('string1').
For example, if you wanted to left-align the SMF date heading, you could use six
blanks at the end of the header string like so:
HEADER(Date
to get the following heading:

Date
----------
If you wanted to center-align the SMF date heading, you could use three blanks at
the end of the header string like so:
HEADER(Date
to get the following heading:

Date
----------
Example 13:
SORT FROM(SMFIN) TO(SMF71) USING(TY71)
DISPLAY FROM(SMF71) LIST(SMF71RPT) TITLE(Low impact central storage frames) BREAK(15,4,CH,LSystem: ) HEADER(Date) ON(11,4,DT1,E9999-99-99) HEADER(Time) ON(7,4,TM1,E99:99:99) -
640
DISPLAY Operator
HEADER(Min Frames) ON(925,8,FL,U10) HEADER(Max Frames) ON(933,8,FL,U10) HEADER(Avg Frames) ON(941,8,FL,U10) BLANK PAGE
This example shows how floating point values can be displayed as integers in a
report on SMF type-71 records with a section for each system id.
The SORT operator selects SMF type-71 records that are at least 19 bytes long and
sorts them by system id, date and time to the SMF71 data set. It uses the following
control statements in TY71CNTL:
OMIT COND=(6,1,BI,NE,+71,OR,1,2,BI,LE,+18)
SORT FIELDS=(15,4,CH,A,11,4,PD,A,7,4,BI,A)
The DISPLAY operator uses the selected type-71 records in SMF71 to print, in the
SMF71RPT data set:
v A title line containing the specified title and the page number.
v A break title containing the specified leading string and the SMF71SID system id
character values in positions 15-18.
v A heading line containing the specified underlined headings.
The SMF71DTE date value. This SMF date value in positions 11-14 is
displayed as a C'yyyy-mm-dd' value.
The SMF71TME time value. This SMF time value in positions 7-10 is
displayed as a C'hh:mm:ss' value.
The SMF71CLM minimum number of low-impact central storage frames
value. This floating-point value in positions 925-932 is displayed as a 10 digit
integer value. (The U10 formatting item reduces the number of digits for the
integer representation of the floating-point value from 20 to 10, decreasing the
column width for the field.)
The SMF71CLX maximum number of low-impact central storage frames
value. This floating-point value in positions 933-940 is displayed as a 10 digit
integer value.
The SMF71CLA average number of low-impact central storage frames value.
This floating-point value in positions 941-948 is displayed as a 10 digit integer
value.
Each system id value starts on a new page and looks as follows (several sections
and records are shown with illustrative values):
Low impact central storage frames
- 1 -
System: SYSA
Date
---------2005-08-01
2005-08-01
2005-08-01
Time
-------11:45:00
12:00:00
12:15:00
Min Frames
----------934215
971599
970192
Low impact central storage frames
Max Frames
----------1001596
1004939
982565
Avg Frames
----------963434
984437
973768
- 2 -
System: SYSB
Date
----------
Time
--------
Min Frames
-----------
Max Frames
-----------
Avg Frames
-----------
641
DISPLAY Operator
2005-08-01
2005-08-01
2005-08-01
11:45:00
12:00:00
12:15:00
947220
980120
980387
985444
1018982
1051920
966581
986360
1011873
MERGE operator
,
,
MERGE
FROM
( indd
USING(xxxx)

,
TO
outdd
)

VSAMTYPE(x)
LOCALE(name)
LOCALE(CURRENT)
LOCALE(NONE)
SERIAL
Merges up to 50 input data sets to an output data set. The records in each input
data set to be merged must already be in sorted order as specified by the control
fields in a supplied DFSORT MERGE statement.
You must specify at least one FROM operand and a USING(xxxx) operand. Each
FROM operand can be used to specify one or more ddnames. You can specify up
to 10 FROM operands. The maximum number of ddnames in all of the FROM
operands must not exceed 50.
DFSORT is called to merge the indd data sets to the outdd data sets using the
DFSORT control statements in xxxxCNTL. You must supply a DFSORT MERGE
statement in the xxxxCNTL data set to indicate the control fields for the merge.
You can use additional DFSORT statements in the xxxxCNTL data set to merge a
subset of the input records (INCLUDE or OMIT statement; OUTFIL INCLUDE,
OMIT, SAVE, STARTREC, ENDREC, SAMPLE, SPLIT, SPLITBY and SPLIT1R
operands), reformat records for output (INREC, OUTREC, and OUTFIL statements;
user exit routines), and so on.
If EQUALS is in effect, records that collate identically are output in the order of
their ddnames in the FROM operands.
The active locale's collating rules affect MERGE processing as explained in
MERGE control statement on page 162. If an INCLUDE or OMIT statement or an
OUTFIL INCLUDE or OMIT operand is specified in the xxxxCNTL data set, the
active locale's collating rules affect INCLUDE and OMIT processing as explained in
the "Cultural Environment Considerations" discussion in INCLUDE control
Note: For a merge application, records deleted during an E35 exit routine are not
sequence checked. If you use an E35 exit routine without an output data set,
sequence checking is not performed at the time the records are passed to the E35
user exit; therefore, you must ensure that input records are in correct sequence.
642
Operand Descriptions
The operands described in this section can be specified in any order:
FROM(indd,...)
Specifies the ddnames of the input data sets to be read by DFSORT for this
operation. Up to 10 FROM operands can be used to specify up to 50 input
ddnames. An indd DD statement must be present for each indd name
specified. Each indd input data set must conform to the rules for DFSORT's
SORTINnn data sets.
USING(xxxx)
An xxxxCNTL DD statement must be present, and the control statements in it
must conform to the rules for DFSORT's SORTCNTL data set.
The xxxxCNTL data set must contain a MERGE statement. If TO is not
specified, the xxxxCNTL data set must also contain either one or more OUTFIL
statements or a MODS statement for an E35 routine that disposes of all
records. Other statements are optional.
TO(outdd,...)
specified, DFSORT is called once to merge the indd data sets to the outdd data
set using SORTOUT processing; the outdd data set must conform to the rules
for DFSORT's SORTOUT data set. If multiple outdd data sets are specified and
SERIAL is not specified, DFSORT is called once to merge the indd data sets to
the outdd data sets using OUTFIL processing; the outdd data sets must
conform to the rules for DFSORT's OUTFIL data sets.
A ddname specified in a FROM operand must not also be specified in the TO
operand.
VSAMTYPE(x)
on page 575.
LOCALE(name)
on page 575.
LOCALE(CURRENT)
on page 575.
LOCALE(NONE)
on page 575.
SERIAL
643
Operand Descriptions
DFSORT is called to merge the indd data set to the first outdd data set using
the DFSORT control statements in the xxxxCNTL data set. If the merge
operation is successful, DFSORT is called as many times as necessary to copy
the first outdd data set to the second and subsequent outdd data sets.
Therefore, for maximum efficiency, use a disk data set as the first in a list of
outdd data sets on both disk and tape. If more than one outdd data set is
specified, DFSORT must be able to read the first outdd data set after it is
written in order to copy it to the other outdd data sets. Do not use a SYSOUT
or DUMMY data set as the first in a list of outdd data sets because:
v If the first data set is SYSOUT, DFSORT abends when it tries to copy the
v If the first data set is DUMMY, DFSORT copies the empty DUMMY data set
to the other outdd data sets (that is, all of the resulting outdd data sets are
empty).
MERGE examples
Although the MERGE operators in the examples in this section could all be
for clarity.
Example 1
//TOOLIN DD *
MERGE FROM(IN01,IN02,IN03,IN04,IN05) TO(OUTPUT) USING(MERG)
//MERGCNTL DD *
OPTION EQUALS
MERGE FIELDS=(21,4,CH,A)
/*
This example merges 5 input files to an output file. EQUALS is used to ensure that
records that collate identically are output in the order specified in the FROM
operand. For example, if IN01, IN03 and IN05 all have records with a key or
'AAAA' in positions 21-24, the output will contain the 'AAAA' record from IN01,
the 'AAAA' record from IN03 and the 'AAAA' record from IN05, in that order.
Example 2
//TOOLIN DD *
MERGE FROM(INPUT1,INPUT2,INPUT3,INPUT4) FROM(INPUT5,INPUT6,INPUT7) VSAMTYPE(F) USING(MRG1)
//MRG1CNTL DD *
MERGE FIELDS=(52,8,UFF,D)
OUTFIL FNAMES=OUT1,INCLUDE=(15,3,SS,EQ,CD21,D33)
OUTFIL FNAMES=OUT2,SAVE
/*
This example merges 7 input files to 2 output files. It uses two OUTFIL statements
to create the two output files; each output file will have a different subset of the
merged records. VSAMTYPE(F) tells DFSORT the record type is F (only needed for
VSAM input files).
MODE operator
MODE
644
STOP
CONTINUE
SCAN
MODE Operator
Specifies one of three modes to control error checking and actions after error
detection. A MODE operator effects the processing (that is, error checking of
ICETOOL statements and calling DFSORT) of the operators that follow it, up to the
next MODE operator (if any). MODE operators allow you to do the following for
groups of operators or all operators:
1. Stop or continue processing operators after a return code of 8, 12 or 16. A
return code of 12 or 16 can be set as the result of a statement or run-time error
detected by ICETOOL or DFSORT. A return code of 8 can be set as the result of
a COUNT operator with RC8.
2. Check for errors in ICETOOL statements, but do not call DFSORT.
STOP
Stops subsequent operations if a return code of 8, 12 or 16 is set. If an error is
detected for an operator, SCAN mode is automatically set in effect; DFSORT is
not called for subsequent operators, although checking ICETOOL statements
for errors continues.
STOP mode can be used to group dependent operators (that is, if an operation
fails, do not process the remaining operators).
STOP MODE is set in effect automatically at the start of the ICETOOL run.
CONTINUE
Continues with subsequent operations regardless of whether or not a return
code of 8, 12 or 16 is set. If an operator results in an error, processing continues
for subsequent operators.
CONTINUE mode can be used to group independent operators (that is,
process each operator regardless of the success or failure of the others).
SCAN
ICETOOL statements are checked for errors, but DFSORT is not called.
SCAN mode can be used to test ICETOOL statements for errors.
Note: SCAN mode is set automatically if an error is detected while in STOP
mode.
MODE example
MODE SCAN
RANGE ...
UNIQUE ...
MODE STOP
VERIFY ...
DISPLAY ...
MODE CONTINUE
COPY ...
SORT ...
STATS ...
SCAN mode: RANGE and UNIQUE are checked for statement errors, but DFSORT
is not called.
STOP mode: DISPLAY is dependent on VERIFY. If the return code for VERIFY is
12 or 16, SCAN mode is entered; DISPLAY is checked for statement errors, but
DFSORT is not called.
645
MODE Operator
CONTINUE mode: COPY, SORT, and STATS are independent of each other. SORT
is processed even if the return code for COPY is 12 or 16. STATS is processed even
if the return code for COPY or SORT is 12 or 16.
Note that the return codes for one group of operators does not affect the other
groups of operators.
OCCUR operator
FROM(indd)
OCCUR
OCCURS
ON(p,m,f)
ON(p,m,HEX)
ON(VLEN)
ON(VLEN,formatting)
ON(VALCNT)
ON(VALCNT,formatting)

TITLE('string1')
LIST(listdd)

TLEFT
TFIRST

PAGE
DATE
DATE(abcd)
DATENS(abc)
YDDD(abc)
YDDDNS(ab)
TIME
TIME(abc)
TIMENS(ab)
BLANK
PLUS
NOCC

HEADER('string1')
HEADER(NONE)
NOHEADER
LINES(n)
INDENT(n)

BETWEEN(n)
ALLDUPS
NODUPS
HIGHER(x)
LOWER(y)
EQUAL(v)
VSAMTYPE(x)
WIDTH(n)
LONGLINE

UZERO
646
TBETWEEN(n)
LISTSDB
LISTNOSDB
OCCUR Operator
Prints each unique value for specified numeric fields (including SMF, TOD, and
ETOD date and time) or character fields, and how many times it occurs, in a
separate list data set. Simple or tailored reports can be produced. The values
printed can be limited to those for which the value count meets specified criteria.
From 1 to 10 fields can be specified, but the resulting list data set line length must
not exceed the limit specified by the WIDTH operand or 8192 bytes if LONGLINE
is specified and WIDTH is not specified, or 2048 bytes if LONGLINE and WIDTH
are not specified. At least one ON(VLEN) or ON(p,m,f) field must be specified; all
such ON fields specified are used to determine whether a record contains a unique
value. A single list data set record is printed for each unique value. If
ON(VALCNT) is specified, the "value count" (that is, the number of times the ON
values occur) is printed in the list data set record along with the other ON values.
Specifying formatting items or the PLUS or BLANK operand, which can
"compress" the columns of output data, can enable you to include more fields in
your report, up to a maximum of 10, if your line length is limited by the
character width your printer or display supports.
ALLDUPS, NODUPS, HIGHER(x), LOWER(y) or EQUAL(v) can be specified to
limit the ON values printed to those for which the value count meets the specified
criteria (for example, ALLDUPS for duplicate values only). The default criteria is
HIGHER(0) resulting in the ON values being printed for each unique value.
DFSORT is called to sort the indd data set to ICETOOL's E35 user exit. ICETOOL
uses its E35 exit to print appropriate titles, headings and data in the list data set.
You must not supply your own DFSORT MODS, INREC, OUTREC, SUM, or
RECORD statement, because they override the DFSORT statements passed by
ICETOOL for this operator.
inappropriate for an OCCUR operator, you can take one of the following actions:
in the DFSPARM data set.

2. Use SORTWKdd DD statements to override the use of dynamic allocation.
Refer to SORTWKdd DD statement on page 71 for details.
Attention: Either of these actions affects the work data sets used for a UNIQUE
operator, or for a SELECT or SPLICE operator for which USING(xxxx) is not
specified.
Simple report
You can produce a simple report by specifying just the required operands. For
example, if you specify FROM and LIST operands, and ON operands for 10-byte
character and 7-byte zoned decimal fields and the value count, the output in the
list data set can be represented as follows:
647
OCCUR Operator
(p,m,f)
characters
.
.
.
(p,m,f)
sddddddddddddddd
.
.
.
VALUE COUNT
ddddddddddddddd
.
.
.
A control character occupies the first byte of each list data set record. Left-justified
standard headings are printed at the top of each page to indicate the contents of
each column, followed by a line for each record showing the characters and
numbers in the fields of that record, and the count of occurrences (value count) of
the specified values.
the OCCUR statement. All fields are left-justified. For numeric fields, leading zeros
are printed, a - is used for the minus sign, and a + is used for the plus sign. For
the value count, leading zeros are printed.
By default, the first column of data starts immediately after the control character,
and three blanks appear between columns. The NOCC operand can be used to
suppress the control character. The INDENT operand can be used to change the
number of blanks before the first column of data. The BETWEEN operand can be
used to change the number of blanks between columns.
The standard column widths are as follows:
v Character data: the length of the character field or 20 bytes if the field length is
less than 21 bytes
v Numeric data: 16 bytes, or 32 bytes if the numeric field is BI or FI with a length
greater than 4, PD with a length greater than 8, or ZD, CSF, FS, UFF or SFF with
a length greater than 15.
v Value count: 15 bytes
HEADER operands can be used to change or suppress the headings. PLUS or
BLANK operands can be used to change the format of numeric fields. PLUS,
BLANK and HEADER operands can be used to change the width of the columns
for numeric and character fields and the justification of headings and fields.
The NOHEADER operand can be used to create list data sets containing only data
records. Data sets created in this way can be processed further by other operators
(for example, STATS or UNIQUE) using CH format for character values or FS
format for numeric values (including the value count).
Tailored report
You can tailor the output in the list data set using various operands that control
title strings, date, time, page number, headings, lines per page and field formats.
The optional operands can be used in many different combinations to produce a
wide variety of report formats. For example, if you specify FROM, LIST, BLANK,
TITLE, PAGE, DATE, TIME, and HEADER operands, and ON operands for 10-byte
character and 7-byte zoned decimal fields and the value count, the output in the
list data set looks as follows:
648
OCCUR Operator
title
- p -
header
---------characters
.
.
.
header
-------sd
.
.
.
mm/dd/yy
hh:mm:ss
header
-------------d
.
.
.
By default, a control character occupies the first byte of each list data set record.
The NOCC operand can be used to suppress the control characters. The title lines
(up to 3) are printed at the top of each page of the list data set. The first title line
contains the elements you specify (title strings,page number, date and time) in the
order in which you specify them. The second and third title lines contain the title
strings you specify. By default, eight blanks appear between title elements, the title
strings are centered with respect to each other, and the title lines appear on every
page. The TBETWEEN(n) operand can be used to change the number of blanks
between title elements. The TLEFT operand can be used to left-justify the title
strings with respect to each other. The TFIRST operand can be used to only print
the title lines on the first page. A blank line is printed after each title line.
Your specified headings (underlined) are printed after the title line on each page to
indicate the contents of each column, followed by a line for each record showing
the characters and numbers in the fields of that record. Your specified headings
can be one, two or three lines. Headings for character fields are left-justified and
headings for numeric fields are right-justified.
the OCCUR statement. Character fields are left-justified and numeric fields are
right justified. For numeric fields, leading zeros are suppressed, a - is used for the
minus sign, and a blank is used for the plus sign (you can specify PLUS rather
than BLANK if you want a + to be used for the plus sign). For the value count,
leading zeros are suppressed.
Formatting items can be used to change the appearance of individual numeric
fields in the report with respect to separators, number of digits, decimal point,
decimal places, signs, leading zeros, leading strings, floating strings, and trailing
strings. Formatting items can also be used to insert leading or trailing strings for
character fields.
The column widths are dynamically adjusted according to the length of the
headings and the maximum number of bytes needed for the character or numeric
data.
FROM(indd)
See the discussion of this operand on the DISPLAY statement in DISPLAY
ON(p,m,f)
used for this operation. '(p,m,f)' is used for the standard column heading (see
649
OCCUR Operator
By default, three blanks appear between columns. You can change the space
between columns with BETWEEN(n).
p specifies the first byte of the field relative to the beginning of the input
record. p is 1 for the first data byte of a fixed-length record and 5 for the first
data byte of a variable-length record as illustrated in the following (RRRR
represents the 4-byte record descriptor word):
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3 4
| p=
1
2
3
4
5
6
7
8
f specifies the format of the field as follows:
650
Format Code
Length
Description
BI
1 to 8 bytes
Unsigned binary
FI
1 to 8 bytes
Signed fixed-point
PD
1 to 16 bytes
ZD
1 to 31 bytes
CH
1 to 4000 bytes
Character
CSF or FS

UFF
SFF
DT1
4 bytes

Z'yyyymmdd'
DT2
4 bytes

Z'yyyymm'
DT3
4 bytes

Z'yyyyddd'
DC1
8 bytes

Z'yyyymmdd'
DC2
8 bytes

Z'yyyymm'
DC3
8 bytes

Z'yyyyddd'
DE1
8 bytes

Z'yyyymmdd'
DE2
8 bytes

Z'yyyymm'
DE3
8 bytes

Z'yyyyddd'
TM1
4 bytes

Z'hhmmss'
TM2
4 bytes

Z'hhmm'
TM3
4 bytes
TM4
4 bytes

Z'hhmmssxx'
OCCUR Operator
Format Code
Length
Description
TC1
8 bytes

Z'hhmmss'
TC2
8 bytes

Z'hhmm'
TC3
8 bytes

Z'hh'
TC4
8 bytes

Z'hhmmssxx'
TE1
8 bytes

Z'hhmmss'
TE2
8 bytes

Z'hhmm'
TE3
8 bytes

Z'hh'
TE4
8 bytes

Z'hhmmssxx'

descriptions.

v A maximum of 31 digits is allowed. If a value with more than 31 digits is
found, ICETOOL issues an error message and terminates the operation.
v If a decimal value contains an invalid digit (A-F), ICETOOL identifies the
bad value in a message and terminates the operation.
v F, E, C, A, 8, 6, 4, 2, and 0 are treated as equivalent positive signs. Thus the
zoned decimal values F2F3C1, F2F3F1 and 020301 are counted as only one
unique value.
v D, B, 9, 7, 5, 3, and 1 are treated as equivalent negative signs. Thus the
zoned decimal values F2F3B0, F2F3D0, and 020310 are counted as only one
unique value.
The fields of records that do not meet the specified criteria are not checked for
invalid digits (PD and ZD) or excessive digits (CSF, FS, UFF, and SFF).
v An invalid SMF date can result in a data exception (0C7 ABEND) or an
incorrect ZD date.
651
OCCUR Operator
Specifies the position, length and format of a numeric or character field to be
used for this operation and how the data for this field is to be formatted for
printing. The BLANK operand is automatically in effect.
See ON(p,m,f) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
Ndd
Udd
item can only be specified once. Any formatting item can be specified for a
numeric field, but only L'string' and T'string' can be specified for a character
field.
mask
See the discussion of mask under ON(p,m,f,formatting) in DISPLAY
E'pattern'
See the discussion of E'pattern' under ON(p,m,f,formatting) in DISPLAY
L'string'
See the discussion of L'string' under ON(p,m,f,formatting) in DISPLAY
F'string'
See the discussion of F'string' under ON(p,m,f,formatting) in DISPLAY
T'string'
See the discussion of T'string' under ON(p,m,f,formatting) in DISPLAY
LZ See the discussion of LZ under ON(p,m,f,formatting) in DISPLAY
Ndd or Udd
specifies the number of digits to be used for the numeric field. Ndd or
Udd can be used to change the column width for numeric fields. dd
specifies the number of digits and must be a two-digit number between 01
and 31.
The default number of digits (d) for a numeric field is the maximum
number of digits for that field. For example, d is 5 for ON(1,5,ZD). If you
know that your numeric field requires less than d digits, you can use a
lower number of digits (dd) instead by specifying Udd, thus reducing the
652
OCCUR Operator
column width if it is determined by d. For example, ON(1,5,ZD,U03)
reduces d from 5 to 3. If you want your numeric field to be displayed with
more than d digits, you can use a higher number of digits (dd) instead by
specifying Ndd or Udd, thus increasing the column width if it is
determined by d. For example, ON(1,5,ZD,U10) increases d from 5 to 10.
Either Ndd or Udd can be used to set d greater than the maximum for a
numeric field, but only Udd can be used to set d less than the maximum
for a numeric field.
For Udd:
dd is used for d. For example:
v If ON(1,16,FS) is specified, 16 digits (default for 16,FS) are used.
If you use Udd and a numeric value overflows dd digits, ICETOOL
terminates the operation. You can prevent the overflow by specifying an
appropriately higher dd value for Udd. For example, if ON(1,12,ZD,U09)
results in overflow, you can use ON(1,12,ZD,U10) instead.
If E'pattern' is specified, Udd is ignored, because the number of digits is
determined from the pattern.
For Ndd:
If dd is greater than or equal to d, dd is used. If dd is less than d, d is
used. For example:
v If ON(1,5,ZD,N03) is specified, 5 digits (from 5,ZD) are used.
If E'pattern' is specified, Ndd is ignored, because d is determined from the
pattern.
ON(p,m,HEX)
Specifies the position and length of a character field to be used for this
operation and printed in hexadecimal format (00-FF for each byte). '(p,m,HEX)'
is used for the standard column heading (see HEADER('string1'),
HEADER(NONE), and NOHEADER for alternative heading options).
See ON(p,m,f) for a discussion of p.
position 32752 or beyond the end of a record. A field can be 1 to 2000 bytes.
ON(VLEN)
Equivalent to specifying ON(1,2,BI); a two-byte binary field starting at position
1. For variable-length records, ON(VLEN) represents the record-length for each
record. 'RECORD LENGTH' is used for the standard column heading. See
alternative heading options.
653
OCCUR Operator
ON(VLEN,formatting)
Equivalent to specifying ON(1,2,BI,formatting); a two-byte binary field starting
at position 1, and how the data for this field is to be formatted for printing.
The BLANK operand is automatically in effect.
See ON(VLEN) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
Ndd
Udd
mask
E'pattern'
See the discussion of E'pattern' under ON(p,m,f,formatting) in DISPLAY
L'string'
F'string'
T'string'
Ndd or Udd
See the discussion of Ndd or Udd under ON(p,m,f,formatting) in OCCUR
ON(VALCNT)
Specifies that the number of occurrences for each unique value is to be printed.
' VALUE COUNT' is used for the standard column heading (see
HEADER('string1','string2','string3'), HEADER(NONE) and NOHEADER for
654
OCCUR Operator
ON(VALCNT,formatting)
Specifies that the number of occurrences for each unique value is to be printed,
and how the value count is to be formatted for printing. The BLANK operand
is automatically in effect.
See ON(VALCNT) for further details.
formatting
,

mask
E'pattern'
L'string'
F'string'
T'string'
LZ
Ndd
Udd
specifies formatting items that indicate how the value count is to be formatted
for printing. Formatting items can be specified in any order, but each item can
only be specified once.
mask
E'pattern'
specifies an edit pattern to be applied to the value count. The pattern (1 to
24 characters) must be enclosed in single apostrophes. Each 9 in the pattern
(up to 15) is replaced by a corresponding digit from the numeric value.
Characters other than 9 in the pattern appear as specified. To include a
When E'pattern' is specified for the value count:
v If the number of significant digits in a value count is less than the
1234 is shown as 001234 with ON(VALCNT,E'999999').
v If the number of significant digits in a value count is greater than the
example, 1234567 is shown as *4567* with ON(VALCNT,E'*9999*').
L'string'
F'string'
T'string'
655
OCCUR Operator
Ndd or Udd
Specifies the number of digits to be used for the value count when
determining the column width. dd specifies the number of digits and must
be a two-digit number between 01 and 15.
The default number of digits (d) for the value count is 15. If you know that
your value counts require less than 15 digits, you can use a lower number
of digits (dd) instead by specifying Ndd or Udd, thus reducing the column
width if it is determined by d. For example, if ON(VALCNT,N06) or
ON(VALCNT,U06) is specified, 6 digits (from N06 or U06) is used instead
of 15 (default for value count).
If you use Ndd or Udd and a value count overflows the number of digits
used, ICETOOL terminates the operation. You can prevent the overflow by
specifying an appropriately higher dd value for Ndd or Udd. For example,
if ON(VALCNT,N05) results in overflow, you can use ON(VALCNT,N06)
instead.
If E'pattern' is specified, Ndd or Udd is ignored, because d is determined
from the pattern.
LIST(listdd)
TITLE('string')
TLEFT
TFIRST
PAGE
DATE
DATE(abcd)
DATENS(abc)
656
OCCUR Operator
YDDD(abc)
YDDDNS(ab)
TIME
TIME(abc)
TIMENS(ab)
BLANK
PLUS
For ON(VALCNT), PLUS is treated as BLANK.
NOCC
HEADER('string1')
HEADER(NONE)
NOHEADER
LINES(n)
INDENT(n)
657
OCCUR Operator
BETWEEN(n)
TBETWEEN(n)
ALLDUPS
Limits the ON values printed to those that occur more than once (that is, those
with duplicate field values). The ON values are printed when value count > 1.
ALLDUPS is equivalent to HIGHER(1).
NODUPS
Limits the ON values printed to those that occur only once (that is, those with
no duplicate field values). The ON values are printed when value count = 1.
NODUPS is equivalent to EQUAL(1) or LOWER(2).
HIGHER(x)
Limits the ON values printed to those that occur more than x times. The ON
values are printed when value count > x.
x must be specified as n or +n where n can be 1 to 15 decimal digits.
LOWER(y)
Limits the ON values printed to those that occur less than y times. The ON
values are printed when value count < y.
y must be specified as n or +n where n can be 1 to 15 decimal digits.
EQUAL(v)
Limits the ON values printed to those that occur v times. The ON values are
printed when value count = v.
v must be specified as n or +n where n can be 1 to 15 decimal digits.
VSAMTYPE(x)
on page 575.
WIDTH(n)
LONGLINE
UZERO
Specifies that -0 and +0 are to be treated as unsigned zero values, that is, as the
same value. With UZERO, -0 and +0 are treated as positive for ON processing.
UZERO overrides the default of treating -0 and +0 as signed zero values, that
is, as different values. Without UZERO, -0 is treated as negative and +0 is
treated as positive for ON processing.
LISTSDB or LISTNOSDB
See the discussion of these operands on the DISPLAY statement in DISPLAY
658
OCCUR Operator
OCCUR examples
Although the OCCUR operators in the examples in this section could all be
for clarity. See DISPLAY operator on page 594 for additional examples of
tailoring the report format.
Example 1
OCCUR FROM(SOURCE) LIST(VOLSERS) ON(40,6,CH) ON(VALCNT)
Prints, in the VOLSERS data set:

v A data line for each unique ON(40,6,CH) value in the standard format
containing:
The characters from positions 40-45 of the SOURCE data set for the unique
value
The count of occurrences in the SOURCE data set of the unique value
The VOLSERS output starts on a new page and looks as follows (the first 2 records
(40,6,CH)
ABC001
ABC002
.
.
.
VALUE COUNT
000000000000025
000000000000011
.
.
.
The heading line appears at the top of each page.
Example 2
OCCUR FROM(IN) LIST(LIST1) TITLE(
3090 Distribution
) PAGE HEADER(Data Centers) ON(VALCNT) HEADER(State) ON(1,16,CH) HEADER(3090s) ON(25,3,PD) BLANK
Prints, in the LIST1 data set:

v A title line containing the specified title and the page number
v A data line for each unique ON(1,16,CH) and ON(25,3,PD) value in the BLANK
format containing:
The count of occurrences in the IN data set of the unique value
The characters from positions 1-16 of the IN data set for the unique value
The packed decimal values from positions 25-27 of the IN data set for the
unique value
The LIST1 output starts on a new page and looks as follows (the first 2 records are
659
OCCUR Operator
3090 Distribution
Data Centers
--------------12
6
.
.
.
- 1 -
State
---------------Alabama
Alabama
.
.
.
3090s
-----1
2
.
.
.
Example 3
OCCUR FROM(FAILURES) LIST(CHECKIT) BLANK HIGHER(4) DATE TITLE(Possible System Intruders on ,System) TITLE(Sysplex, in ,Location) PAGE TBETWEEN(2) HEADER(,Userid) ON(23,8,CH) HEADER(Logon failures,(More than 4)) ON(VALCNT)
Assume that the SYMNAMES data set contains the following symbols:
System,S&SYSNAME
Sysplex,S&SYSPLEX
Location,San Jose
Prints, in the CHECKIT data set:

v A first title line containing the date, the string and System name specified in the
first TITLE operand, and the page number. Two blanks appear between the title
elements in the first title line as specified by the TBETWEEN(2) operand. A
blank line follows the first title line.
v A second title line containing the Sysplex name, string and Location specified in
the second TITLE operand. A blank line follows the second title line.
v An underlined three-line heading. The three-line heading for the first ON field
has blanks on line1 and line2 and 'Userid' on line3. The heading for the second
ON field has 'Number of' on line1, 'Logon Failures' on line2 and '(More than 4)'
on line3.
v A data line for each unique ON(23,8,CH) value for which there are more than 4
occurrences, in the BLANK format, containing:
The characters from positions 23-30 of the FAILURES data set
The count of occurrences of the characters from positions 23-30 of the
FAILURES data set
The CHECKIT output starts on a new page and looks as follows (the first 2 records
07/22/08
Possible System Intruders on EDS3
- 1 -
MAS3 in San Jose
Userid
-------B7234510
D9853267
...
Logon failures
(More than 4)
--------------5
11
The title lines and underlined three-line heading lines appear at the top of each
page.
660
OCCUR Operator
Example 4
OCCUR FROM(VARIN) LIST(ONCE) TITLE(Record lengths that occur only once) TIME(12:) DATE(DMY.) ON(VLEN) NODUPS BLANK
Prints, in the ONCE data set:

v A title line containing the specified title and the time and date
v A heading line containing the standard heading
v A data line for each record length for which there is only one occurrence, in the
BLANK format, containing the record length
The ONCE output starts on a new page and looks as follows (the first 2 records
Record lengths that occur only once
09:52:17 am
21.10.92
RECORD LENGTH
57
61
.
.
.
The title line and heading line appear at the top of each page.
Example 5
OCCUR FROM(BRANCH) LIST(CALLRPT)DATENS(4MD)TITLE(Yearly Branch Phone Call Counts)HEADER(Phone Number) ON(7,10,ZD,E(999)-999-9999)HEADER(Calls) ON(VALCNT,A1,N05)INDENT(5) BETWEEN(10)
Prints, in the CALLRPT data set:

v A title line containing the date (without separators) and the specified title.
v A heading line containing the specified underlined headings.
v A data line for each unique ON(7,10,ZD) value containing:
The zoned decimal value from positions 7-16 of the BRANCH data set printed
as (ddd)-ddd-dddd according to the E'pattern' formatting item.
The count of occurences of this value printed as dd,ddd according to the A1
and N05 formatting items.
The report is indented by five blanks as specified by the INDENT(5) operand, and
ten blanks appear between the columns as specified by the BETWEEN(10)
operand.
The CALLRPT output starts on a new page and looks as follows:
20020316
Phone Number
-------------(037)-325-1807
(216)-721-5530
(218)-062-7214
Yearly Branch Phone Call Counts

Calls
------3,125
2,087
872
661
RANGE Operator
RANGE operator
RANGE FROM(indd)
ON(p,m,f)
ON(VLEN)
HIGHER(x)
LOWER(y)
HIGHER(x) LOWER(y)
EQUAL(v)
NOTEQUAL(w)

VSAMTYPE(x)
Prints a message containing the count of values in a specified range for a specific
numeric field.
prints a message containing the range count as determined by its E35 user exit.
The range can be specified as higher than x, lower than y, higher than x and lower
than y, equal to v, or not equal to w, where x, y, v, and w are signed or unsigned
decimal values. If the range is specified as higher than x and lower than y, it must
be a valid range (for example, higher than 5 and lower than 6 is not a valid range,
because there is no integer value that satisfies the criteria).
operator.
FROM(indd)
ON(p,m,f)
Specifies the position, length, and format of the numeric field to be used for
this operation.
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3 4
| p=
1
2
3
4
5
6
7
8
662
Format Code
Length
Description
BI
1 to 8 bytes
Unsigned binary
FI
1 to 8 bytes
Signed fixed-point
PD
1 to 16 bytes
ZD
1 to 31 bytes
RANGE Operator
Format Code
Length
Description
CSF or FS

UFF
SFF

descriptions.

bad value in a message and terminates the operation.
For a ZD, PD, CSF, FS, or SFF format field, a negative zero value is treated as a
positive zero value.
ON(VLEN)
HIGHER(x)
Values higher than x are counted as contained in the range. If only HIGHER(x)
is specified, the range count is incremented when x < value. If LOWER(y) is
also specified, the range count is incremented when x < value < y.
x must be specified as n, +n, or -n where n can be 1 to 31 decimal digits.
LOWER(y)
Values lower than y are counted as contained in the range. If only LOWER(y)
is specified, the range count is incremented when value < y. If HIGHER(x) is
also specified, the range count is incremented when x < value < y.
y must be specified as n, +n, or -n where n can be 1 to 31 decimal digits.
EQUAL(v)
Values equal to v are counted as contained in the range. The range count is
incremented when v = value.
v must be specified as n, +n, or -n where n can be 1 to 31 decimal digits.
NOTEQUAL(w)
Values not equal to w are counted as contained in the range. The range count
is incremented when w = value.
w must be specified as n, +n, or -n where n can be 1 to 31 decimal digits.
VSAMTYPE(x)
on page 575.
663
RANGE Operator
RANGE example
RANGE FROM(DATA1) ON(VLEN) HIGHER(10)
RANGE FROM(DATA2) ON(31,18,ZD) LOWER(+123456789012345678)
RANGE FROM(DATA3) ON(29001,4,FI) HIGHER(-10000) LOWER(27)
RANGE FROM(DATA2) ON(45,3,PD) EQUAL(-999)
RANGE FROM(DATA3) ON(40,1,BI) NOTEQUAL(199)
The first RANGE operator prints a message containing the count of binary values
from positions 1-2 of the DATA1 data set that are higher than 10.
The second RANGE operator prints a message containing the count of zoned
decimal values from positions 31-48 of the DATA2 data set that are lower than
123456789012345678.
The third RANGE operator prints a message containing the count of fixed-point
values from positions 29 001-29 004 of the DATA3 data set that are higher than -10
000 but lower than 27.
The fourth RANGE operator prints a message containing the count of packed
decimal values from positions 45-47 of the DATA2 data set that are equal to -999.
The fifth RANGE operator prints a message containing the count of binary values
from position 40 of the DATA3 data set that are not equal to 199. This RANGE
operator could be used to count the number of records that do not have 'G' in
position 40, because 199 (X'C7') is the EBCDIC code for 'G'. Alternatively, the
COUNT operator could be used with OMIT COND=(40,1,CH,EQ,C'G').
RESIZE operator
RESIZE FROM(indd) TO(outdd) TOLEN(n)

USING(xxxx)
Produces fixed length output records, with the specified TOLEN length, from fixed
length input records of a different length, as follows:
v If the input length is smaller than the requested TOLEN size, multiple smaller
input records are combined into one larger output record of size TOLEN. If the
combined input records do not completely fill up an output record, the output
record will be padded on the right with blanks. For example, if we have five
input records of 20 bytes and TOLEN is 70, two output records of 70 bytes will
be created. The first output record will have input records 1-3 (60 bytes)
followed by 10 blanks, and the second output record will have input records 4-5
(40 bytes) followed by 30 blanks. Note that the first 10 bytes of input record 4
are not used for the first output record; bytes or words are not "wrapped"
between records.
To illustrate, if the following RESIZE operator was specified:
RESIZE FROM(FB20) TO(FB70) TOLEN(70)
and FB20 had these 20 byte input records:

<111111111111111111>
<222222222222222222>
<333333333333333333>
<444444444444444444>
<555555555555555555>
FB70 would have these 70 byte output records:
664
RESIZE Operator
<111111111111111111><222222222222222222><333333333333333333>
<444444444444444444><555555555555555555>
v If the input length is larger than the requested TOLEN size, each larger input
record is broken up into multiple smaller output records of size TOLEN. If a
broken up input record does not completely fill up the multiple output records,
the last of the multiple output records will be padded on the right with blanks.
For example, if we have two input records of 50 bytes and TOLEN is 20, six
output records of 20 bytes each will be created. The first two output records will
have bytes 1-20 and 21-40 of the first input record (40 bytes), respectively. The
third output record will have bytes 41-50 of the first input record (10 bytes)
followed by 10 blanks. Likewise, the fourth output record will have bytes 1-20 of
the second input record, the fifth output record will have bytes 21-40 of the
second input record, and the sixth output record will have bytes 41-50 of the
second input record followed by 10 blanks.
To illustrate, if the following RESIZE operator was specified:
RESIZE FROM(F50) TO(F20) TOLEN(20)
and F50 had these 50 byte input records:

<111111111111111111><222222222222222222><33333333>
<444444444444444444><555555555555555555><66666666>
F20 would have these 20 byte output records:

<111111111111111111>
<222222222222222222>
<33333333>
<444444444444444444>
<555555555555555555>
<66666666>
You must specify the FROM(indd), TO(outdd) and TOLEN(n) operands. The
FROM data set must be fixed-length (for example, RECFM=FB). The TO data set
must also be fixed-length (unless you use an OUTFIL statement with the FTOV
operand to change the fixed-length resized records to variable-length output
records). If a VSAM input data set is used, it will be treated as TYPE=F
(fixed-length) by default. If you specify a variable-length input data set (or use
TYPE=V for a VSAM input data set), the RESIZE operation will be terminated.
The USING(xxxx) operand is optional; do not supply your own MODS or
OUTREC statement.
DFSORT is called to copy or sort the indd data set, as appropriate, before the
records are resized. ICETOOL uses its E35 exit to create a larger record from
multiple smaller records, or to create multiple smaller records from a larger record.
ICETOOL passes the EQUALS option to DFSORT to ensure that if records are
sorted, duplicate records are kept in their original input order when resized.
If USING(xxxx) is specified, any SORT, INCLUDE, OMIT, INREC, or SUM
statement specified in xxxxCNTL is processed before the records are resized. The
order of the records, and the input length, will be affected by these control
statements. Any OUTFIL statements specified in xxxxCNTL are processed after the
records are resized.
inappropriate for a RESIZE operator, you can take one of the following actions:
665
RESIZE Operator
FROM(indd)
on page 575.
TO(outdd)
See the discussion of this operand on the DATASORT statement in
DATASORT operator on page 585.
TOLEN(n)
Specifies the record length you want ICETOOL to use for the resized output
records. n can be 1 to 32760. n must not be equal to the input record length.
USING(xxxx)
valid in a ddname of the form xxxxCNTL. xxxx must not be SYSx. If
USING(xxxx) is specified, an xxxxCNTL DD statement must be present, and
the control statements in it must conform to the rules for DFSORT's
SORTCNTL data set.
RESIZE examples
Although the RESIZE operators in the examples in this section could all be
for clarity.
Example 1
RESIZE FROM(IN1) TO(OUT1) TOLEN(40)
This example illustrates how you can create larger records from smaller records.
The IN1 data set has RECFM=FB and LRECL=10 with these 10-byte records:
Bird
Bluejay
4
Charlie
Rodent
Rat
2
Sara
The OUT1 data set has RECFM=FB and LRECL=40 with these 40 byte records:
Bird
Rodent
Bluejay
Rat
4
2
Charlie
Sara
Example 2
RESIZE FROM(OLD) TO(NEW) TOLEN(15) USING(CTL1)
This example illustrates how you can create larger records from a subset of smaller
sorted records.
666
RESIZE Operator
OMIT COND=(2,4,ZD,EQ,0)
The OLD data set has RECFM=FB and LRECL=5 with these 5-byte records:
C0005
B0000
A0008
I1234
F0053
D0123
H0001
G0000
E0022
The NEW data set will have RECFM=FB and LRECL=15 with these 15-byte
records:
A0008C0005D0123
E0022F0053H0001
I1234
Note that before the records were resized, the two records with 0 in positions 2-5
were omitted, and the remaining records were sorted as directed by the DFSORT
control statements in CTL1CNTL. The last output record was padded with blanks
on the right to 15 bytes.
Example 3
RESIZE FROM(IN3) TO(OUT3) TOLEN(3) USING(CTL2)
This example illustrates how you can break up large records into multiple smaller
records.
OUTFIL FNAMES=OUT3,OMIT=(1,3,CH,EQ,C ),OVERLAY=(10:X)
The IN3 data set has RECFM=FB and LRECL=18 with these 18-byte records:
000111222333444555
666777888999
Every 3-byte field in each large IN3 record will be broken up into a single 3-byte
field and then padded on the right with blanks to 10-bytes. TOLEN(3) indicates
that the resized records will have a length of 3 bytes. OVERLAY=(10:X) expands
each resized record to 10 bytes in OUT3. OMIT=(1,3,CH,EQ,C' ') removes any
resized records that are completely blank (that is, the two blank resized records
resulting from the blanks in positions 13-18 of the second input record).
The OUT3 data set will have RECFM=FB and LRECL=10 with these 10-byte
records:
000
111
222
333
444
555
666
777
888
999
667
SELECT Operator
SELECT operator
SELECT FROM(indd)
ALLDUPS
NODUPS
HIGHER(x)
LOWER(y)
EQUAL(v)
FIRST
FIRST(u)
LAST
FIRSTDUP
FIRSTDUP(w)
LASTDUP
TO(outdd)
DISCARD(savedd)
TO(outdd) DISCARD(savedd)
ON(p,m,f)
ON(VLEN)

VSAMTYPE(x)
UZERO
USING(xxxx)
Selects records from an input data set based on meeting criteria for the number of
times specified numeric or character field values occur. This makes it possible to
only keep records with duplicate field values, only keep records with no duplicate
field values, only keep records with field values that occur more than, less than, or
exactly n times, only keep the first or first n duplicate records with each field
value, or only keep the first or last record with each unique or duplicate field
value. From 1 to 10 fields can be specified. At least one ON(VLEN) or ON(p,m,f)
field must be specified; all such ON fields specified will be used to determine the
"value count" (that is, the number of times the ON values occur) to be matched
against the criteria.
DISCARD(savedd) can be used to save the records that do not meet the criteria
(that is, the discarded records), in the savedd data set. DISCARD(savedd) can be
used with or without TO(outdd).
DFSORT is called to sort the indd data set. ICETOOL uses its E35 exit to determine
which records to include in the outdd data set or savedd data set. ICETOOL passes
the EQUALS option to DFSORT to ensure that duplicates are kept in their original
input order.
The DFSORT control statements in xxxxCNTL are used if USING(xxxx) is specified.
Do not supply your own MODS, SUM or OUTREC statement.
You can use comment statements. You can use INCLUDE, OMIT, INREC, OPTION,
SORT, or OUTFIL statements providing you observe these rules:
v You can use an INCLUDE or OMIT statement to remove input records before
SELECT processing.
v You can use an INREC statement to reformat input records before SELECT
processing. You can use INREC's PARSE, BUILD (FIELDS), OVERLAY, FINDREP,
IFTHEN, or IFOUTLEN functions. If your INREC statement changes the starting
668
SELECT Operator
position of an ON field, you must specify the new starting position for that ON
field. For example, if your input records have a CH key at positions 1-5 and you
use an INREC statement like this:
INREC FIELDS=(25:1,50)
you must specify ON(25,5,CH) instead of ON(1,5,CH).

v If you specify a SORT statement, you must specify each ON field as a p,m,f,A
sort field and these sort fields must be in the same order as the ON fields. After
these sort fields, you can specify additional p,m,f,A or p,m,f,D sort fields. The
additional sort fields will be used for sorting, but not for selecting. For example,
if you use a SELECT statement like this:
SELECT FROM(IN) TO(OUT) ON(21,5,CH) FIRST(3) USING(CTL1)
you can use a SORT statement in CTL1CNTL like this:

SORT FIELDS=(21,5,CH,A,41,6,CH,D)
The records will be sorted by the 21,5,CH,A field and the 41,6,CH,D field, but
only selected by the 21,5,CH field. This would allow you to select the three
highest 41,6,CH values for each 21,5,CH value.
v If you specify TO(outdd) without DISCARD(savedd), you can further process
the outdd records after SELECT processing using an OUTFIL statement like this:

...
v If you specify DISCARD(savedd) without TO(outdd), you can further process

the savedd records after SELECT processing using one (and only one) OUTFIL
statement like this:
OUTFIL FNAMES=savedd,...
v If you specify TO(outdd) and DISCARD(savedd), you can further process the
outdd and savedd records after SELECT processing using two (and only two)
OUTFIL statements like this:
Both statements must be specified in the order shown with at least the FNAMES
parameter. For example, to further modify only the DISCARD data set, you
could use statements like this:
OUTFIL FNAMES=OUT
OUTFIL FNAMES=SAVE,INCLUDE=(21,3,ZD,GT,+25)
ICETOOL requires extra storage for SELECT processing, over and above what is
normally needed by ICETOOL and DFSORT, in order to save your records until it
can determine whether or not they meet your specified criteria. In most cases, only
a small amount of storage is needed and can be obtained (above 16MB virtual).
However, for a FROM data set with a large record length and criteria requiring
many saved records, a large amount of storage is needed. For example, with a
record length of 32756 and HIGHER(99), over 3 MBs of storage is needed. If
ICETOOL cannot get the storage it needs, it issues a message and terminates the
SELECT operation. Increasing the REGION by the amount indicated in the
message may allow ICETOOL to run successfully.
669
SELECT Operator
inappropriate for a SELECT operator, you can specify USING(xxxx) and take one
of the following actions:

FROM(indd)
on page 575.
TO(outdd)
Specifies the ddname of the output data set to which DFSORT will write the
records it selects for the operation (that is, the records that meet the specified
criteria). Thus, the outdd data set will contain the records selected by
ALLDUPS, NODUPS, HIGHER(x), LOWER(y), EQUAL(v), FIRST, FIRST(u),
LAST, FIRSTDUP, FIRSTDUP(w) or LASTDUP.
that conforms to the rules for DFSORT's SORTOUT data set (if the DISCARD
operand is not specified) or OUTFIL data set (if the DISCARD operand is
specified).
TO and DISCARD can both be specified. If DISCARD is not specified, TO must
be specified. If TO is not specified, DISCARD must be specified.
specified in the FROM or DISCARD operand.
DISCARD(savedd)
records it does not select for this operation (that is, the records that do not
meet the specified criteria). Thus, the savedd data set will contain the records
discarded by ALLDUPS, NODUPS, HIGHER(x), LOWER(y), EQUAL(v), FIRST,
FIRST(u), LAST, FIRSTDUP, FIRSTDUP(w) or LASTDUP.
A savedd DD statement must be present and must define an output data set
that conforms to the rules for DFSORT's OUTFIL data set.
The ddname specified in the DISCARD operand must not be the same as the
ddname specified in the FROM or TO operand.
ON(p,m,f)
used for this operation.
670
SELECT Operator
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3
4
| p=
1
2
3
4
5
6
7
8
If INREC is specified, p must refer to the record as reformatted by INREC.

position 32752, or beyond the end of a record. If INREC is specified, a field
must not extend beyond the end of the record as reformatted by INREC. The
maximum length for a field depends on its format.
f specifies the format of the field as shown in the following table.
Format Code
Length
Description
BI
1 to 1500 bytes
Unsigned binary
FI
1 to 256 bytes
Signed fixed-point
PD
1 to 16 bytes
ZD
1 to 31 bytes
CH
1 to 4000 bytes
Character
CSF or FS
1 to 32 bytes

UFF
1 to 44 bytes
SFF
1 to 44 bytes
descriptions.

zoned decimal values F2F3C1, F2F3F1 and 020301 are counted as only one
unique value.
unique value.
v Digits are not checked for validity.
ON(VLEN)
ALLDUPS
Limits the records selected to those with ON values that occur more than once
(value count > 1). You can use this operand to keep just those records with
duplicate field values.
ALLDUPS is equivalent to HIGHER(1).
NODUPS
Limits the records selected to those with ON values that occur only once
(value count = 1). You can use this operand to keep just those records with no
duplicate field values.
NODUPS is equivalent to EQUAL(1) or LOWER(2).
671
SELECT Operator
HIGHER(x)
Limits the records selected to those with ON values that occur more than x
times (value count > x). You can use this operand to keep just those records
with field values that occur more than x times.
x must be specified as n or +n where n can be 0 to 99.
LOWER(y)
Limits the records selected to those with ON values that occur less than y
times (value count < y). You can use this operand to keep just those records
with field values that occur less than y times.
y must be specified as n or +n where n can be 0 to 99.
EQUAL(v)
Limits the records selected to those with ON values that occur v times (value
count = v). You can use this operand to keep just those records with field
values that occur v times.
FIRST
(value count = 1) and the first record of those with ON values that occur more
than once (value count > 1). You can use this operand to keep just the first
record for each unique field value.
FIRST is equivalent to FIRST(1).
FIRST(u)
(value count = 1) and the first u records of those with ON values that occur
more than once (value count > 1). You can use this operand to keep just the
first u records for each unique field value.
LAST
(value count = 1) and the last record of those with ON values that occur more
than once (value count > 1). You can use this operand to keep just the last
record for each unique field value.
FIRSTDUP
Limits the records selected to the first record of those with ON values that
occur more than once (value count > 1). You can use this operand to keep just
the first record of those records with duplicate field values.
FIRSTDUP is equivalent to FIRSTDUP(1).
FIRSTDUP(w)
Limits the records selected to the first w records of those with ON values that
the first w records of those records with duplicate field values.
w must be specified as n or +n where n can be 1 to 999999999999999.
LASTDUP
Limits the records selected to the last record of those with ON values that
the last record of those records with duplicate field values.
672
SELECT Operator
VSAMTYPE(x)
on page 575.
UZERO
See the discussion of this operand on the OCCUR statement in OCCUR
USING(xxxx)
2. Should generally be used only for an INCLUDE or OMIT statement, an
INREC statement, a SORT statement, comment statements, or OUTFIL
statements as described for SELECT operator on page 668.
SELECT examples
Although the SELECT operators in the examples in this section could all be
for clarity.
Example 1
SELECT FROM(INPUT) TO(DUPS) ON(11,8,CH) ON(30,44,CH) ALLDUPS
Sorts the INPUT data set to the DUPS data set, selecting only the records from
INPUT with characters in positions 11-18 and characters in positions 30-73 that
occur more than once (that is, only records with duplicate ON field values).
The DUPS data set might look as follows (several records are shown for illustrative
purposes):
USR002
DFSRT2
DFSRT5
DFSRT1
SYS003
DFSRT2
USR003
.
.
.
BETTEN
BETTEN
BOENIG
BOENIG
BOENIG
BOENIG
BOENIG
.
.
.
12
5
20
20
20
20
20
.
.
.
DOC.EXAMPLES
DOC.EXAMPLES
MYDATA
MYDATA
MYDATA
SORTST1.TEST
SORTST1.TEST
.
.
.
Example 2
SELECT FROM(INPUT) TO(ONLYONE) ON(23,3,FS) NODUPS
Sorts the INPUT data set to the ONLYONE data set, selecting only the records
from INPUT with floating sign values in positions 23-25 that occur just once (that
is, only records with no duplicate ON field values).
The ONLYONE data set might look as follows (several records are shown for
illustrative purposes):
DFSRT2
DFSRT1
USR002
SYS003
BOENIG
PACKER
BETTEN
YAEGER
5
8
12
32
DOC.EXAMPLES
ICETOOL.SMF.RUNS
DOC.EXAMPLES
ICETOOL.TEST.CASES
673
SELECT Operator
DFSRT2
.
.
.
CORNELL
.
.
.
108
.
.
.
FS.TEST.CASES
.
.
.
Example 3
SELECT FROM(FAILURES) TO(CHECKOUT) ON(28,8,CH) ON(1,5,CH) HIGHER(3)
Sorts the FAILURES data set to the CHECKOUT data set, selecting only the
records from FAILURES with characters in positions 28-35 and characters in
positions 1-5 that occur more than three times (that is only records with four or
more duplicate ON field values).
The CHECKOUT data set might look as follows (several records are shown for
03/12/04
03/12/04
03/12/04
03/12/04
03/06/04
03/06/04
03/06/04
03/06/04
03/06/04
.
.
.
08:36:59
09:27:32
09:03:18
08:56:13
15:12:01
14:57:00
15:43:19
16:06:39
15:22:08
.
.
.
A3275647
A3275647
A3275647
A3275647
C3275647
C3275647
C3275647
C3275647
C3275647
.
.
.
Example 4
SELECT FROM(BOOKS) TO(PUBLISHR) ON(29,10,CH) FIRST
Sorts the BOOKS data set to the PUBLISHR data set, selecting only the records
from BOOKS with characters in positions 29-38 that occur only once and the first
record of those with characters in positions 29-38 that occur more than once (that
is, one record for each unique ON field value).
The PUBLISHR data set might look as follows (several records are shown for
Banana Slugs I Have Known
Toads on Parade
Pets Around the World
.
.
.
Brent
Cooper
Davis
.
.
.
Animals
Animals
Animals
.
.
.
Example 5
SELECT FROM(BOOKS) TO(PUBLISHR) ON(29,10,CH) FIRST DISCARD(SAVEREST)
This example creates the same PUBLISHR data set as Example 4. In addition, it
creates a SAVEREST data set that contains all of the records not written to the
PUBLISHR data set. The SAVEREST data set might look as follows (several records
are shown for illustrative purposes):
How to Talk to Your Amoeba
What Buzzards Want
Birds of Costa Rica
.
.
.
674
Brent
Davis
Davis
Animals
Animals
Animals
SELECT Operator
Example 6
SELECT FROM(MASTPULL) TO(MATCH) ON(5,8,CH) FIRSTDUP
This example shows how you can use a list of account numbers in a "pull" data set
to only select records with those account numbers from a "master" data set. The
MASTPULL DD would have the "master" data set and "pull" data set concatenated
together (in that order).
The SELECT operator sorts the concatenated data sets and selects only the first
record of those with characters in positions 5-12 that occur more than once (that is,
one record for each duplicate ON field value). Because the "master" data set is first
in the concatenation, the selected records will come from the "master" data set.
If the "master" data set looked like this:
A52
N92
B12
J73
Q28
RB172832
MX328126
LB018725
AB007231
SP973004
2001/03/15
2001/01/27
2000/12/28
2001/02/13
2000/11/19
and the "pull" data set looked like this:

AB007231
RS859276
QN005001
MX328126
the MATCH data set would look like this:

J73 AB007231 2001/02/13
N92 MX328126 2001/01/27
Note: This example assumes that there are not any duplicate account numbers in
either the "master" or "pull" data sets. If that is not true, you can use SELECT with
FIRST or LAST, for the appropriate data set, to make it true. For example, if your
"master" data set has duplicate account numbers and you want to select the first
account number from the "master" data set for each account number in the "pull"
data set, you could use the following statements:
SELECT FROM(MASTER) TO(TEMP) ON(5,8,CH) FIRST
SELECT FROM(TEMPPULL) TO(MATCH) ON(5,8,CH) FIRSTDUP
The TEMPPULL DD would have the temporary data set and "pull" data set
concatenated together (in that order).
Example 7
SELECT FROM(INPUT) TO(ONLYONE) ON(23,3,FS) NODUPS USING(CTL1)
This example shows how you can use USING(xxxx) to supply an OUTFIL
statement to modify the TO data set. SELECT chooses the same output records as
for Example 2 on page 673, but an OUTFIL statement is used to further modify
those records for output to the ONLYONE data set.
OUTFIL FNAMES=ONLYONE,
REMOVECC,
INCLUDE=(23,3,FS,LT,100),
OUTREC=(1:1,7,8:C|,11:11,7,19:C|,23:23,3,FS,M11,
27:C|,30:30,15),
TRAILER1=(/,TOTAL= ,TOT=(23,3,FS,M11,LENGTH=6))
675
SELECT Operator
and the ONLYONE data set might look as follows:
DFSRT2
DFSRT1
USR002
SYS003
|
|
|
|
EISSLER
PACKER
EISSLER
YAEGER
|
|
|
|
005
008
012
032
|
|
|
|
DOC.EXAMPLES
ICETOOL.SMF.RUN
DOC.EXAMPLES
ICETOOL.TEST.CA
TOTAL= 000057
Example 8
This example shows how you can use USING(xxxx) to supply an OMIT statement
to remove certain input records and an INREC statement to reformat certain input
records before SELECT processing begins.
SELECT FROM(INB) TO(OUTB) ON(8,10,CH) FIRSTDUP USING(BIRD)
Let's say the INB data set looks like this:

TYPE1
TYPE1
TYPE1
TYPE1
TYPE1
TYPE1
TYPE1
TYPE2
TYPE2
TYPE2
TYPE2
TYPE3
TYPE3
CROWS
PIGEONS
HAWKS
ROBINS
SPARROWS
CHICKENS
RAVENS
PIGEONS
ROBINS
HAWKS
TERNS
EAGLES
STARLINGS
We want to remove the TYPE3 records and then select one record for each type of
bird that appears in both a TYPE1 and a TYPE2 record. We could use the following
control statements in BIRDCNTL:
* Remove the TYPE3 records.
OMIT COND=(1,5,CH,EQ,CTYPE3)
* Reformat the TYPE2 records to place the bird name in
* same place as in the TYPE1 records so we can match them.
INREC IFTHEN=(WHEN=(1,5,CH,EQ,CTYPE2),
BUILD=(8:18,10))
The OUTB data set would look like this:

TYPE1
TYPE1
TYPE1
HAWKS
PIGEONS
ROBINS
Example 9
This example shows how you can use USING(xxxx) to supply a SORT statement to
alter the records that are selected.
SELECT FROM(IN) TO(OUT) ON(1,5,CH) FIRST(3) USING(CTL1)
Let's say the IN data set looks like this:

FRANK
FRANK
FRANK
FRANK
FRANK
FRANK
VICKY
VICKY
676
00015
00012
00003
00018
00005
00035
00022
00007
SELECT Operator
VICKY
VICKY
VICKY
VICKY
00014
00028
00002
00015
We want to select the three records with each name that have the highest counts. If
we just used ON(1,5,CH) without any CTL1CNTL statements, we'd get the first
three records for each name without regard to the count. The OUT data set would
look like this:
FRANK
FRANK
FRANK
VICKY
VICKY
VICKY
00015
00012
00003
00022
00007
00014
To get the three records with the highest counts for each name, we can use the
following SORT statement in CTL1CNTL:
SORT FIELDS=(1,5,CH,A,7,5,ZD,D)
The records will be sorted in ascending order on the name field, and in descending
order on the count field. By sorting descending on the count, we ensure that the
three records with the highest counts are the first three records for each name.
Thus, when ON(1,5,CH) selects the first three records, they will be those with the
highest counts. The OUT data set will look like this:
FRANK
FRANK
FRANK
VICKY
VICKY
VICKY
00035
00018
00015
00028
00022
00015
SORT operator
SORT
FROM(indd)
JKFROM
USING(xxxx)

,
TO( outdd
VSAMTYPE(x)
)

LOCALE(name)
LOCALE(CURRENT)
LOCALE(NONE)
SERIAL
Sorts a data set to one or more output data sets.

DFSORT is called to sort the indd data set to the outdd data sets using the
DFSORT control statements in xxxxCNTL. You must supply a DFSORT SORT
statement in the xxxxCNTL data set to indicate the control fields for the sort. You
can use additional DFSORT statements in the xxxxCNTL data set to sort a subset
of the input records (INCLUDE or OMIT statement; SKIPREC and STOPAFT
options; OUTFIL INCLUDE, OMIT, SAVE, STARTREC, ENDREC, SAMPLE, SPLIT,
SPLITBY, and SPLIT1R operands; user exit routines), reformat records for output
(INREC, OUTREC, and OUTFIL statements, user exit routines), and so on.
The active locale's collating rules affect SORT processing as explained in SORT
control statement on page 442. If an INCLUDE or OMIT statement or an OUTFIL
677
SORT Operator
INCLUDE or OMIT operand is specified in the xxxxCNTL data set, the active
locale's collating rules affect INCLUDE and OMIT processing as explained in the
Cultural Environment Considerations discussion in INCLUDE control
inappropriate for a SORT operator, you can take one of the following actions:
in the xxxxCNTL data set in conjunction with the USING(xxxx) operand.

2. Use xxxxWKdd DD statements to override the use of dynamic allocation in
conjunction with the USING(xxxx) operand. Refer to SORTWKdd DD
statement on page 71 for details.
FROM(indd)
on page 575.
JKFROM
Specifies you are using a JOINKEYS application with this SORT operator to
sort the joined records. You must provide a USING(xxxx) operand. In
xxxxCNTL, you must provide a SORT statement, a JOINKEYS statement with
F1=ddname1 for the F1 file, and a JOINKEYS statement with F2=ddname2 for
the F2 file, as well as JOIN and REFORMAT statements as needed.
USING(xxxx)
An xxxxCNTL DD statement must be present, and the control statements in it
must conform to the rules for DFSORT's SORTCNTL data set.
The xxxxCNTL data set must contain a SORT statement. If TO is not specified,
the xxxxCNTL data set must also contain either one or more OUTFIL
statements or a MODS statement for an E35 routine that disposes of all
records. Other statements are optional.
If you want to override dynamic allocation of work data sets for this operation,
you can use xxxxWKdd DD statements for that purpose.
TO(outdd,...)
specified, DFSORT is called once to sort the indd data set to the outdd data set
using SORTOUT processing; the outdd data set must conform to the rules for
DFSORT's SORTOUT data set. If multiple outdd data sets are specified and
SERIAL is not specified, DFSORT is called once to sort the indd data set to the
678
SORT Operator
outdd data sets using OUTFIL processing; the outdd data sets must conform to
the rules for DFSORT's OUTFIL data sets.
A ddname specified in the FROM operand must not also be specified in the
TO operand.
VSAMTYPE(x)
on page 575.
LOCALE(name)
on page 575.
LOCALE(CURRENT)
on page 575.
LOCALE(NONE)
on page 575.
SERIAL
DFSORT is called to sort the indd data set to the first outdd data set using the
DFSORT control statements in the xxxxCNTL data set. If the sort operation is
successful, DFSORT is called as many times as necessary to copy the first
outdd data set to the second and subsequent outdd data sets. Therefore, for
maximum efficiency, use a disk data set as the first in a list of outdd data sets
on both disk and tape. If more than one outdd data set is specified, DFSORT
must be able to read the first outdd data set after it is written in order to copy
it to the other outdd data sets. Do not use a SYSOUT or DUMMY data set as
the first in a list of outdd data sets because:
v If the first data set is SYSOUT, DFSORT abends when it tries to copy the
v If the first data set is DUMMY, DFSORT copies the empty DUMMY data set
to the other outdd data sets (that is, all of the resulting outdd data sets are
empty).
Sort examples
This section includes 14 sort examples.
Example 1
* Method 1
SORT FROM(MASTER) TO(PRINT,TAPE,DISK) USING(ABCD)
* Method 2
SORT FROM(MASTER) TO(DISK,TAPE,PRINT) USING(ABCD) SERIAL
679
SORT Operator
This example shows two different methods for creating multiple sorted output data
sets. Assume that the ABCDCNTL data set contains:
SORT FIELDS=(15,20,CH,A,1,5,PD,D)
Method 1 requires one call to DFSORT, one pass over the input data set, and
allows the output data sets to be specified in any order. The SORT operator sorts
all records from the MASTER data set to the PRINT (SYSOUT), TAPE, and DISK
data sets, using the SORT statement in the ABCDCNTL data set and OUTFIL
processing.
Method 2 requires three calls to DFSORT, three passes over the input data set, and
imposes the restriction that the SYSOUT data set must not be the first TO data set.
The SORT operator sorts all records from the MASTER data set to the DISK data
set, using the SORT statement in the ABCDCNTL data set, and then copies the
resulting DISK data set to the TAPE and PRINT (SYSOUT) data sets. Because the
first TO data set is processed three times (written, read, read), placing the DISK
data set first is more efficient than placing the TAPE data set first. PRINT must not
be the first in the TO list because a SYSOUT data set cannot be read.
Example 2
* Method 1
SORT FROM(IN) TO(DEPT1) USING(DPT1)
* Method 2
SORT FROM(IN) USING(ALL3)
This example shows two different methods for creating sorted subsets of an input
data set. Assume that:
SORT FIELDS=(51,2,BI,A,18,5,CH,A,58,4,BI,A)


v The ALL3CNTL data set contains:

Method 1 requires three calls to DFSORT and three passes over the input data set:
v The first SORT operator sorts the records from the IN data set that contain D01
in positions 5-7 to the DEPT1 data set
v The second COPY operator sorts the records from the IN data set that contain
D02 in positions 5-7 to the DEPT2 data set
v The third COPY operator sorts the records from the IN data set that contain D03
in positions 5-7 to the DEPT3 data set.
Method 2 accomplishes the same result as method 1 but, because it uses OUTFIL
statements instead of TO operands, requires only one call to DFSORT and one pass
over the input data set.
680
SORT Operator
Example 3
SORT FROM(IN1) TO(FRANCE) USING(SRT1) LOCALE(FR_FR)
SORT FROM(IN1) TO(CANADA) USING(SRT1) LOCALE(FR_CA)
SORT FROM(IN1) TO(BELGIUM) USING(SRT1) LOCALE(FR_BE)
This example shows how sorted data for three different countries can be produced.
Assume that the SRT1CNTL data set contains:
SORT FIELDS=(5,20,CH,A,31,15,CH,A,1,4,FI,D,63,10,CH,D)
The first SORT operator sorts all records from the IN1 data set to the FRANCE
data set, using the SORT statement in the SRT1CNTL data set. The character (CH)
control fields are sorted according to the collating rules defined in locale FR_FR
(French language for France).
The second SORT operator sorts all records from the IN1 data set to the CANADA
control fields are sorted according to the collating rules defined in locale FR_CA
(French language for Canada).
The third SORT operator sorts all records from the IN1 data set to the BELGIUM
control fields are sorted according to the collating rules defined in locale FR_BE
(French language for Belgium).
SORT operator with JOINKEYS example

Here is an example of using one SORT operator for a simple JOINKEYS
application.
//SRTJK EXEC PGM=ICETOOL
//JNA DD DSN=MY.INPUTA,DISP=SHR
//JNB DD DSN=MY.INPUTB,DISP=SHR
//OUT DD SYSOUT=*
//TOOLIN DD *
* SORT operator with JOINKEYS application.
SORT JKFROM TO(OUT) USING(CTL1)
/*
//CTL1CNTL DD *
* JOINKEYS application control statements for SORT operator.
JOINKEYS F1=JNA,FIELDS=(5,4,A)
JOINKEYS F2=JNB,FIELDS=(11,4,A),SORTED
* Main task control statement for SORT operator
OPTION EQUALS
/*
SPLICE operator
SPLICE FROM(indd) TO(outdd) ON(p,m,f)
WITH(p,m)

WITHALL
WITHANY
WITHEACH
681
SPLICE Operator

KEEPNODUPS
KEEPBASE
USING(xxxx)
VSAMTYPE(x)
UZERO

VLENMAX
VLENOVLY
Splices together specified fields from records with matching numeric or character
field values (that is, duplicate values), but different information. This makes it
possible to join fields from different types of input records to create an output
record with information from two or more records.
Typically, you will want to reformat the records from two or more data sets to a
temporary MOD data set, and use that temporary MOD data set as input to the
SPLICE operator.
SPLICE examples on page 691 shows some techniques for splicing records from
different data sets together in a variety of ways to perform various file "join" and
"match" operations.
By default (when WITHALL, WITHANY and WITHEACH are not specified), one
spliced record is created for each set of duplicates by splicing the first duplicate
with specified fields from the last duplicate.
The first duplicate is treated as a "base record". The last duplicate is treated as an
"overlay record". Specified fields from the overlay record are overlaid on to the
base record. Thus, the output record consists of fields from the base (first) record
intermixed with specified fields from the overlay (last) record.
The records to be spliced can originate from two or more different input data sets.
From 1 to 10 ON fields can be used for the fields to match on. At least one
ON(p,m,f) field must be specified; all such ON fields specified will be used to
determine the matching records to be spliced together.
From 1 to 50 WITH fields can be used to specify the fields to be overlaid on the
base record from the overlay record. At least one WITH(p,m) field must be
specified; all such WITH fields specified will be overlaid on to the base record. All
other fields in the base record will be kept unchanged.
To illustrate the splicing process, if we had the following two fixed-length input
records with the base fields, ON field and WITH fields as shown:
BASE1
ON1
ON1
BASE2
BASE3 BASE4
WITHA
GGGGG
WITHB
the resulting spliced output record would be:

BASE1
ON1
BASE2
WITHA
BASE3 BASE4
WITHB
For variable-length records, by default (without VLENMAX or VLENOVLY), the

spliced record has the same length as the base record, and WITH fields from the
overlay record that are beyond the end of the base record do not appear in the
682
SPLICE Operator
spliced record. For example, if we had the following two records with the lengths
(in the RDW), ON field and WITH fields as shown:
25 | ON1 BASE1
BASE2
35 | ON1
WITH1
WITH2
the resulting spliced output record would be:

25 | ON1 BASE1 WITH1 BASE2
The WITH2 field is beyond the end of the base record, so it is not spliced.
However, if you specify VLENMAX, the spliced record is given the larger of the
base record length or overlay record length. If we specify VLENOVLY, the spliced
record is given the overlay record length. In either case, if the overlay record
length is larger, bytes in the extended spliced record that are not overlaid are filled
in with blanks.
The resulting spliced output record with either VLENMAX or VLENOVLY would
be:
35
ON1 BASE1 WITH1 BASE2
WITH2
You can use VLENMAX when you want the spliced record to have the maximum
length of the base or overlay record. You can use VLENOVLY when you want the
spliced record to have the length of the overlay record, regardless of whether it's
longer or shorter than the base record. Without VLENMAX or VLENOVLY, the
spliced record has the length of the base record regardless of whether it's longer or
shorter than the overlay record.
For fixed-length records, the length of the base, overlay and spliced records are all
the same. Thus, VLENOVLY and VLENMAX have no meaning for fixed-length
records and are ignored.
WITHALL can be used to create multiple spliced records for each set of duplicates.
The first duplicate is spliced with the specified fields from the second duplicate.
Then the first duplicate is spliced with the specified fields from the third duplicate,
and so on.
The first duplicate is treated as a base record. Each subsequent duplicate is treated
as an overlay record. The specified fields from each overlay record are overlaid on
to the base record. Thus, the output records consist of fields from the base record
intermixed with specified nonblank and blank fields from the overlay records.
The records to be spliced can originate from multiple input data sets.
To illustrate the splicing process when WITHALL is specified, if we had the
following four fixed-length records with the base fields, ON field and WITH fields
as shown:
BASE1
ON1
ON1
ON1
ON1
BASE2
BASE3 BASE4
WITHA
WITHC
WITHE
GGGGG
WITHB
WITHF
The resulting three spliced output records would be:

BASE1
BASE1
BASE1
ON1
ON1
ON1
BASE2
BASE2
BASE2
WITHA
WITHC
WITHE
BASE3 BASE4
BASE3 BASE4
BASE3 BASE4
WITHB
WITHF
683
SPLICE Operator
Note that without WITHALL, the resulting single spliced output record would be:
BASE1
ON1
BASE2
WITHE
BASE3 BASE4
WITHF
For variable-length records, by default (without VLENMAX or VLENOVLY), the

spliced record has the same length as the base record, and WITH fields from the
overlay record that are beyond the end of the base record do not appear in the
spliced record. For example, with WITHALL, if we had the following four records
with the lengths (in the RDW), ON field and WITH fields as shown:
30
25
50
40
| BASE1 ON1
|
ON1
|
ON1
|
ON1
BASE2
WITHA
WITHB
WITHE
WITHC
WITHF
WITHD
the resulting three spliced output records would be:

30 | BASE1 ON1
30 | BASE1 ON1
30 | BASE1 ON1
WITHA BASE2
WITHB BASE2
WITHE BASE2
The WITHC, WITHD and WITHF fields are beyond the end of the base record, so
they are not spliced.
However, if you specify VLENMAX, the spliced record is given the larger of the
base record length or overlay record length. If we specify VLENOVLY, the spliced
record is given the overlay record length. In either case, if the overlay record
length is larger, bytes in the extended spliced record that are not overlaid are filled
in with blanks. The resulting three spliced output records with WITHALL and
VLENMAX would be:
30 | BASE1 ON1
50 | BASE1 ON1
40 | BASE1 ON1
WITHA BASE2
WITHB BASE2
WITHE BASE2
WITHC
WITHF
WITHD
The resulting three spliced output records with WITHALL and VLENOVLY would
be:
25 | BASE1 ON1
50 | BASE1 ON1
40 | BASE1 ON1
WITHA
WITHB BASE2
WITHE BASE2
WITHC
WITHF
WITHD
WITHANY can be used to create one spliced record for each set of duplicates. The
first duplicate is spliced with the nonblank values of each subsequent duplicate for
specified fields.
as an overlay record. Each specified field with a nonblank value in each overlay
record is overlaid on to the base record. Thus, the output record consists of fields
from the base record intermixed with specified nonblank fields from each overlay
record. The value from the last overlay record with each nonblank value will
appear in the output record. Note that a specified "field" from an overlay record
can actually consist of multiple fields from the record that have previously been
reformatted into one contiguous field.
The records to be spliced can originate from multiple input data sets.
To illustrate the splicing process when WITHANY is specified, if we had the
following four fixed-length records with the base fields, ON field and WITH fields
as shown:
684
SPLICE Operator
BASE1
ON1
ON1
ON1
ON1
BASE2
WITHC
WITHA
WITHB
The resulting spliced output record would be:

BASE1
ON1
BASE2
WITHB
WITHC
WITHA
For variable-length records, by default (without VLENMAX), the spliced record has
the same length as the base record. For example, with WITHANY, if we had the
following four records with the lengths (in the RDW), ON field and WITH fields as
shown:
30
50
25
40
| BASE1 ON1
|
ON1
|
ON1
|
ON1
BASE2
WITHB
WITHA
WITHC
the resulting spliced output records would be:

30 | BASE1 ON1
WITHA BASE2
The WITHB and WITHC fields are beyond the end of the base record, so they are
not spliced. However, if you specify VLENMAX, the spliced record is given the
largest of the base record length or overlay record lengths. If the largest overlay
record length is larger than the base record length, bytes in the extended spliced
record that are not overlaid are filled in with blanks. The resulting spliced output
record with WITHANY and VLENMAX would be:
50 | BASE1 ON1
WITHA BASE2
WITHC
WITHB
VLENOVLY cannot be specified with WITHANY.

WITHEACH can be used to create one spliced record for each set of duplicates.
The first duplicate is spliced with one specified field from each subsequent
duplicate.
as an overlay record. The specified blank or nonblank field from each overlay
record is overlaid on to the base record. Thus, the output record consists of fields
from the base record intermixed with a specified nonblank or blank field from each
overlay record. Note that the specified "field" from an overlay record can actually
consist of multiple fields from the record that have previously been reformatted
into one contiguous field.
The records to be spliced can originate from multiple input data sets
To illustrate the splicing process when WITHEACH is specified, if we had the
following four records with the base fields, ON field and WITH fields as shown:
BASE1
ON1
ON1
ON1
ON1
BASE2
WITHA
WITHB
WITHC
The resulting spliced output record would be:

BASE1
ON1
BASE2
WITHA
WITHB
WITHC
685
SPLICE Operator
For variable-length records, by default (without VLENMAX), the spliced record has
the same length as the base record. For example, with WITHEACH, if we had the
following four records with the lengths (in the RDW), ON field and WITH fields as
shown:
30
25
50
40
| BASE1 ON1
|
ON1
|
ON1
|
ON1
BASE2
WITHA
WITHB
WITHC
the resulting spliced output records would be:

30 | BASE1 ON1
WITHA BASE2
The WITHB and WITHC fields are beyond the end of the base record, so they are
not spliced.
However, if you specify VLENMAX, the spliced record is given the largest of the
base record length or overlay record lengths. If the largest overlay record length is
larger than the base record length, bytes in the extended spliced record that are not
overlaid are filled in with blanks. The resulting spliced output record with
WITHEACH and VLENMAX would be:
50 | BASE1 ON1
WITHA BASE2
WITHC
WITHB
VLENOVLY cannot be specified with WITHEACH.

KEEPNODUPS can be used to keep the non-duplicate records as well as the
spliced records. The non-duplicate records will be unchanged.
To illustrate the splicing process when KEEPNODUPS is specified, if we had the
following six records with the base fields, ON fields and WITH fields as shown:
UNIQA
BASEA
DUPAA
UNIQB
BASEB
DUPBB
ONA
ONB
ONB
ONC
OND
OND
WITHA
WITHB
The two unique records (ONA and OND) would be kept along with the two
spliced records (ONB and OND). The resulting four unspliced and spliced output
records would be:
UNIQA
BASEA
UNIQB
BASEB
ONA
ONB
ONC
OND
WITHA
WITHB
Note that without KEEPNODUPS, the two unique records (ONA and ONC) would
not be kept. The resulting two spliced output records would be:
BASEA
BASEB
ONB
OND
WITHA
WITHB
KEEPBASE can be used to keep the base records (first duplicate) as well as the
spliced records. The base records will be unchanged.
To illustrate the splicing process when KEEPBASE is specified, if we had the
following six records with the base fields, ON fields and WITH fields as shown:
686
SPLICE Operator
UNIQA
BASEA
DUPAA
UNIQB
BASEB
DUPBB
ONA
ONB
ONB
ONC
OND
OND
WITHA
WITHB
The two base records with duplicates (first ONB record and first OND record)
would be kept along with the two spliced records (ONB and OND). The resulting
four unspliced and spliced output records would be:
BASEA
BASEA
BASEB
BASEB
ONB
ONB
OND
OND
WITHA
WITHB
Note that without KEEPBASE, the two base records with duplicates (first ONB
record and first OND record) would not be kept. The resulting two spliced output
records would be:
BASEA
BASEB
ONB
OND
WITHA
WITHB
If we used KEEPNODUPS and KEEPBASE with the original six records, the
resulting six unspliced and spliced output records would be:
UNIQA
BASEA
BASEA
UNIQB
BASEB
BASEB
ONA
ONB
ONB
ONC
OND
OND
WITHA
WITHB
DFSORT is called to sort the indd data set. ICETOOL uses its E35 exit to determine
which records to splice and include in the outdd data set. ICETOOL passes the
EQUALS option to DFSORT to ensure that duplicates are kept in their original
input order.
Do not supply your own MODS, SUM, OUTREC, or SORT statement.
You can use comment statements. You can use INCLUDE, OMIT, INREC, OPTION,
and OUTFIL statements providing you observe these rules:
v You can use an INCLUDE or OMIT statement to remove input records before
SPLICE processing.
v You can use an INREC statement to reformat input records before SPLICE
processing; the base and overlay records are reformatted according to the INREC
statement. You can use INREC's PARSE, BUILD (FIELDS), OVERLAY, FINDREP,
IFTHEN, or IFOUTLEN functions. If your INREC statement changes the starting
position of an ON field or WITH field, you must specify the new starting
position for that ON field or WITH field. For example, if your input records
have a CH key at positions 1-5 and a WITH field at 6-8 and you use an INREC
INREC FIELDS=(31:1,50)
you must specify ON(31,5,CH) instead of ON(1,5,CH) and WITH(36,3) instead of

WITH(6,3).
v You can further process the outdd records associated with TO(outdd) after
SPLICE processing using an OUTFIL statement like this:
687
SPLICE Operator

...
For example, with TO(OUT1) you could further modify the OUT1 records after
they have been spliced, with a statement like this:
OUTFIL FNAMES=OUT1,FTOV,VLTRIM=X40

inappropriate for a SPLICE operator, you can specify USING(xxxx) and take one of
the following actions:

FROM(indd)
on page 575.
TO(outdd)
records it produces for the operation (that is, the spliced records, the
non-duplicate records if KEEPNODUPS is specified, and the base records if
KEEPBASE is specified).
specified in the FROM operand.
ON(p,m,f)
See the discussion of this operand on the SELECT statement in SELECT
WITH(p,m)
Specifies the position and length of a field to be overlaid from the overlay
record on to the base record.
p specifies the first byte of the field relative to the beginning of the overlay
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3 4
| p=
1
2
3
4
5
6
7
8
688
SPLICE Operator
If INREC is specified, p must refer to the record as reformatted by INREC.
position 32752.
A WITH field will not be used to overlay the RDW of a variable-length base
record or to overlay bytes from beyond the end of an overlay record on to a
base record. When necessary, WITH fields will be adjusted to prevent these
situations. For example, if WITH(1,6) is specified for a variable-length record, it
will be treated as WITH(5,2) and if WITH(75,10) is specified for an 80-byte
overlay record, it will be treated as WITH(75,6).
A WITH field will not be used to overlay bytes beyond the end of a base
record. When necessary, WITH fields will be adjusted to prevent this situation.
For example, if WITH(75,10) is specified for an 80-byte base record, it will be
treated as WITH(75,6). However, if you specify VLENMAX or VLENOVLY, a
WITH field can be used to overlay bytes beyond the end of a base record
provided that WITH field is present in the overlay record. For example, if
VLENMAX and WITH(75,10) is specified for an 80-byte base record and a
90-byte overlay record, the spliced record will have a length of 90 bytes and
the WITH(75,10) field will be present at positions 75-84 followed by 6 blanks in
positions 85-90.
WITHALL
Specifies that the first duplicate is spliced with specified fields from the second
duplicate, and then from each subsequent duplicate in turn. All of the WITH
fields (nonblank and blank) from each overlay record are overlaid on to the
base record.
With WITHALL, a spliced output record is created from each base record and
overlay record, resulting in n-1 spliced records for each set of n duplicates.
WITHALL overrides the default of splicing the first duplicate with all of the
specified fields from the last duplicate.
WITHANY
Specifies that the first duplicate is spliced with specified nonblank fields from
each subsequent duplicate. Each nonblank WITH field from each overlay
record is overlaid on to the base record.
With WITHANY, a single spliced output record is created using the base record
and each nonblank field from each overlay record. If more than one overlay
record has a nonblank value for a WITH field, the nonblank value from the
last overlay record for that WITH field will appear in the output record. Note
that the specified "field" from an overlay record can actually consist of multiple
fields from the record that have previously been reformatted into one
contiguous field.
WITHANY overrides the default of splicing the first duplicate with all of the
VLENOVLY cannot be specified with WITHANY.
WITHEACH
Specifies that the first duplicate is spliced with one specified field from each
subsequent duplicate. One WITH field (nonblank or blank) from each overlay
record is overlaid on to the base record. The first WITH field specifies the
bytes to be overlaid from the second duplicate record on to the first duplicate
record. The second WITH field specifies the bytes to be overlaid from the third
duplicate record on to the first duplicate record, and so on. For any set of
689
SPLICE Operator
duplicates, extra overlay records without matching WITH fields, or extra
WITH fields without matching overlay records are ignored.
With WITHEACH, a single spliced output record is created using the base
record and one field from each overlay record. Note that the specified "field"
from an overlay record can actually consist of multiple fields from the record
that have previously been reformatted into one contiguous field.
WITHEACH overrides the default of splicing the first duplicate with all of the
VLENOVLY cannot be specified with WITHEACH.
KEEPNODUPS
Specifies that non-duplicate records are to be kept as well as spliced records.
The non-duplicate records will be unchanged.
KEEPBASE
Specifies that base records (first duplicate) are to be kept as well as spliced
records. The base records will be unchanged.
VLENMAX
Specifies that for variable-length records, the length of the spliced record is set
to the maximum length of the base record and overlay record. VLENMAX
overrides the default of setting the length of the spliced record to the length of
the base record.
If VLENMAX is specified with or without WITHALL, the spliced record is
given the larger of the base record length or overlay record length. If the
overlay record length is larger than the base record length, bytes in the
extended spliced record that are not overlaid are filled in with blanks.
If VLENMAX is specified with WITHANY or WITHEACH, the spliced record
is given the largest of the base record length or overlay record lengths. If the
largest overlay record length is larger than the base record length, bytes in the
extended spliced record that are not overlaid are filled in with blanks.
For fixed-length records, VLENMAX is ignored since the base, overlay and
spliced records all have the same length.
VLENOVLY
Specifies that for variable-length records, the length of the spliced record is set
to the length of the overlay record. VLENOVLY overrides the default of setting
the length of the spliced record to the length of the base record.
If VLENOVLY is specified with or without WITHALL, the spliced record is
given the overlay record length. If the overlay record length is larger than the
base record length, the spliced record is padded with blanks from the end of
the base record to the new length. If the overlay record length is smaller than
the base record length, bytes in the extended spliced record that are not
overlaid are filled in with blanks.
VLENOVLY cannot be specified with WITHANY or WITHEACH.
For fixed-length records, VLENOVLY is ignored since the base, overlay and
spliced records all have the same length.
USING(xxxx)
690
SPLICE Operator
2. Should generally be used only for an INCLUDE or OMIT statement, an
INREC statement, comment statements, or appropriate OUTFIL statements
as described for SPLICE operator on page 681.
VSAMTYPE(x)
on page 575.
UZERO
See the discussion of this operand on the OCCUR statement in OCCUR
SPLICE examples
SPLICE normally requires reformatting the records of two or more data sets so
they can be joined, so complete JCL examples are shown in this section to illustrate
the suggested techniques. These techniques and others can be employed with
SPLICE to perform a variety of tasks.
Because SPLICE overlays the WITH fields from the overlay record to the base
record using matching ON fields, it's usually necessary to do some initial setup
before using SPLICE, to ensure that:
v the ON fields are in the same positions in the base and overlay records
v the WITH fields in the overlay records are in the positions they will occupy in
the base records
v the base records and overlay records are the same length. This is always
required for fixed-length records, and is required for variable-length records
unless VLENMAX or VLENOVLY is specified.
For optimum efficiency, it is also a good idea to remove any records that are not
needed for the SPLICE operation as part of the initial setup before the SPLICE
operation, by using appropriate INCLUDE or OMIT statements.
Example 1 - Create one spliced record for each match in two

files
This example shows how you can splice data together for each pair of records with
the same ON field in two different input data sets.
//S1
EXEC PGM=ICETOOL
//TOOLMSG
DD SYSOUT=*
//DFSMSG
DD SYSOUT=*
//IN1 DD *
Y12 89503 MKT
Y12 57301 MKT
Z35 02316 DEV
Y12 91073 MKT
Z35 18693 DEV
/*
//IN2 DD *
89503 27M $9,185,354 SAN JOSE
CA
72135 08M
$317,632 BOSTON
MA
18693 10M $8,732,105 BUFFALO
NY
57301 50M
$30,000 NEWARK
NJ
/*
//TEMP1 DD DSN=&&TEMP1,DISP=(MOD,PASS),SPACE=(TRK,(5,5)),UNIT=SYSDA
//COMBINE DD SYSOUT=*
//TOOLIN
DD *
691
SPLICE Operator
* Reformat the File1 records for splicing
COPY FROM(IN1) TO(TEMP1) USING(CTL1)
* Reformat the File2 records for splicing
COPY FROM(IN2) TO(TEMP1) USING(CTL2)
* Splice the needed data from File1 and File2 together
SPLICE FROM(TEMP1) TO(COMBINE) ON(5,5,ZD) WITH(15,17)
/*
//CTL1CNTL DD *
file1 data
31:X)
add blanks for spliced file2 data
/*
//CTL2CNTL DD *
OUTREC FIELDS=(5:1,5,
put file2 key in same place as file1 key
15:7,15,
file2 data
30:33,2)
file2 data
/*
The base records originate from the IN1 data set and are copied and reformatted to
the TEMP1 data set. The reformatted TEMP1 records are 31 bytes long and look
like this:
Y12
Y12
Z35
Y12
Z35
89503
57301
02316
91073
18693
MKT
MKT
DEV
MKT
DEV
The overlay records originate from the IN2 data set and are copied and
reformatted to the end (MOD) of the TEMP1 data set. The reformatted TEMP1
records are 31 bytes long and look like this:
89503
72135
18693
57301
27M $9,185,354 CA
08M
$317,632 MA
10M $8,732,105 NY
50M
$30,000 NJ
Note that MOD is used for the TEMP1 data set, so the reformatted records from
IN1 and IN2 will be output to the TEMP1 data set in that order, ensuring that they
are spliced in that order.
The base and overlay records from the TEMP1 data set are sorted and spliced to
the COMBINE data set.
The records look like this after they are sorted on the 5,5,ZD field, but before they
are spliced. As a visual aid, the WITH fields in the overlay records are shown in
bold.
Z35 02316 DEV
Z35 18693 DEV
18693
Y12 57301 MKT
57301
72135
Y12 89503 MKT
89503
Y12 91073 MKT
10M $8,732,105 NY
50M
08M
$30,000 NJ
$317,632 MA
27M $9,185,354 CA
The spliced COMBINE records are 31 bytes long and look like this:
Z35 18693 DEV 10M $8,732,105 NY
Y12 57301 MKT 50M
$30,000 NJ
Y12 89503 MKT 27M $9,185,354 CA
Note that the base records for 18693, 57301 and 89503 have been spliced together
with their respective overlay records.
692
SPLICE Operator
Here is what the various ICETOOL operators do in this job:
v The first COPY operator creates reformatted IN1 records in TEMP1. The second
COPY operator creates reformatted IN2 records in TEMP1. The reformatted IN1
records have blanks where the reformatted IN2 WITH fields will go. The
reformatted IN2 records have the ON field from IN2 in the same place as in the
reformatted IN1 records, and have the IN2 data where we want it to go in the
reformatted IN1 records. We made the reformatted IN1 and reformatted IN2
records the same size so we can put them all in the TEMP1 data set and use
TEMP1 as input to the SPLICE operator.
v The SPLICE operator sorts the records from TEMP1 using the ON field. TEMP1
has the reformatted IN1 records before the reformatted IN2 records. The spliced
records are created from the base records and the overlay records in TEMP1.
Whenever two records are found with the same ON field, the WITH field from
the second record (reformatted IN2 overlay record) is overlaid on to the first
record (reformatted IN1 base record). The resulting spliced records are written to
the COMBINE data set.
Example 2 - Combine complete records from four files

This example shows how you can use the WITHEACH operand to splice complete
records from four different data sets together.
//S2
EXEC PGM=ICETOOL
//TOOLMSG
DD SYSOUT=*
//DFSMSG
DD SYSOUT=*
//FILE1 DD DSN=... input file1 - 300-byte records
//** BE SURE TO USE MOD FOR T1
//T1 DD DSN=&&TX,UNIT=SYSDA,SPACE=(CYL,(5,5)),
// DISP=(MOD,PASS)
//ALLRCDS DD DSN=... output file - 870-byte records
//TOOLIN DD *
* Reformat the File1 records for splicing on added sequence
COPY FROM(FILE1) TO(T1) USING(CTL1)
* Splice record-by-record on added sequence number
SPLICE FROM(T1) TO(ALLRCDS) ON(871,8,PD) WITHEACH WITH(301,400) WITH(701,150) WITH(851,20) USING(CTL5)
/*
//CTL1CNTL DD *
* Reformat records to:
* 1
301
701
851
871
* File1bytes|400 blanks|150 blanks|20 blanks |seqno
OUTREC FIELDS=(1,300,871:SEQNUM,8,PD)
/*
//CTL2CNTL DD *
* 1
301
701
851
871
* 300 blanks|File2bytes|150 blanks|20 blanks |seqno
OUTREC FIELDS=(301:1,400,871:SEQNUM,8,PD)
/*
//CTL3CNTL DD *
* 1
301
701
851
871
* 300 blanks|400 blanks|File3bytes|20 blanks |seqno
number
number
number
number
693
SPLICE Operator
/*
//CTL4CNTL DD *
* 1
301
701
851
871
* 300 blanks|400 blanks|150 blanks|File4bytes|seqno
/*
//CTL5CNTL DD *
* Remove added sequence number from spliced records to get:
* File1bytes|File2bytes|File3bytes|File4bytes
OUTFIL FNAMES=ALLRCDS,OUTREC=(1,870)
/*
Because the data sets do not have a common key, we add sequence numbers to the
records from each data set and use the sequence numbers as the ON field for
SPLICE. Using this technique, we can splice together the 300-byte records from
FILE1, the 400-byte records from FILE2, the 150-byte records from FILE3 and the
20-byte records from FILE4, to produce 870-byte records in ALLRCDS.
Conceptually, the 870-byte records in ALLRCDS would look like this:
File1 Record1 ... File2Record1 ... File3Record1 ... File4Record1 ...
File1 Record2 ... File2Record2 ... File3Record2 ... File4Record2 ...
...
The base records originate from the FILE1 data set and the overlay records
originate from the FILE2, FILE3 and FILE4 data sets.
Here is what the various ICETOOL operators do in this job:
The first COPY operator creates reformatted records in the T1 data set with the
FILE1 records in positions 1-300, blanks in all other positions up to 870, and a
sequence number in positions 871-878. The sequence number will be 1 for the first
FILE1 record, 2 for the second FILE1 record, and so on.
The second COPY operator creates reformatted records in the T1 data set with the
The third COPY operator creates reformatted records in the T1 data set with the
The fourth COPY operator creates reformatted records in the T1 data set with the
Note that MOD is used for the T1 data set, so the reformatted records from FILE1,
FILE2, FILE3 and FILE4 will be output in that order in T1, ensuring that they are
sorted and spliced in that order.
The SPLICE operator sorts the records from T1 using the sequence number as the
ON field. With WITHEACH, the reformatted FILE1 records are treated as the base
records, and the reformatted FILE2, FILE3 and FILE4 records are treated as the
overlay records; each WITH field is associated with an overlay record in turn. So
the first WITH field specifies the bytes to be used from the second duplicate
694
SPLICE Operator
(FILE2 record), the second WITH field specifies the bytes to be used from the third
duplicate (FILE3 record) and the third WITH field specifies the bytes to be used
from the fourth duplicate (FILE4 record).
SPLICE matches each base and overlay record by their sequence numbers, and
creates a new combined 878-byte record. The OUTFIL statement in CTL5CNTL is
used to remove the sequence number so that the 870-byte spliced record is written
to the ALLRCDS data set.
Example 3 - Create files with matching and non-matching

records
This example shows how you can match records in input data sets 1 and 2 to
produce three output data sets with:
v ON fields that appear in both input data set 1 and input data set 2
v ON fields that appear only in input data set 1
v ON fields that appear only in input data set 2
//S3 EXEC PGM=ICETOOL
//IN1
DD *
Vicky
Frank
Carrie
Holly
Paul
/*
//IN2
DD *
Karen
Holly
Carrie
Vicky
Mary
/*
//OUT12
DD SYSOUT=*
//OUT1
DD SYSOUT=*
//OUT2
DD SYSOUT=*
//T1
DD DSN=&&T1,DISP=(MOD,PASS),UNIT=SYSDA,SPACE=(TRK,(5,5))
//TOOLIN
DD *
* Add 11 identifier for FILE1 records.
COPY FROM(IN1) TO(T1) USING(CTL1)
* Add 22 identifier for FILE2 records.
COPY FROM(IN2) TO(T1) USING(CTL2)
* SPLICE to match up records and write them to their
* appropriate output files.
SPLICE FROM(T1) TO(OUT12) ON(1,10,CH) WITH(13,1) USING(CTL3) KEEPNODUPS
/*
//CTL1CNTL DD *
* Mark FILE1 records with 11
OUTREC FIELDS=(1,10,12:C11)
/*
//CTL2CNTL DD *
* Mark FILE2 records with 22
OUTREC FIELDS=(1,10,12:C22)
/*
//CTL3CNTL DD *
* Write matching records to OUT12 file. Remove id.
OUTFIL FNAMES=OUT12,INCLUDE=(12,2,CH,EQ,C12),OUTREC=(1,10)
* Write FILE1 only records to OUT1 file. Remove id.
* Write FILE2 only records to OUT2 file. Remove id.
/*
695
SPLICE Operator
We copy the IN1 records to the T1 data set and add an identifier of '11' to show
they come from FILE1.
We copy the IN2 records to the end (MOD) of the T1 data set and add an identifier
of '22' to show they come from FILE2.
We sort the records of T1 on positions 1-3 and splice the second id byte for
matching records. We use KEEPNODUPS to keep non-duplicate records.
The records look like this after they are sorted, but before they are spliced:
Carrie
Carrie
Frank
Holly
Holly
Karen
Mary
Paul
Vicky
Vicky
11
22
11
11
22
22
22
11
11
22
The records look like this after they are spliced, but before we do the OUTFIL
processing specified by CTL3CNTL with USING(CTL3) for SPLICE:
Carrie
Frank
Holly
Karen
Mary
Paul
Vicky
12
11
12
22
22
11
12
An id of 12 indicates an ON field that appears in IN1 and IN2. An id of 11

indicates an ON field that appears only in IN1. An id of 22 indicates an ON field
that appears only in IN2.
The OUTFIL statements in CTL3CNTL write the records to their appropriate
output data sets (without the ids) as follows:
OUT12 contains:
Carrie
Holly
Vicky
OUT1 contains:
Frank
Paul
OUT2 contains:
Karen
Mary
Example 4 - Create a matrix of values

This example shows how you can use the WITHANY operand to create a matrix of
values. We combine multiple rows for different types of data with a common key
into a single row of data for that key, even if some rows are missing.
//IN DD *
696
SPLICE Operator
001 3 150
001 2 120
001 1 100
002 2 140
002 3 250
003 1 050
003 3 920
004 3 005
/*
//OUT DD SYSOUT=*
//TOOLIN DD *
* Splice nonblank type 1 values (in 5-7), type 2 values (9-11)
* and type 3 values (13-15) for each key.
SPLICE FROM(IN) TO(OUT) ON(1,3,CH) WITHANY KEEPNODUPS WITH(5,3) WITH(9,3) WITH(13,3) USING(CTL1)
/*
//CTL1CNTL DD *
* Before SPLICE:
* Reformat type 1 records to:
* 1 5
* key val
INREC IFTHEN=(WHEN=(5,1,CH,EQ,C1),
BUILD=(1,3,5:8,3)),
* 1
9
* key
val
BUILD=(1,3,9:8,3)),
* 1
13
* key
val
BUILD=(1,3,13:8,3))
/*

001
001
001
002
002
003
003
004
3
2
1
2
3
1
3
3
150
120
100
140
250
050
920
005
We have a key (for example, '001') in positions 1-3, a record type (1, 2 or 3) in
position 5 and a numeric value in positions 8-10. We want to set up a single row
for each key with the values for the three record types and blanks for missing
record types.
Before we SPLICE the records, we use the INREC IFTHEN clauses to put the
values from the type 1 records in positions 5-7, the values from the type 2 records
in positions 9-11 and the values from the type 3 records in positions 13-15. The
reformatted INREC records look like this:
001
150
001
120
001 100
002
140
002
250
003 050
003
920
004
005
697
SPLICE Operator
We use SPLICE with WITHANY and appropriate WITH fields to create one
combined record for each key with the values for the record types. We use
KEEPNODUPS to keep records for keys with only one value (for example, '004').
The OUT records are 15 bytes long and look like this:
001 100 120 150
002
140 250
003 050
920
004
005
Note that if we used WITHEACH instead of WITHANY, the output would not be
what we wanted since record types are missing for some keys (for example, type 1
is missing for key 002), and record types are out of order for some keys (for
example, we have type 3, type 2 and type 1 in that order for key 001). With
WITHEACH, the OUT records would look like this:
001
002
003
004
150
140
005
Example 5 - Create multiple spliced records for each match in

two types of records
This example shows how you can use the WITHALL operand to tell ICETOOL to
splice data together for a single record of one type (A records) and multiple
records of another type (H records), in the same input data set, that all have the
same ON field (duplicate records). It also shows how to ensure that duplicates of
the second type without a match of the first type are not written to the output data
set. IFTHEN clauses are used in an INREC statement to reformat the two types of
records appropriately before they are sorted and spliced
//S5
EXEC PGM=ICETOOL
//TOOLMSG
DD SYSOUT=*
//DFSMSG
DD SYSOUT=*
//MAST DD *
A0000B0000KRSC0000D000000E0000F00G000
A1111B1111FLYC1111D111111E1111F11G111
H02KRSI000002J002K002L02
H03FLYI000003J003K003L03
H04VQXI000004J004K004L04
H05FLYI000005J005K005L05
H06KHNI000006J006K006L06
H07KRSI000007J007K007L07
H08FLYI000008J008K008L08
H09KHNI000009J009K009L09
/*
//OUT DD SYSOUT=*
//TOOLIN
DD *
* Splice needed base and overlay data together.
* Do NOT splice identifier.
SPLICE FROM(MAST) TO(OUT) WITH(1,7) WITH(13,4) ON(20,3,CH) WITH(23,3) WITH(26,3) WITHALL USING(CTL1)
/*
//CTL1CNTL DD *
* Before SPLICE:
* Set up fields in base (A) records. Add B id in position 33.
INREC IFTHEN=(WHEN=(1,1,CH,EQ,CA),
BUILD=(8:14,5,17:31,3,20:11,3,29:34,4,33:CB)),
* Set up fields in overlay (H) records. Add V id in position 33.
IFTHEN=(WHEN=(1,1,CH,EQ,CH),
BUILD=(1:7,7,13:18,4,20:4,3,23:1,3,26:22,3,33:CV))
* After SPLICE:
698
SPLICE Operator
* Remove duplicate overlay records without matching base record.
* Remove base or overlay indicator.
OUTFIL FNAMES=OUT,OMIT=(33,1,CH,EQ,CV),OUTREC=(1,32)
/*
The base records are the records in MAST with an 'A' in column 1. They are
reformatted by the INREC statement as 33 byte records that look like this:
C0000
C1111
F00KRS
F11FLY
G000B
G111B
We put a 'B' in position 33 to identify these records as base records.

The overlay records are the records in MAST with an 'H' in column 1. They are
reformatted by the INREC statement as 33 bytes records that look like this:
I000002
I000003
I000004
I000005
I000006
I000007
I000008
I000009
K002
K003
K004
K005
K006
K007
K008
K009
KRSH02L02
FLYH03L03
VQXH04L04
FLYH05L05
KHNH06L06
KRSH07L07
FLYH08L08
KHNH09L09
V
V
V
V
V
V
V
V
We put a 'V' in position 33 to identify these records as overlay records.

The base and overlay records set up by the INREC statement are sorted and
spliced.
The records look like this after they are sorted on the 20,3,CH field, but before
they are spliced. As a visual aid, the WITH fields in the overlay records are shown
in bold.
C1111
F11FLY
G111B
I000003
K003
FLYH03L03
V
I000005
K005
FLYH05L05
V
I000008
K008
FLYH08L08
V
I000006
K006 KHNH06L06
V
I000009
K009
KHNH09L09
V
C0000
F00KRS
G000B
I000002
K002
KRSH02L02
V
I000007
K007
KRSH07L07
V
I000004
K004 VQXH04L04
V
The spliced output records are 33 bytes long and look like this:
I000003C1111K003F11FLYH03L03G111B
I000005C1111K005F11FLYH05L05G111B
I000008C1111K008F11FLYH08L08G111B
I000009
K009
KHNH09L09
V
I000002C0000K002F00KRSH02L02G000B
I000007C0000K007F00KRSH07L07G000B
Note that the base record (type A) for FLY has been spliced together with each of
the three overlay records (type H) for FLY. Likewise, the base record (type A) for
KRS has been spliced together with each of the two overlay records (type H) for
KRS.
But also note that the overlay records (type H) for KHN have been spliced
together. Because KHN does not appear as a base record (type A) we don't want
the KHN records to appear in the OUT data set. So we will use the 'V' we put in
position 33 for the overlay records to identify and delete spliced overlay records
without a matching base record. We only have to do this if we have duplicate
699
SPLICE Operator
overlay records without a matching base record. Single overlay records without a
matching base record will be deleted automatically (unless you specify
KEEPNODUPS).
After we eliminate the spliced overlay records and the position 33 indicator, the
OUT records are 32 bytes long and look like this:
I000003C1111K003F11FLYH03L03G111
I000005C1111K005F11FLYH05L05G111
I000008C1111K008F11FLYH08L08G111
I000002C0000K002F00KRSH02L02G000
I000007C0000K007F00KRSH07L07G000
Note that if we had not specified WITHALL, only the first and last records for
each set of duplicates would have been spliced, producing the following output:
I000008C1111K008F11FLYH08L08G111
I000007C0000K007F00KRSH07L07G000
Example 6 - Pull records from a master file in sorted order

This example shows how you can use the WITHALL operand to tell ICETOOL to
use ON fields in a "PULL" data set to select one or more records from a "MASTER"
data set. In other words, you can use a PULL list to select records from a MASTER
list. In this case, the PULL data set has VB records and the MASTER data set has
FB records. The ON field is a City Name that can be 1-20 bytes long, and the
selected MASTER records are to be sorted by the City Name. (Example 7 - Pull
records from a master file in their original order on page 702 shows how to keep
the MASTER records in their original order.)
//S6
EXEC PGM=ICETOOL
//TOOLMSG
DD SYSOUT=*
//DFSMSG
DD SYSOUT=*
//PULL DD DSN=VAR.PULL.FILE,DISP=SHR
//MASTER DD DSN=FIXED.MASTER.FILE,DISP=SHR
//TEMP1 DD DSN=&&T1,DISP=(MOD,PASS),SPACE=(TRK,(5,5)),UNIT=SYSDA
//OUT DD DSN=FIXED.OUTPUT.FILE,DISP=(NEW,CATLG,DELETE),
//
SPACE=(TRK,(5,5)),UNIT=SYSDA
//TOOLIN DD *
* Convert PULL records from VB to FB and add P identifier.
COPY FROM(PULL) USING(CTL1)
* Add M identifier to MASTER records.
COPY FROM(MASTER) TO(TEMP1) USING(CTL2)
* Splice PULL and MASTER records (do NOT splice identifier):
*
Spliced MASTER records with matching PULL records have P id.
*
Spliced MASTER records without matching PULL records
*
have M id.
* Eliminate records with M id.
SPLICE FROM(TEMP1) TO(OUT) ON(1,20,CH) WITHALL WITH(1,40) USING(CTL3)
/*
//CTL1CNTL DD *
OUTFIL FNAMES=TEMP1,VTOF,OUTREC=(5,20,41:CP)
/*
//CTL2CNTL DD *
* Add M identifier to MASTER records.
OUTREC FIELDS=(1,40,41:CM)
/*
//CTL3CNTL DD *
* Eliminate MASTER records without matching PULL records.
OUTFIL FNAMES=OUT,OMIT=(41,1,CH,EQ,CM),OUTREC=(1,40)
/*
700
SPLICE Operator
The base records originate from the PULL data set (VAR.PULL.FILE). The PULL
data set has variable-length (VB) records with the RDW in positions 1-4 and the
variable-length City Name starting in position 5 for 1-20 bytes. Conceptually, the
PULL records look like this:
Length
12
12
11
15
|
|
|
|
|
Data
SAN JOSE
NEW YORK
DENVER
LOS ANGELES
The overlay records originate from the MASTER data set (FIXED.MASTER.FILE).
The MASTER data set has 40-byte fixed-length (FB) records with the City Name in
positions 1-20.
The PULL records are copied and reformatted to the TEMP1 data set as 41-byte
fixed-length (FB) records with the City Name in positions 1-20 (padded on the
right with blanks as necessary), and a 'P' in position 41 to identify them as PULL
records. The VTOF and OUTREC parameters of DFSORT's OUTFIL statement are
used to convert the VB records to FB records with blank padding. The reformatted
PULL records in TEMP1 look like this:
SAN JOSE
NEW YORK
DENVER
LOS ANGELES
P
P
P
P
The MASTER records are copied and reformatted to the end (MOD) of the TEMP1
data set as 41-byte fixed-length (FB) records with an 'M' added in position 41 to
identify them as MASTER records. The reformatted MASTER records in TEMP1
look like this:
SAN JOSE
PHOENIX
LOS ANGELES
SAN JOSE
NEW YORK
SAN JOSE
TUCSON
NEW YORK
PHOENIX
8630 SUSAN
7993 PAUL
9203 MICHAEL
0052 VICKY
5218 CARRIE
3896 FRANK
1056 LISA
6385 MICHAEL
5831 HOLLY
M
M
M
M
M
M
M
M
M
The base and overlay records from the TEMP1 data set are sorted and spliced.
in bold.
DENVER
LOS ANGELES
LOS ANGELES
NEW YORK
NEW YORK
NEW YORK
PHOENIX
PHOENIX
SAN JOSE
SAN JOSE
SAN JOSE
SAN JOSE
TUCSON
9203
MICHAEL
5218
6385
7993
5831
CARRIE
MICHAEL
PAUL
HOLLY
8630 SUSAN
0052 VICKY
3896 FRANK
1056 LISA
P
P
M
P
M
M
M
M
P
M
M
M
M
The spliced records look like this:
701
SPLICE Operator
LOS ANGELES
NEW YORK
NEW YORK
PHOENIX
SAN JOSE
SAN JOSE
SAN JOSE
9203
5218
6385
5831
8630
0052
3896
MICHAEL
CARRIE
MICHAEL
HOLLY
SUSAN
VICKY
FRANK
P
P
P
M
P
P
P
Finally, we use the OUTFIL statement for SPLICE to remove each spliced record
with an 'M' in position 41, because that represents a base record without a
matching overlay record. The OUTFIL statement also removes the 'P' indicator in
position 41 from each record, because it is not needed in the OUT data set.
Thus, for each MASTER record that matches a PULL record, we've overlaid the
PULL record with the MASTER record. This effectively selects all of the MASTER
records on the PULL list. The resulting OUT data set (FIXED.OUTPUT.FILE) has
the following 40-byte fixed-length records:
LOS
NEW
NEW
SAN
SAN
SAN
ANGELES
YORK
YORK
JOSE
JOSE
JOSE
9203
5218
6385
8630
0052
3896
MICHAEL
CARRIE
MICHAEL
SUSAN
VICKY
FRANK
Example 7 - Pull records from a master file in their original order

This example is similar to Example 6 - Pull records from a master file in sorted
order on page 700, except that we want to keep the resulting MASTER records in
their original order instead of sorting them by the City Name field. We use
DFSORT's SEQNUM parameter to add a sequence number to each MASTER record
before the records are spliced, and we splice that sequence number along with the
data. After SPLICE sorts by the City Name, we SORT again by the sequence
number to get the resulting MASTER records back in their original order.
//S7
EXEC PGM=ICETOOL
//TOOLMSG
DD SYSOUT=*
//DFSMSG
DD SYSOUT=*
//PULL DD DSN=VAR.PULL.FILE,DISP=SHR
//MASTER DD DSN=FIXED.MASTER.FILE,DISP=SHR
//TEMP1 DD DSN=&&TEMP1,DISP=(MOD,PASS),SPACE=(TRK,(5,5)),UNIT=SYSDA
//TEMP2 DD DSN=&&TEMP2,DISP=(,PASS),SPACE=(TRK,(5,5)),UNIT=SYSDA
//OUT DD DSN=FIXED.OUTPUT.FILE,DISP=(NEW,CATLG,DELETE),
//
SPACE=(TRK,(5,5)),UNIT=SYSDA
//TOOLIN DD *
COPY FROM(PULL) USING(CTL1)
* Add sequence number and M identifier to MASTER records.
COPY FROM(MASTER) TO(TEMP1) USING(CTL2)
* Splice PULL and MASTER records (splice sequence number, but
* do NOT splice identifier):
*
Spliced MASTER records with matching PULL records have P id.
*
Spliced MASTER records without matching PULL records
*
have M id.
* Eliminate records with M id.
SPLICE FROM(TEMP1) TO(TEMP2) ON(1,20,CH) WITHALL WITH(1,48) USING(CTL3)
* Sort resulting spliced records on original sequence number
* to get them back in their original order.
* Remove id and sequence number.
SORT FROM(TEMP2) TO(OUT) USING(CTL4)
/*
//CTL1CNTL DD *
* Convert PULL records from VB to FB and add P identifier
OUTFIL FNAMES=TEMP1,VTOF,OUTREC=(5,20,49:CP)
/*
702
SPLICE Operator
//CTL2CNTL DD *
* Add sequence number and M identifier to MASTER records.
OUTREC FIELDS=(1,40,41:SEQNUM,8,BI,49:CM)
/*
//CTL3CNTL DD *
* Eliminate MASTER records without matching PULL records.
OUTFIL FNAMES=TEMP2,OMIT=(49,1,CH,EQ,CM)
/*
//CTL4CNTL DD *
* Sort on sequence number and remove id and sequence number.
OUTREC FIELDS=(1,40)
/*
The resulting OUT data set (FIXED.OUTPUT.FILE) has the following 40-byte
fixed-length records:
SAN
LOS
SAN
NEW
SAN
NEW
JOSE
ANGELES
JOSE
YORK
JOSE
YORK
8630
9203
0052
5218
3896
6385
SUSAN
MICHAEL
VICKY
CARRIE
FRANK
MICHAEL
Example 8 - Create a report showing if needed parts are on-hand

This example shows how you can use the KEEPNODUPS operand to tell
ICETOOL to compare the ON fields in a list of needed parts to the ON fields in a
list of on-hand parts, and produce a report showing if each needed part is on-hand
or not.
//S8
EXEC PGM=ICETOOL
//DFSMSG
DD SYSOUT=*
//ONHAND DD *
P62 Blue
P62 Red
G73 Blue
A27 Green
L90 Red
P63 Blue
/*
//NEEDED DD *
2003/05/07
A27 Green
2002/12/29
P62 Blue
2003/03/17
A27 Blue
2003/06/14
M92 Yellow
2002/12/18
L90 Red
/*
//COMBINED DD DSN=&&T1,UNIT=SYSDA,SPACE=(TRK,(5,5)),
// DISP=(MOD,PASS)
//RPT DD SYSOUT=*
//TOOLIN
DD *
* Reformat the ONHAND records for splicing.
* Add Yes for found in ONHAND data set.
* Add O to indicate ONHAND record.
COPY FROM(ONHAND) TO(COMBINED) USING(CTL1)
* Reformat the NEEDED records for splicing.
* Add No for missing from ONHAND data set.
* Add N to indicate NEEDED record.
COPY FROM(NEEDED) TO(COMBINED) USING(CTL2)
* Splice ONHAND and NEEDED records (splice identifier):
*
NEEDED records found in ONHAND list will have Yes
*
and N.
*
NEEDED records not found in ONHAND list will have No
*
and N.
*
ONHAND records that are not needed will have Yes
*
and O.
703
SPLICE Operator
* Eliminate records with O.
SPLICE FROM(COMBINED) TO(RPT) ON(1,12,CH) WITH(24,10) WITH(40,1) KEEPNODUPS USING(CTL3)
/*
//CTL1CNTL DD *
* Reformat ONHAND records with part in 1-12, Yes in 15-17
* and O in 40.
OUTREC FIELDS=(1:1,12,15:CYes,40:CO)
/*
//CTL2CNTL DD *
* Reformat NEEDED records with part in 1-12, No in 15-17,
* date in 24-33 and N in 40.
OUTREC FIELDS=(1:15,12,15:CNo,24:2,10,40:CN)
/*
//CTL3CNTL DD *
* Eliminate ONHAND parts that do not appear in NEEDED list.
* Create the report showing if needed parts are on-hand.
OUTFIL FNAMES=RPT,OMIT=(40,1,CH,EQ,CO),OUTREC=(1,33),
HEADER2=(1:Part,15:On-Hand,24:Needed by,/,
1:------------,15:-------,24:----------)
/*
The base records originate from the ONHAND data set and are copied and
reformatted to the COMBINED data set. We put an 'O' in position 40 to identify
these records as ONHAND records. The overlay records originate from the
NEEDED data set and are copied and reformatted to the COMBINED data set. We
put an 'N' in position 40 to identify these records as NEEDED records. Because
MOD is used for the COMBINED data set, it contains the reformatted ONHAND
records followed by the reformatted NEEDED records. The COMBINED records
are 40 bytes long and look like this:
P62
P62
G73
A27
L90
P63
A27
P62
A27
M92
L90
Blue
Red
Blue
Green
Red
Blue
Green
Blue
Blue
Yellow
Red
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
No
No
2003/05/07
2002/12/29
2003/03/17
2003/06/14
2002/12/18
O
O
O
O
O
O
N
N
N
N
N
The base and overlay records from the COMBINED data set are sorted and spliced.
in bold.
A27
A27
A27
G73
L90
L90
M92
P62
P62
P62
P63
Blue
Green
Green
Blue
Red
Red
Yellow
Blue
Blue
Red
Blue
No
Yes
No
Yes
Yes
No
No
Yes
No
Yes
Yes
2003/03/17
2003/05/07
2002/12/18
2003/06/14
2002/12/29
N
O
N
O
O
N
N
O
N
O
O
The spliced output records are 40 bytes long and look like this:
704
SPLICE Operator
A27
A27
G73
L90
M92
P62
P62
P63
Blue
Green
Blue
Red
Yellow
Blue
Red
Blue
No
Yes
Yes
Yes
No
Yes
Yes
Yes
2003/03/17
2003/05/07
2002/12/18
2003/06/14
2002/12/29
N
N
O
N
N
N
O
O
We have three types of records as follows:

1. Records with 'Yes and 'N' are NEEDED records with an ONHAND match that
have been spliced together. We want these for our report.
2. Records with 'No' and 'N' are NEEDED records without an ONHAND match
that have been kept because we used KEEPNODUPS. We want these for our
report.
3. Records with 'Yes' and 'O' are ONHAND records without a NEEDED match
that have been kept because we used KEEPNODUPS. We do not want these for
our report.
We use the OUTFIL statement for SPLICE to further process the spliced records. It
omits the 'O' records, removes the 'N' byte, and sets up the headers for the report.
The resulting RPT data set looks like this:
Part
-----------A27 Blue
A27 Green
L90 Red
M92 Yellow
P62 Blue
On-Hand
------No
Yes
Yes
No
Yes
Needed by
---------2003/03/17
2003/05/07
2002/12/18
2003/06/14
2002/12/29
Example 9 - Create a report showing if needed parts are on-hand

- advanced
This example is a more complex variation of Example 8 - Create a report showing
if needed parts are on-hand on page 703. It shows how you can use the
WITHALL, KEEPBASE, and KEEPNODUPS operands to tell ICETOOL to compare
the ON fields in a list of needed parts to the ON fields in a list of on-hand parts,
and produce a report showing if each needed part is on-hand or not. However, it
also has duplicate parts in the NEEDED data set, and produces a report with more
information from the ONHAND and NEEDED records.
//S9
EXEC PGM=ICETOOL
//DFSMSG
DD SYSOUT=*
//ONHAND DD *
P62 Blue
Dallas
G73 Blue
San Jose
A27 Green
Vancouver
/*
//NEEDED DD *
Rachel
A27 Green
Phoenix
Monica
P62 Blue
Phoenix
Phoebe
A27 Blue
Toronto
Chandler
M92 Yellow
Los Angeles
Joey
M92 Yellow
Paris
Ross
A27 Green
Paris
/*
//COMBINED DD DSN=&&C1,UNIT=SYSDA,SPACE=(TRK,(5,5)),
// DISP=(MOD,PASS)
//TEMP1 DD DSN=&&T1,UNIT=SYSDA,SPACE=(CYL,(5,5)),DISP=(,PASS)
//RPT DD SYSOUT=*
//TOOLIN
DD *
* Reformat the ONHAND records for splicing.
705
SPLICE Operator
* Add Yes for found and D for delete record.
COPY FROM(ONHAND) TO(COMBINED) USING(CTL1)
* Reformat the NEEDED records for splicing.
* Add No for missing and K for keep record.
COPY FROM(NEEDED) TO(COMBINED) USING(CTL2)
* Splice ONHAND and NEEDED records.
* Splice in Requested by, Ship to and id fields.
* Eliminate spliced records with D.
SPLICE FROM(COMBINED) TO(TEMP1) ON(1,12,CH) WITHALL KEEPBASE KEEPNODUPS USING(CTL3) WITH(24,10) WITH(53,13) WITH(66,1)
* Print report.
DISPLAY FROM(TEMP1) LIST(RPT) INDENT(2) BETWEEN(2) BLANK HEADER(Part) ON(1,12,CH) HEADER(On-Hand) ON(15,3,CH) HEADER(Requested by) ON(24,12,CH) HEADER(Ship from) ON(38,13,CH) HEADER(Ship to) ON(53,13,CH)
/*
//CTL1CNTL DD *
* Reformat ONHAND records with Part in 1-12, Yes for found in
* 15-17, From City in 38-50 and D in 66.
OUTREC FIELDS=(1:1,12,15:CYes,38:20,13,66:CD)
/*
//CTL2CNTL DD *
* Reformat NEEDED records with Part in 1-12, No for missing in
* 15-17, Requester Name in 24-35, n/a for From City in 38-40,
* To City in 53-65 and K in 66.
OUTREC FIELDS=(1:15,12,15:CNo ,24:2,10,38:Cn/a,
53:31,13,66:CK)
/*
//CTL3CNTL DD *
* Eliminate ONHAND parts that do not appear in NEEDED list.
OUTFIL FNAMES=TEMP1,OMIT=(66,1,CH,EQ,CD)
/*
The base records originate from the ONHAND data set. They are copied and
reformatted to the COMBINED data set. The reformatted records look like this:
P62 Blue
G73 Blue
A27 Green
Yes
Yes
Yes
Dallas
San Jose
Vancouver
D
D
D
The overlay records originate from the NEEDED data set and are copied and
reformatted to the COMBINED data set. The reformatted records look like this
A27
P62
A27
M92
M92
A27
Green
Blue
Blue
Yellow
Yellow
Green
No
No
No
No
No
No
Rachel
Monica
Phoebe
Chandler
Joey
Ross
n/a
n/a
n/a
n/a
n/a
n/a
Phoenix
Phoenix
Toronto
Los Angeles
Paris
Paris
K
K
K
K
K
K
The base and overlay records from the COMBINED data set are sorted and spliced.
However, we need to make sure that all parts which appear in more than one
NEEDED record, but do not appear in the ONHAND list, will appear in the
report. For example, we have two M92 Yellow parts in the NEEDED data set that
do not appear in the ONHAND data set. These two records are reformatted and
appear in the COMBINED data set as follows:
M92 Yellow
M92 Yellow
706
No
No
Chandler
Joey
n/a
n/a
Los Angeles
Paris
K
K
SPLICE Operator
ICETOOL would normally treat the first record as the base record and the second
record as the overlay record. As a result, these two records would be spliced
together into one record instead of two. To prevent this, and ensure that we keep
both M92 Yellow parts, we must specify KEEPBASE. As a result, two records are
kept: the unchanged first M92 Yellow record, and the spliced first and second M92
Yellow records (which in this case looks identical to the unspliced second record).
in bold.:
A27
A27
A27
A27
G73
M92
M92
P62
P62
Blue
Green
Green
Green
Blue
Yellow
Yellow
Blue
Blue
No
Yes
No
No
Yes
No
No
Yes
No
Phoebe
Rachel
Ross
Chandler
Joey
Monica
n/a
Vancouver
n/a
n/a
San Jose
n/a
n/a
Dallas
n/a
Toronto
Phoenix
Paris
Los Angeles
Paris
Phoenix
K
D
K
K
D
K
K
D
K
The spliced records look like this:

A27
A27
A27
A27
G73
M92
M92
P62
P62
Blue
Green
Green
Green
Blue
Yellow
Yellow
Blue
Blue
No
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Phoebe
Rachel
Ross
Chandler
Joey
Monica
n/a
Vancouver
Vancouver
Vancouver
San Jose
n/a
n/a
Dallas
Dallas
Toronto
Phoenix
Paris
Los Angeles
Paris
Phoenix
K
D
K
K
D
K
K
D
K
Records with 'D' are not needed, so we use the OUTFIL statement for SPLICE to
omit them. The TEMP1 records look like this:
A27
A27
A27
M92
M92
P62
Blue
Green
Green
Yellow
Yellow
Blue
No
Yes
Yes
No
No
Yes
Phoebe
Rachel
Ross
Chandler
Joey
Monica
n/a
Vancouver
Vancouver
n/a
n/a
Dallas
Toronto
Phoenix
Paris
Los Angeles
Paris
Phoenix
K
K
K
K
K
K
Although we could have used the OUTFIL statement for SPLICE to print the
report, we've chosen instead to use a separate DISPLAY operator. DISPLAY
requires an extra pass over the spliced records in TEMP1, but is easier to use than
OUTFIL for reports. The resulting RPT data set looks like this:
Part
-----------A27 Blue
A27 Green
A27 Green
M92 Yellow
M92 Yellow
P62 Blue
On-Hand Requested by Ship from

Ship to
------- ------------ ------------- ------------No
Phoebe
n/a
Toronto
Yes
Rachel
Vancouver
Phoenix
Yes
Ross
Vancouver
Paris
No
Chandler
n/a
Los Angeles
No
Joey
n/a
Paris
Yes
Monica
Dallas
Phoenix
Example 10 - Create spliced variable-length records from two

files
This example shows how you can splice data together for each pair of records with
the same ON field in two different VB input data sets, even when records are of
different lengths.
707
SPLICE Operator
//CON DD DSN=VAR.INPUT1,DISP=SHR
//
DD DSN=VAR.INPUT2,DISP=SHR
//OUT DD DSN=VAR.OUTPUT,DISP=(NEW,CATLG,DELETE),
//
SPACE=(CYL,(5,5)),UNIT=SYSDA
//TOOLIN DD *
* Splice the needed data from the two VB files together
SPLICE FROM(CON) TO(OUT) ON(5,5,CH) WITHALL WITH(12,5) WITH(22,20) VLENMAX
/*
VAR.INPUT1 has RECFM=VB and LRECL=25. It contains the base records which
look like this:
Length
25
15
25
25
|
|
|
|
|
Data
DIV01
DIV02
DIV03
DIV05
L2
L6
L8
VAR.INPUT2 has RECFM=VB and LRECL=50. It contains the overlay records

which look like this:
Length
42
33
39
19
47
43
|
|
|
|
|
|
|
Data
DIV01
DIV01
DIV02
DIV05
DIV05
DIV06
83201
73268
00589
57003
01381
37982
FERN BROTHERS INTL

ROSS INC.
ACME PAINT SHOP
FLOWERS BY RENEE
EVERYTHING FOR PETS
Because some of the overlay records are longer than their corresponding base
records, we use VLENMAX to ensure that none of the data from the overlay
records is lost. VLENMAX ensures that the larger length between the base record
and overlay record is used for the spliced record, and that blanks are added to the
end of the spliced record when needed.
The base and overlay records from the concatenated data sets are sorted and
spliced. VAR.OUTPUT has RECFM=VB and LRECL=50. It contains the spliced
records, which look like this:
Length
42
33
39
25
47
|
|
|
|
|
|
Data
DIV01
DIV01
DIV02
DIV05
DIV05
83201
73268
00589
57003
01381
L2
L2
L8
L8
FERN BROTHERS INTL

ROSS INC.
ACME PAINT SHOP
FLOWERS BY RENEE
Notice that VLENMAX prevented any data from being lost. Without VLENMAX,
data would have been lost; the spliced records would have looked like this:
Length
25
25
15
25
25
|
|
|
|
|
|
Data
DIV01
DIV01
DIV02
DIV05
DIV05
83201
73268
0058
57003
01381
L2
L2
FERN
ROSS
L8
L8
FLOW
STATS operator
708
STATS Operator
STATS FROM(indd)
ON(p,m,f)
ON(VLEN)

VSAMTYPE(x)
LMSG
Prints messages containing the minimum, maximum, average, and total for
specified numeric fields. From 1 to 10 fields can be specified.
prints messages containing the minimum, maximum, average, and total for each
field as determined by its E35 exit.
The average (or mean) is calculated by dividing the total by the record count and
rounding down to the nearest integer (examples: 23 / 5 = 4, -23 / 5 = - 4).
operator.
FROM(indd)
ON(p,m,f)
Specifies the position, length, and format of a numeric field to be used for this
operation.
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3
4
| p=
1
2
3
4
5
6
7
8
Format Code
Length
Description
BI
1 to 8 bytes
Unsigned binary
FI
1 to 8 bytes
Signed fixed-point
PD
1 to 16 bytes
ZD
1 to 31 bytes
CSF or FS

UFF
SFF
descriptions.
709
STATS Operator
If the total for a field overflows 31 digits, ICETOOL continues processing, but
prints asterisks for the average and total for that field.
bad value in a message and prints asterisks for the minimum, maximum,
average and total for that field.
For a ZD, PD, CSF, FS, or SFF format field, a negative zero value is treated as a
positive zero value.
ON(VLEN)
VSAMTYPE(x)
on page 575.
LMSG
Specifies that the minimum, maximum, average and total for all numeric fields
are to be printed using messages that display 31 digits (overriding the default
of printing messages that display 15 digits when possible). LMSG ensures that
only message ICE648I is used to display the statistics. Without LMSG, a
combination of messages ICE608I, ICE609I and ICE648I can be used to display
the statistics.
STATS example
STATS
FROM(DATA1)
ON(VLEN)
ON(15,4,ZD)
Prints messages containing the minimum, maximum, average and total of the
binary values in positions 1-2 of the DATA1 data set. For variable-length records,
this gives statistics about the length of the records. Prints messages containing the
minimum, maximum, average and total of the zoned decimal values in positions
15-18 of the DATA1 data set.
SUBSET operator
SUBSET FROM(indd)
TO(outdd)
DISCARD(savedd)
TO(outdd) DISCARD(savedd)
KEEP
REMOVE

HEADER
FIRST
HEADER(u)
FIRST(u)
710
INPUT
OUTPUT
TRAILER
LAST
TRAILER(v)
LAST(v)
RRN(q)
RRN(q,r)
RRN(r,q)
RRN(q,*)
USING(xxxx)
SUBSET Operator

VSAMTYPE(x)
Keeps or removes input or output records based on meeting criteria for the first n
records, specific relative record numbers, and the last n records. DFSORT writes
the records that are kept or not removed to the outdd data set.
DFSORT is called to copy or sort the indd data set, as appropriate. ICETOOL uses
its E15 or E35 exit to determine which records to include in the outdd or savedd
data set. ICETOOL passes the EQUALS option to DFSORT to ensure that
duplicates are kept in their original input order if records are sorted.
If the criteria includes the last n records, ICETOOL may call DFSORT twice. For
the first pass, ICETOOL counts the indd records without opening the output data
sets. For the second pass, ICETOOL opens the output data sets and does SUBSET
processing against the indd data set using the count obtained in the first pass.
DISCARD(savedd) can be used to write the records that are removed or not kept
in the savedd data set. DISCARD(savedd) can be used with or without TO(outdd).
If you do not specify a header operand (HEADER, FIRST, HEADER(u), FIRST(u)),
a relative record number operand (RRN(q), RRN(q,r), RRN(q,*)), or a trailer
operand (TRAILER, LAST, TRAILER(v), LAST(v)), all of the records will be kept or
removed. You can specify a header operand, relative record number operands, and
a trailer operand in any combination. Records will be kept or removed according
to the criteria you specify.
You can only specify one header operand. You can specify from 1 to 300 relative
record number operands in any combination. You can only specify one trailer
operand.
However, you must observe these rules for control statements in the xxxxCNTL
data set:
v MODS and OUTREC statements should not be present.
v SKIPREC and STOPAFT operands, and INCLUDE and OMIT statements, should
not be present.
v A SORT statement can be present unless INPUT and DISCARD(savedd) are
specified.
v INREC and SUM statements can be present.
v If INPUT is specified, the records selected will not be affected by INREC, SORT
or SUM. If OUTPUT is specified, the records selected will be affected by INREC,
SORT and SUM.
v Comment statements can be present.
v If you specify TO(outdd) without DISCARD(savedd), you can further process
the outdd records after SUBSET processing using an OUTFIL statement like this:

711
SUBSET Operator
v If you specify DISCARD(savedd) without TO(outdd), you can further process
the savedd records after SUBSET processing using one (and only one) OUTFIL
v If you specify TO(outdd) and DISCARD(savedd), you can further process the
outdd and savedd records after SUBSET processing using two (and only two)
OUTFIL statements like this:

Both statements must be specified in the order shown with at least the FNAMES
parameter. For example, to further modify only the DISCARD data set, you
could use statements like this:
OUTFIL FNAMES=OUT
OUTFIL FNAMES=SAVE,OMIT=(21,3,ZD,GT,+25)
If you specify a SORT statement in xxxxCNTL, the DYNALLOC option is passed

to DFSORT to ensure that work space is available for the sort. If your installation
defaults for dynamic allocation are inappropriate for a SUBSET operator, you can
take one of the following actions:

FROM(indd)
on page 575.
TO(outdd)
records it selects for the operation (that is, the records that are kept or not
removed according to the specified criteria).
specified in the FROM or DISCARD operand.
DISCARD(savedd)
records it does not select for the operation.
712
SUBSET Operator
A savedd DD statement must be present and must define an output data set
that conforms to the rules for DFSORT's OUTFIL data set.
The ddname specified in the DISCARD operand must not be the same as the
ddname specified in the FROM or TO operand.
KEEP
Specifies that the records that meet the criteria are to be kept.
REMOVE
Specifies that the records that meet the criteria are to be removed.
INPUT
Specifies that the criteria are to be applied using the first n input records,
relative input record numbers, and the last n input records. Use INPUT if you
want to apply the criteria directly to the records from the indd data set.
The criteria will be applied to keep or remove records before they are
reformatted by INREC, sorted by SORT and summed by SUM. The kept
records will subsequently be reformatted, sorted and summed. OUTFIL will be
applied to the resulting records.
As an example, if the input data set contains these records:
AAAA
AAAA
BBBB
CCCC
CCCC
CCCC
DDDD
DDDD
EEEE
EEEE
EEEE
R01
R02
R03
R04
R05
R06
R07
R08
R09
R10
R11
and the following SUBSET operator and DFSORT control statements were
specified:
...
//TOOLIN DD *
SUBSET FROM(IN) TO(OUT) KEEP INPUT RRN(3,10) USING(CTL1)
//CTL1CNTL DD *
SUM FIELDS=NONE
/*
First, input records 3-10 would be kept, so the intermediate result would be:
BBBB
CCCC
CCCC
CCCC
DDDD
DDDD
EEEE
EEEE
R03
R04
R05
R06
R07
R08
R09
R10
Then the SORT statement would be applied, so the intermediate result would
be:
713
SUBSET Operator
EEEE
EEEE
DDDD
DDDD
CCCC
CCCC
CCCC
BBBB
R09
R10
R07
R08
R04
R05
R06
R03
Finally, the SUM statement would be applied, so the final output in the OUT
data set would be:
EEEE
DDDD
CCCC
BBBB
R09
R07
R04
R03
OUTPUT
Specifies that the criteria are to be applied using the first n output records,
relative output record numbers, and the last n output records. Use OUTPUT if
you want to apply the criteria to the indd records after they are processed by
INREC, SORT and SUM as specified.
The criteria will be applied to keep or remove records after they are
reformatted by INREC, sorted by SORT and summed by SUM. OUTFIL will be
applied to the resulting records.
As an example, if the input data set contains these records:
AAAA
AAAA
BBBB
CCCC
CCCC
CCCC
DDDD
DDDD
EEEE
EEEE
EEEE
R01
R02
R03
R04
R05
R06
R07
R08
R09
R10
R11
and the following SUBSET operator and DFSORT control statements were
specified:
...
//TOOLIN DD *
SUBSET FROM(IN) TO(OUT) KEEP OUTPUT RRN(2,3) USING(CTL1)
//CTL1CNTL DD *
SUM FIELDS=NONE
/*
First, the SORT statement would be applied, so the intermediate result would
be:
EEEE
EEEE
EEEE
DDDD
DDDD
CCCC
CCCC
CCCC
BBBB
AAAA
AAAA
714
R09
R10
R11
R07
R08
R04
R05
R06
R03
R01
R02
SUBSET Operator
Then the SUM statement would be applied, so the intermediate result would
be:
EEEE
DDDD
CCCC
BBBB
AAAA
R09
R07
R04
R03
R01
Finally, output records 2-3 would be kept, so the final output in the OUT data
set would be:
DDDD R07
CCCC R04
HEADER or FIRST
Specifies one header record (the first record) is to be kept or removed.
HEADER and FIRST are equivalent to HEADER(1) and FIRST(1).
HEADER(u) or FIRST(u)
Specifies u header records (the first u records) are to be kept or removed. For
example, HEADER(3) or FIRST(3) keeps or removes the first three records.
TRAILER or LAST
Specifies one trailer record (the last record) is to be kept or removed.
TRAILER and LAST are equivalent to TRAILER(1) and LAST(1).
TRAILER(v) or LAST(v)
Specifies v trailer records (the last v records) are to be kept or removed. For
example, TRAILER(4) or LAST(4) keeps or removes the last 4 records.
RRN(q)
Specifies relative record number q is to be kept or removed. For example,
RRN(8) keeps or removes the eighth record.
RRN(q,r) or RRN(r,q)
Specifies relative record numbers q through r are to be kept or removed. q can
be less than, equal to, or greater than r. For example, RRN(5,10) and RRN(10,5)
both keep or remove the fifth through tenth records.
q and r must be specified as n or +n where n can be 1 to 99999999999999.
RRN(q,*)
Specifies relative record numbers q through the last record are to be kept or
removed. For example, RRN(7,*) keeps or removes the seventh through last
records.
Note: RRN(*,q) is not valid
USING(xxxx)
If USING is specified, an xxxxCNTL DD statement must be present. Control
statements in xxxxCNTL can be used as described for "SUBSET Operator" on
page xxx, and must conform to the rules for DFSORT's SORTCNTL data set.
715
SUBSET Operator
VSAMTYPE(x)
on page 575.
SUBSET examples
Although the SUBSET operators in the examples in this section could all be
for clarity.
Example 1
SUBSET FROM(IN1) TO(OUT1) REMOVE INPUT HEADER TRAILER
This example shows how you can remove the header record (first record) and
trailer record (last record).
The variable-length input records look like this:
Len|Data
33|01A
23|01A
26|02B
22|03A
24|04B
21|01A
RODENTS
VOLE
HAMSTER
RAT
MOUSE
COUNT
FFFFF
BINKY
GARFIELD
JUNE
MICKEY
004
We just want to keep the data records. We use REMOVE, INPUT, HEADER and
TRAILER to indicate we want to remove the header and trailer input records.
23|01A
26|02B
22|03A
24|04B
VOLE
HAMSTER
RAT
MOUSE
BINKY
GARFIELD
JUNE
MICKEY
Note that we could use FIRST and LAST instead of HEADER and TRAILER.
Example 2
SUBSET FROM(IN2) TO(OUT2) DISCARD(OUT3) KEEP INPUT RRN(3,4) LAST(3)
This example shows how you can create one output file with relative records and
the last n records, and another output file with the remaining records.
Algebra
Astronomy
Biology
Calculus
French
Geography
Geometry
Greek
History
Latin
Psychology
Russian
In the first output file (OUT2), we want the third and fourth input records and the
last three input records. In the second output file (OUT3), we want the records that
are not in the first output file. We use TO(OUT2) and DISCARD(OUT3) for the two
716
SUBSET Operator
output files. We use KEEP, INPUT, RRN(3,4) and LAST(3) to indicate we want to
keep relative input records 3 and 4 and the last 3 input records.
The OUT2 records would look like this:
Biology
Calculus
Latin
Psychology
Russian
The OUT3 records would look like this:

Algebra
Astronomy
French
Geography
Geometry
Greek
History
Note that we could use TRAILER(3) instead of LAST(3).
Example 3
SUBSET FROM(IN3) TO(OUT4) KEEP OUTPUT LAST(5) USING(CTL1)
This example illustrates how you can keep the last 5 sorted records.

Psychology
Biology
Russian
French
History
Geography
Calculus
Geometry
Algebra
Greek
Astronomy
Latin
We want to sort the records by the CH field in positions 1-15 and keep the last 5
sorted records. We use KEEP, OUTPUT and LAST(5) to keep the last 5 output
records. We use the SORT statement to sort the records before they are output.
After the SORT statement is executed, the intermediate records look like this:
Algebra
Astronomy
Biology
Calculus
French
Geography
Geometry
Greek
History
Latin
Psychology
Russian
717
SUBSET Operator
After the SUBSET operator is executed, the final output records look like this:
Greek
History
Latin
Psychology
Russian
Note that we could use TRAILER(5) instead of LAST(5).
UNIQUE operator
UNIQUE FROM(indd)
ON(p,m,f)
ON(VLEN)

VSAMTYPE(x)
UZERO
Prints a message containing the count of unique values for a specified numeric or
character field.
DFSORT is called to sort the indd data set to ICETOOL's E35 user exit. ICETOOL
prints a message containing the unique count as determined by its E35 user exit.
inappropriate for a UNIQUE operator, you can take one of the following actions:
OPTION DYNALLOC=(8)
in the DFSPARM data set.

2. Use SORTWKdd DD statements to override the use of dynamic allocation.
Refer to SORTWKdd DD statement on page 71 for details.
Attention: Either of these actions affects the work data sets used for an OCCUR
operator, or for a SELECT or SPLICE operator for which USING(xxxx) is not
specified.
You must not supply your own DFSORT MODS, INREC, OUTREC, SUM or
RECORD statement, because they override the DFSORT statements passed by
ICETOOL for this operator.
FROM(indd)
ON(p,m,f)
used with this operation.
718
UNIQUE Operator
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3
4
| p=
1
2
3
4
5
6
7
8
f specifies the format of the field as shown in the following table:
Format Code
Length
Description
BI
1 to 256 bytes
Unsigned binary
FI
1 to 256 bytes
Signed fixed-point
PD
1 to 32 bytes
ZD
1 to 32 bytes
CH
1 to 4000 bytes
Character
CSF or FS
1 to 32 bytes

UFF
1 to 44 bytes
SFF
1 to 44 bytes
descriptions.

zoned decimal values F2F3C1, F2F3F1, and 020301 are counted as only one
unique value.
unique value.
v Digits are not checked for validity.
ON(VLEN)
VSAMTYPE(x)
on page 575.
UZERO
See the discussion of this operand on the OCCUR statement on OCCUR
UNIQUE example
UNIQUE FROM(DATAIN)
UNIQUE FROM(DATAIN)
ON(20,40,CH)
ON(5,3,ZD)
The first UNIQUE operator prints a message containing the count of unique
character data in positions 20-59 of the DATAIN data set.
The second UNIQUE operator prints a message containing the count of unique
zoned decimal values in positions 5-7 of the DATAIN data set.
719
VERIFY Operator
VERIFY operator
VERIFY FROM(indd) ON(p,m,f)

NOSIGN
LIMIT(n)

VSAMTYPE(x)
Examines particular decimal fields in a data set and prints a message identifying
each invalid value found for each field. From 1 to 10 fields can be specified.
uses its E35 user exit to examine the digits and sign of each value for validity, and
for each invalid value found, prints an error message containing the record
number and field value (in hexadecimal).
operator.
Note:
1. Values with invalid decimal digits are also identified for the DISPLAY, OCCUR,
RANGE, and STATS operators.
2. The DISPLAY operator can be used to print a report identifying the relative
record number, hexadecimal value, and associated fields for each invalid (and
valid) decimal value, as shown in Example 9 under DISPLAY operator on
page 594.
FROM(indd)
ON(p,m,f)
Specifies the position, length, and format of a decimal field to be used for this
operation.
Fixed-length record
|
| D | A | T | A | ... |
| R | R | R | R | D | A | T | A | ...
p= 1 2 3 4
| p=
1
2
3
4
5
6
7
8
f specifies the format of the field as shown later in this section:
720
VERIFY Operator
Format Code
Length
Description
PD
1 to 16 bytes
ZD
1 to 31 bytes
descriptions.
A value is considered invalid under one of the following circumstances:

v it contains A-F as a digit (examples: a PD field of 00AF or a ZD field of
F2FB)
v it contains 0-9 as a sign and the NOSIGN operand is not specified
(examples: a PD field of 3218 or a ZD field of F235).
If the number of bad values reaches the LIMIT for invalid decimal values,
ICETOOL terminates the operation. If the LIMIT operand is not specified, a
default of 200 is used for the invalid decimal value limit.
NOSIGN
Specifies that the sign of the decimal values is not to be validity checked
(overriding the default of checking for 0-9 as invalid signs).
LIMIT(n)
VSAMTYPE(x)
on page 575.
VERIFY example
VERIFY FROM(NEW) ON(22,16,PD) ON(7,9,ZD)
VERIFY FROM(NEW) ON(22,16,PD) ON(7,9,ZD) NOSIGN LIMIT(10)
The first VERIFY operator checks for invalid digits (A-F) and invalid signs (0-9) in
the packed decimal values from positions 22-37 and the zoned decimal values from
positions 7-15 of the NEW data set. A message is printed identifying each value (if
any) that contains an invalid digit or sign. If 200 invalid values are found, the
operation is terminated.
The second VERIFY operator checks for invalid digits (A-F) in the packed decimal
values from positions 22-37 and the zoned decimal values from positions 7-15 of
the NEW data set. A message is printed identifying each value (if any) that
contains an invalid digit. If 10 invalid values are found, the operation is
terminated.
Note: The DISPLAY operator can be used to print a report identifying the relative
record number, hexadecimal value, and associated fields for each invalid (and
valid) decimal value, as shown in Example 9 under DISPLAY operator on page
594.
721
Calling ICETOOL from a Program
Calling ICETOOL from a program

ICETOOL can be called from an assembler program using the LINK, ATTACH, or
XCTL system macros. Standard linkage conventions must be used. When ICETOOL
finishes processing, it returns to the calling program with register 15 set to the
highest operation return code encountered. See ICETOOL return codes on page
728 for an explanation of the ICETOOL return codes.
When you call ICETOOL from a program, you have a choice of two different
interfaces: the TOOLIN Interface and the Parameter List Interface.
TOOLIN interface
With the TOOLIN Interface, you supply ICETOOL statements in the TOOLIN data
set. ICETOOL prints messages in the TOOLMSG data set, but does not return
information directly to your program.
To use the TOOLIN interface, set a value of 0 in register 1, or place the address of
a 4-byte field containing X'80000000' in register 1, before calling ICETOOL as
follows:
...
SLR
R1,R1
LINK EP=ICETOOL
...
LA
R1,NOPLIST
LINK EP=ICETOOL
...
NOPLIST DC
X80,AL3(0)
...
TOOLIN INTERFACE - METHOD 1

CALL ICETOOL
TOOLIN INTERFACE - METHOD 2
CALL ICETOOL
TOOLIN INTERFACE INDICATOR
Parameter list interface

The Parameter List Interface allows you to use the information derived by
ICETOOL in your program. With this interface, you supply ICETOOL statements in
a parameter list. ICETOOL prints messages in the TOOLMSG data set, and puts an
operation status indicator and operation-specific values in the parameter list for
use by your calling program.
Figure 32 shows the format of the parameter list used with the Parameter List
Interface. Table 86 on page 724 shows the operation-specific values returned to the
calling program.
Register 1
Flags
Statement Area 1 Address
Return Area 1 address
Starement Area n Address

Return Area n Address
X'FFFFFFFF"
Figure 32. Parameter List for Parameter List Interface
722

The flags field must be specified. A 4-byte field containing X' FFFFFFFF' must be
used to indicate the end of the parameter list. It can be coded after any pair of
statement/return addresses.
All addresses in the parameter list must be 31-bit addresses or clean 24-bit
addresses (the first 8 bits contain zeros).
Explanation of fields
Flags
Bit 0 = 0:
Use the Parameter List Interface. Process ICETOOL statements from
this parameter list and return information to this parameter list. Ignore
TOOLIN.
Bit 0 = 1:
Use the TOOLIN Interface. Process ICETOOL statements from
TOOLIN. Ignore this parameter list.
Bits 1-31:
Reserved. Must be set to zero to ensure future extendability.
Statement Area Address and Statement Area
Each statement area address gives the location of a statement area that
describes an ICETOOL operation to be performed. If the statement area
address is 0, ICETOOL ignores this statement area/return area pair. Otherwise,
the statement area address must point to a statement area in the following
format:
v A 2-byte length field containing the length of the statement area for this
operation. If this field is 0, ICETOOL ignores this statement area/return area
pair.
v One or more 80-character ICETOOL statement images in the format
described in ICETOOL statements on page 574. Each statement area must
have one operator statement. Comment and blank statements before the
operator statement are processed. Comment, blank, and additional operator
statements after the end of the first operator statement are ignored.
Return Area Address and Return Area
Each return area address gives the location of a return area in which ICETOOL
is to return operation-specific information for the operation described in the
corresponding statement area. If the return area address is 0, ICETOOL does
not return any information for this operation. Otherwise, the return area
address must point to a return area in the following general format:
v A 2-byte length field containing the length of the return area for this
operation. If this field is 0, ICETOOL does not return any information for
this operation.
v A 1-byte operation status indicator which is set by ICETOOL as follows:
0=
This operation was run and completed with return code 0 or 4.

Operation-specific values (see the following bullet) were returned.
This operation was not run (for example, scan mode was in effect) or
did not complete with return code 0 or 4. Operation-specific values
(see the following bullet) were not returned.
v Operation-specific values. Each count value returned by ICETOOL is an
8-byte packed decimal value with a C for a positive sign or a D for a
negative sign. For a STATS operator, each minimum, maximum, average and
total value returned by ICETOOL is a 16-byte packed decimal value with a
4=
723

C for a positive sign or a D for a negative sign. If ICETOOL set the
operation status to 4, it did not return any values for this operation.
Note: Programs in LPALIB that call ICETOOL must provide return areas that
ICETOOL can store into.
The required return area length and the operation-specific values returned for
each operator are shown in Table 86. If the return area length is less than the
length required, ICETOOL issues a message and terminates the operation.
Table 86. Return Area Lengths/Operation-Specific Values
Return Area Length (Bytes)
724
Operation-Specific Values
Returned
COPY
None
COUNT
Count of records processed,

or 0 if any criteria specified.
If SUB(q) or ADD(r) is
specified, the modified count
is returned.
DATASORT
None
DEFAULTS
None
DISPLAY
Count of records processed
MERGE
None
MODE
None
OCCUR
17

count of records resulting
from criteria
RANGE
17

count of values in range
RESIZE
None
SELECT
17

from criteria
SORT
None
SPLICE
17

from criteria
STATS
64*n+9

minimum for ON field 1,
maximum for ON field 1,
average for ON field 1, total
for ON field 1, ... minimum
for ON field n, maximum for
ON field n, average for ON
field n, total for ON field n
SUBSET
None
UNIQUE
17

count of unique values
VERIFY
Count of records processed
Parameter list interface example

The example in Figure 33 on page 726 shows a portion of an assembler language
program that uses the Parameter List Interface. Figure 34 on page 727 shows the
JCL you might use to run the program in Figure 33 on page 726.
725
DEPTVIEW CSECT
...
* SET UP PARAMETER LIST AND CALL ICETOOL
LA
R1,PARLST
LOAD ADDRESS OF PARAMETER LIST
LINK EP=ICETOOL
CALL ICETOOL
LTR R15,R15
IF ANY OPERATIONS WERE NOT SUCCESSFUL,
BNZ CKSTAT1
DETERMINE WHICH ONE FAILED
* ALL OPERATIONS WERE SUCCESSFUL
* CHECK EMPLOYEES PER DEPARTMENT AGAINST ACCEPTABLE LEVEL
CP
RT2AVG1,EMAVGCK
IF AVERAGE IS ACCEPTABLE,
BNH CKQUAL
NO MESSAGE IS NEEDED
* ISSUE A MESSAGE SHOWING AVERAGE, MINIMUM, MAXIMUM, AND
* TOTAL NUMBER OF EMPLOYEES PER DEPARTMENT.
...
* CHECK EXPENSES PER DEPARTMENT AGAINST ACCEPTABLE LEVEL
CKQUAL CP
RT2AVG2,TLAVGCK
IF AVERAGE IS ACCEPTABLE,
BNH PCTCALC
NO MESSAGE IS NEEDED
* ISSUE A MESSAGE SHOWING AVERAGE, MINIMUM, MAXIMUM, AND
* TOTAL EXPENSES PER DEPARTMENT.
...
* CALCULATE THE PERCENTAGE OF DEPARTMENTS OVER/UNDER EMPLOYEE LIMIT
PCTCALC MVC WORK+2(4),RT3RCDS+4 COPY NUMBER OF DEPARTMENTS
SP
WORK+2(4),RT3RNG+4(4) SUBTRACT NUMBER WITHIN LIMITS TO
*
GET NUMBER OVER/UNDER LIMIT
CP
WORK+2(4),P0
IF NONE OVER/UNDER LIMIT,
BE
PCTPRT
PERCENTAGE IS ZERO
MP
WORK+2(4),P100
MULTIPLY NUMBER OVER/UNDER BY 100
DP
WORK(6),RT3RCDS+4(4) DIVIDE BY NUMBER OF DEPARTMENTS
* ISSUE A MESSAGE SHOWING THE PERCENTAGE OF DEPARTMENTS THAT ARE
* OVER/UNDER EMPLOYEE LIMIT
PCTPRT UNPK PCTVAL,WORK(2)
CONVERT AVERAGE TO PRINTABLE EBCDIC
OI
PCTVAL+2,XF0
ENSURE LAST DIGIT IS PRINTABLE
...
* ONE OR MORE OPERATIONS FAILED
CKSTAT1 CLI RT1STAT,0
IF OPERATION 1 WORKED,
BNE CKSTAT2
CHECK OPERATION 2
* ISSUE MESSAGE: OPERATION 1 FAILED - CHECK TOOLMSG
...
* PARAMETER LIST
PARLST DC
A(0)
USE PARAMETER LIST INTERFACE
DC
A(ST1A)
STATEMENT AREA 1 ADDRESS
DC
A(RT1A)
RETURN AREA 1 ADDRESS
DC
A(ST2A)
DC
A(RT2A)
DC
A(ST3A)
DC
A(RT3A)
DC
F.*-1
END OF PARAMETER LIST* OPERATOR STATEMENT AREAS
726
* COPY OPERATION
ST1A
DC
AL2(ST1E-ST1)
LENGTH OF STATEMENT AREA 1
ST1
DC
CL80* CREATE TWO COPIES OF THE DENVER SITE
DC
CL80* DEPARTMENT RECORDS
DC
CL80COPY FROM(IN) TO(OUT1,OUT2) USING(CTL1)
ST1E
EQU *
* STATS OPERATION
ST2A
DC
AL2(ST2E-ST2)
ST2
DC
CL80* GET STATISTICS FOR NUMBER OF EMPLOYEES
DC
CL80* AND TRAVEL EXPENSES PER DEPARTMENT
DC
CL80STATS FROM(OUT1) ON(15,2,PD) ON(28,8,ZD)
ST2E
EQU *
* RANGE OPERATION
ST3A
DC
AL2(ST3E-ST3)
ST3
DC
CL80* DETERMINE THE NUMBER OF DEPARTMENTS THAT ARE
DC
CL80* WITHIN THE LIMIT FOR NUMBER OF EMPLOYEES
DC
CL80RANGE FROM(OUT1) ON(15,2,PD) -
DC
CL80 HIGHER(10) LOWER(21)
ST3E
EQU *
* RETURN AREAS
COPY OPERATION
RT1A
DC
AL2(RT1E-RT1STAT) LENGTH OF RETURN AREA 1
RT1STAT DS
C
OPERATION STATUS
RT1E
EQU *
* STATS OPERATION
RT2A
DC
AL2(RT2E-RT2STAT) LENGTH OF RETURN AREA 2
RT2STAT DS
C
OPERATION STATUS
RT2RCDS DS
PL8
COUNT OF RECORDS PROCESSED
RT2MIN1 DS
PL16
FIELD 1 - MINIMUM VALUE
RT2MAX1 DS
PL16
FIELD 1 - MAXIMUM VALUE
RT2AVG1 DS
PL16
FIELD 1 - AVERAGE VALUE
RT2TOT1 DS
PL16
FIELD 1 - TOTAL VALUE

//INVOKE EXEC PGM=DEPTVIEW,REGION=1024K
//STEPLIB DD DSN=... Link library containing DEPTVIEW
//IN DD DSN=ALL.DEPTS,DISP=SHR
//OUT1 DD DSN=ALL.DEPTS.BACKUP1,DISP=OLD
//OUT2 DD DSN=ALL.DEPTS.BACKUP2,DISP=OLD
//CTL1CNTL DD *
* SELECT ONLY THE DENVER SITE DEPARTMENT RECORDS
INCLUDE COND=(1,12,CH,EQ,CDENVER)
/*
Figure 34. JCL for Parameter List Interface Program Example
ICETOOL notes and restrictions

v Small REGION values can cause storage problems when ICETOOL calls
DFSORT. Large REGION values give DFSORT the flexibility to use the storage it
needs for best performance. We recommend that you use a REGION value of at
least 1024K for ICETOOL.
v Each ICETOOL operation results in a set of ICETOOL messages in the
TOOLMSG data set, and a corresponding set of DFSORT messages in the
DFSMSG data set. For a particular call to DFSORT, you can relate the sets of
messages in the TOOLMSG and DFSMSG data sets by using the unique
identifier for that call. Just match the identifier printed in ICETOOL message
ICE606I or ICE627I to the same identifier printed in DFSORT message ICE200I.
This is particularly important if an ICETOOL operation fails due to an error
detected by DFSORT (return code 16).
v Because ICETOOL calls DFSORT, the installation defaults used for DFSORT are
those in effect for the appropriate program-invoked environment, that is, the
ICEAM2 or ICEAM4 defaults, or the ICETDx defaults activated for ICEAM2 or
ICEAM4. The DFSORT installation defaults apply only to DFSORT, not to
ICETOOL. For example, installation option MSGCON=ALL causes DFSORT, but
not ICETOOL, to write messages to the master console. The one exception is the
SDBMSG installation option; the value in effect from ICEAM2 or ICEAM4 is
used for ICETOOL's TOOLMSG data set.
v When ICETOOL calls DFSORT, it passes control statements and options
appropriate to the specific operations being performed. You should not override
the DFSORT control statements or options passed by ICETOOL unless you
understand the ramifications of doing so.
For example, ICETOOL passes the NOABEND option to DFSORT to ensure that
ICETOOL will regain control if DFSORT issues an error message. If you specify:
//DFSPARM DD *
DEBUG ABEND
you cause DFSORT to abend when it issues an error message, thus preventing
ICETOOL from performing subsequent operators.
v Tape work data sets cannot be used with ICETOOL.
v An ON field must not include bytes beyond the fixed part of variable length
input records. The entire field specified must be present in every input record,
otherwise, DFSORT issues message ICE015A, ICE218A, or ICE027A and
terminates.
v
727
Notes on Using JOINKEYS with COPY and SORT
Notes on using JOINKEYS with COPY and SORT

v
v
v
You can use a JOINKEYS application with a COPY or SORT operator of

ICETOOL, but not with any of the other ICETOOL operators. Use the JKFROM
operand instead of the FROM operand to indicate you are using a JOINKEYS
application.
See Chapter 4, Using a JOINKEYS application for joining two files, on page
457 for complete details on JOINKEYS applications.
If you want to sort the joined records, use a SORT operator with USING(xxxx) in
TOOLIN and a SORT control statement along with the appropriate JOINKEYS,
JOIN and REFORMAT statements in xxxxCNTL (main task).
If you want to copy the joined records, use a COPY operator with USING(yyyy)
in TOOLIN and the appropriate JOINKEYS, JOIN and REFORMAT statements in
yyyyCNTL (main task).
For multiple JOINKEYS applications within a single ICETOOL step:
You can reference different JOINKEYS F1 and F2 input files for the different
SORT and/or COPY operators by using F1=ddname and F2=ddname on the
various JOINKEYS statements as appropriate
You can reference different subtaskn message and control files for the
different SORT and/or COPY operators by using TASKID=id on the various
JOINKEYS statements as appropriate.
ICETOOL return codes

ICETOOL sets a return code for each operation it performs in STOP or CONTINUE
mode and passes back the highest return code it encounters to the operating
system or the invoking program.
For successful completion of all operations, ICETOOL passes back a return code of
0 or 4 to the operating system or the invoking program.
For unsuccessful completion due to an unsupported operating system, ICETOOL
passes back a return code of 24 to the operating system or invoking program.
For unsuccessful completion of one or more operations, ICETOOL passes back a
return code of 8, 12, 16, or 20 to the operating system or the invoking program.
The meanings of the return codes that ICETOOL passes back (in register 15) are:
0
Successful completion. All operations completed successfully.
Successful completion. All operations completed successfully. Either:

v RC4 was specified for an ICETOOL COUNT operator and the record
count met the specified criteria, or
v DFSORT passed back a return code of 4 for one or more operations. See
DFSORT messages and return codes on page 25 for details.
728
Unsuccessful completion. RC8 was specified for an ICETOOL COUNT

operator and the record count met the specified criteria.
12
Unsuccessful completion. ICETOOL detected one or more errors that

prevented it from completing successfully. Messages for these errors were
printed in the TOOLMSG data set.
16
Unsuccessful completion. DFSORT detected one or more errors that
ICETOOL Return Codes

prevented ICETOOL from completing successfully. Messages for these
errors were printed in the DFSMSG data set.
20
Message data set error. The TOOLMSG DD statement was not present or
the TOOLMSG data set was not opened.
24
Unsupported operating system. This operating system is not supported by

this release of DFSORT.
729
ICETOOL Return Codes
730
Chapter 8. Using symbols for fields and constants

Field and constant symbols overview
This chapter describes DFSORT's simple and flexible method for using symbols in
DFSORT and ICETOOL statements. You can define and use a symbol for any field,
constant, or output column that is recognized in a DFSORT control statement or
ICETOOL operator. You can also use symbols for numbers associated with various
keywords in DFSORT and ICETOOL statements. This makes it easy to create and
reuse collections of symbols (that is, mappings) representing information associated
with various record layouts.
In addition, you can obtain and use collections of DFSORT symbols created
specifically for records produced by other products (for example, RACF,
DFSMSrmm and DCOLLECT) or by your site. Visit the DFSORT home page at the
following URL to obtain information about downloading DFSORT symbol
mappings for records produced by other products, and examples that use these
symbols:
http://www.ibm.com/storage/dfsort
Symbols can increase your productivity by automatically providing the positions,

lengths and formats of the fields, the values of the constants, and the positions of
the output columns associated with the particular records you are processing with
DFSORT or ICETOOL.
You can even use system symbols (for example, &JOBNAME.) and SET and PROC
symbols, in your symbol constants.
To use symbols for DFSORT or ICETOOL, you just:
1. Create or obtain the DFSORT symbol data sets you need. Symbol data sets
contain symbols that map the fields in your records, and constants used for
comparisons, titles, headings and so on. The symbols are specified in DFSORT's
simple but flexible SYMNAMES statement format. Symbols can be easily added
or modified using an editor, such as ISPF EDIT.
2. Include a SYMNAMES DD statement in your job. SYMNAMES specifies one or
more symbol data sets (sequential, partitioned member, DD *) to be used for
your DFSORT or ICETOOL application. SYMNAMES can be used to
concatenate as many symbol data sets as you like.
3. Use the symbols from SYMNAMES where appropriate in DFSORT control
statements or ICETOOL operators. You can mix symbols (for example,
Last_Name) with regular fields (for example, p,m,f) and constants (for example,
C'string').
DFSORT or ICETOOL will read SYMNAMES and use the symbols it contains to
"transform" your statements by performing symbol substitution. DFSORT or
ICETOOL will then use the transformed statements as if you had specified them
directly.
If your record layout changes, just make a corresponding change to your DFSORT
symbol data set. DFSORT will use the new mapping to "transform" your symbols
731
Using Symbols for Fields and Constants

correctly, even if positions change, so you won't have to change your statements.
Be sure that your symbol definitions match your record layout before you attempt
to use them.
DFSORT example
The example in this section shows the JCL and control statements for a simple
DFSORT job that uses symbols.
Let's say you created a symbols data set named MY.SYMBOLS that contains the
following SYMNAMES statements:
* Fields
First_Name,6,20,CH
Last_Name,*,=,=
Account_Number,53,3,PD
SKIP,2
Balance,*,6,ZD
Type,*,8,CH
* Constants
Loan,LOAN
Check,CHECKING
Level1,50000
Level2,-100
Here's the JCL and control statements for the example:

//RUNIT EXEC PGM=ICEMAN
//SYMNAMES DD DSN=MY.SYMBOLS,DISP=SHR
//SYMNOUT DD SYSOUT=*
//SORTIN DD ...
//SORTOUT DD ...
//SYSIN
DD
*
INCLUDE COND=((Type,EQ,Loan,AND,Balance,GT,Level1),OR,
(Type,EQ,Check,AND,Balance,LE,Level2))
SORT FIELDS=(Last_Name,A,First_Name,A,
Type,A,Account_Number,D)
/*
This example is only meant to give you a quick overview of how symbols can be
used. The rest of this chapter will explain all of the details, but here are a few
important things to take note of:
v The SYMNAMES DD indicates you want DFSORT or ICETOOL to do symbol
processing. The SYMNAMES data set contains the symbols for fields and
constants.
v DFSORT or ICETOOL will print your original symbols and the symbol table
constructed from them in the SYMNOUT data set, if you specify it. You might
want to use SYMNOUT while debugging a set of symbols and then remove it,
or you might want to keep SYMNOUT permanently so you can always see your
original symbols and the symbol table.
v The simple, yet flexible, format for SYMNAMES statements is:
symbol,value remark
where value can represent a field (p,m,f or p,m or p), a parsed field (%nn), or a
constant (C'string', c'string', 'string', S'string', s'string', X'string', x'string', B'string',
b'string', n, +n or -n). Leading blanks are allowed before symbol so indentation
can be used. For example, the following SYMNAMES statements could be
specified:
732

Div1_Department,8,1,BI
Research,B0001....
Marketing,B0010....
Development,B0100....
Division 1 Department
Research Departments
Marketing Departments
Development Departments
v Symbols are case-sensitive: Frank, FRANK and frank are three different
symbols.
v An asterisk (*) can be used to assign the next position to p. For example:
Symbola,6,20,CH
Symbolb,*,5,BI
Symbolc,*,12,ZD
is the same as specifying:

Symbola,6,20,CH
Symbolb,26,5,BI
Symbolc,31,12,ZD
By using * for p, you can map consecutive fields in your records without having
to compute their actual positions.
v SKIP,n can be used to advance the next position by n bytes so it can be used for *.
For example:
Symbola,6,20,CH
SKIP,2
Symbolb,*,5,BI

Symbola,6,20,CH
Symbolb,28,5,BI
SKIP,n gives you an easy way to skip unused bytes. Other mapping aids allow
you to reset the next position (POSITION,q or POSITION,symbol), or align the
next position on a halfword (ALIGN,H), fullword (ALIGN,F) or doubleword
(ALIGN,D).
v An equal sign (=) can be used for p, m or f to assign the previous position,
length or format to p, m, or f, respectively. For example:
Symbola,6,20,CH
Symbola1,=,8,=
Symbola2,*,12,=
Symbold,*,=,ZD

Symbola,6,20,CH
Symbola1,6,8,CH
Symbola2,14,12,CH
Symbold,26,12,ZD
By using = and *, you can easily map fields onto other fields.
v Symbols for fields and constants can be specified in any order. However, the use
of * and = imposes order dependencies on symbols for fields.
v Comment statements and blank statements are allowed in SYMNAMES.
SYMNAMES DD statement
A SYMNAMES DD statement indicates you want DFSORT or ICETOOL to do
symbol processing. It specifies the SYMNAMES data set (SYMNAMES for short),
which can consist of one DFSORT symbol data set or many concatenated symbol
data sets.
733

A symbol data set can be a sequential data set, a partitioned member or a DD *
data set; all three types can be concatenated together for the SYMNAMES DD.
Each symbol data set must contain SYMNAMES statements describing the symbols
for fields and constants to be used for the DFSORT or ICETOOL application. Each
symbol data set must have the following attributes: RECFM=F or RECFM=FB and
LRECL=80.
For best performance, use a large block size, such as the system determined
optimum block size, for all DFSORT symbol data sets.
If a SYMNAMES DD statement is not present, or SYMNAMES is empty, symbol
processing is not performed.
SYMNOUT DD statement
A SYMNOUT DD statement specifies a data set in which you want DFSORT or
ICETOOL to print your original SYMNAMES statements and the symbol table
constructed from them. DFSORT or ICETOOL uses RECFM=FBA, LRECL=121 and
the specified BLKSIZE for the SYMNOUT data set (SYMNOUT for short).
If the BLKSIZE you specify is not a multiple of 121, or you do not specify the
BLKSIZE:
v the system determined optimum block size is used, if supported
v otherwise, BLKSIZE=121 is used.
For best performance, use a large block size, such as the system determined
optimum block size, for the SYMNOUT data set.
SYMNAMES statements
Each symbol in SYMNAMES must be described using a SYMNAMES statement. A
SYMNAMES statement can be a symbol statement, keyword statement, comment
statement or blank statement.
Comment and blank statements

A statement with an asterisk (*) in column 1 is treated as a comment statement. It
is printed in SYMNOUT (if specified), but otherwise not processed. A statement
with blanks in columns 1 through 80 is treated as a blank statement. It is printed
in SYMNOUT (if specified), but otherwise not processed.
Symbol statements
The general format for a symbol statement is:
symbol,value remark
The general coding rules are as follows:

v Columns 1 through 80 are scanned.
v The symbol can start in column 1 or in any column after 1.
v A remark is optional, but if specified, must be separated from the value by at
least one blank. A remark is printed in SYMNOUT (if specified), but otherwise
not processed.
v A semicolon (;) can be used instead of a comma (,) between the symbol and the
value.
734

v Continuation is not allowed. Each symbol and value must be completely
specified on one line.
The specific syntax for symbol statements is:
symbol,
constant
field
parsed field
Symbol: A symbol can be 1 to 50 EBCDIC characters consisting of uppercase

letters (A-Z), lowercase letters (a-z), numbers (0-9), the number sign (#), the dollar
sign ($), the commercial at sign (@), the underscore(_) and the hyphen(-). The first
character of a symbol must not be a number or a hyphen. Symbols are treated as
case-sensitive: Frank, FRANK and frank are three different symbols.
The following DFSORT/ICETOOL reserved words (uppercase only as shown) are
not allowed as symbols: A, AC, ADD, ADDDAYS, ADDMONS, ADDYEARS, ALL,
AND, AQ, ASL, AST, BI, CH, CLO, COPY, COUNT, COUNT15, CSF, CSL, CST,
CTO, D, DATE, DATEDIFF, DATE1, DATE1..., DATE2, DATE2..., DATE3, DATE3...,
DATE4, DATE5, DC1, DC2, DC3, DC4, DE1, DE2, DE3, DE4, DIV, DT1, DT2, DT3,
D1, D2, E, F, FI, FL, FS, H, HEX, LASTDAYM, LASTDAYQ, LASTDAYW,
LASTDAYY, LC, LN, LS, MAX, MC, MIN, MN, MOD, MUL, Mn, Mnn, NEXTDday,
NONE, NUM, OL, OR, OT, PAGE, PAGEHEAD, PD, PDC, PDF, PD0, PREVDday,
SEQNUM, SFF, SS, SUB, SUBCOUNT, SUBCOUNT15, SUBDAYS, SUBMONS,
SUBYEARS, TC1, TC2, TC3, TC4, TE1, TE2, TE3, TE4, TIME, TIME1, TIME1P,
TIME2, TIME2P, TIME3, TIME3P, TM1, TM2, TM3, TM4, TS, UC, UFF, UN,
VALCNT, VLEN, X, Y2x, Y2xx, Y4x, Z, ZD, ZDC, and ZDF, where n is 0-9, day is
SUN, MON, TUE, WED, THU, FRI or SAT, and x is any character. Lower case and
mixed case forms of these words, such as None and page, can be used as symbols.
POSITION, SKIP and ALIGN (uppercase only) are treated as keywords as
discussed in Keyword statements on page 744 and thus are not recognized as
symbols. However, lowercase and mixed case forms of these words, such as
Position and skip, can be used as symbols.
Some examples of valid symbols are: Account_Number, CON12, PHONE#, count,
SORT-KEY, and _Invalid.
Some examples of invalid symbols are: 123_Account (starts with a number),
COUNT (reserved word), DATE1-20 (reserved word), and -Invalid (starts with a
hyphen).
Constant: A constant can be a character string, system symbol string, hexadecimal
string, bit string, decimal number, or two-digit year date string.
A symbol for a constant value must be used only where such a constant is allowed
and has the desired result. Otherwise, substitution of the constant for the symbol
will result in an error message or unintended processing. For example, if the
following SYMNAMES statement is specified:
SYMB,B10110001
SYMB can be used in this INCLUDE statement:

INCLUDE COND=(12,1,BI,EQ,SYMB)
735

because a bit string can be compared to a binary field. However, SYMB will result
in an error message if used in this INCLUDE statement:
INCLUDE COND=(12,1,CH,EQ,SYMB)
because a bit string cannot be compared to a character field.

Make sure the constants that will be substituted for your symbols are appropriate.
If in doubt, check the rules for constants given in the description of the relevant
operand.
A symbol can represent one of the following types of constants:
v A character string in the format 'xx...x', C'xx...x' or c'xx...x'.
The value x may be any EBCDIC character. You can specify up to 64 characters
for the string. c'xx...x' will be treated like C'xx...x'.
Note: If you want to use system symbols (for example, &JOBNAME.) in your
character strings, use a system symbol string (S'string' or s'string') as described
later in this section instead of a character string.
If you want to include a single apostrophe in the character string, you must
specify it as two single apostrophes (each pair of apostrophes counts as two
characters towards the 64 character limit for the string). Thus:
Required:
ONeill
Specify:
CONeill
Double-byte data may be used in a character string (each pair of

shift-in/shift-out characters and each double-byte character counts as two
characters towards the 64 character limit for the string). See INCLUDE control
statement on page 96 for details on double-byte data.
Some examples of valid character strings are: '+0.193', c'Title', C'O''Neil',
C'J62,J82,M72' and ''.
Some examples of invalid character strings are: C'AB'' (apostrophes not paired),
c'title (ending apostrophe missing) and C'O'NEIL' (one apostrophe after O
instead of two).
You can use C'xx...x' and 'xx...x' interchangeably. 'xx...x' will be substituted for
symbols where appropriate even if C'xx...x' is specified in SYMNAMES.
Likewise, C'xx...x' will be substituted for symbols where appropriate even if
'xx...x' is specified in SYMNAMES. For example, if these SYMNAMES statements
are specified:
My_Title,cMy Report
My_Heading,CJanuary
DEPT1,J82
DEPT2,cM72
the ICETOOL operator:

DISPLAY TITLE(My_Title) HEADER(My_Heading) ...
will be transformed to:

DISPLAY TITLE(My Report) HEADER(January) ...
and the INCLUDE statement:

INCLUDE COND=(5,3,EQ,DEPT1,OR,5,3,EQ,DEPT2),FORMAT=CH

INCLUDE COND=(5,3,EQ,CJ82,OR,5,3,EQ,CM72),FORMAT=CH
736

Although the rules for character strings used as symbols generally follow the
rules for INCLUDE/OMIT character strings, keep in mind that the same rules
do not apply for character strings in all DFSORT and ICETOOL operands, so use
symbols representing character strings appropriately. For example, ICETOOL
only allows up to 50 characters for a TITLE string, so TITLE(MYCON) would
result in an error message if MYCON is a 64-character string, even though
MYCON could be used without error in an INCLUDE statement. As another
example, double-byte characters would be recognized in a character string
substituted for a symbol in an INCLUDE statement, but would not be
recognized in a character string substituted in an OUTREC statement.
v A system symbol string in the format S'original_string' or s'original_string'
The original_string can contain any combination of EBCDIC characters and
system symbols (&SYMBOL or &SYMBOL.) you want to use to form a character
string. DFSORT will replace each system symbol in the system symbol string
with its substitution text to create a character string in the format
C'result_string'.
For example, you could specify the following symbol statement in SYMNAMES:
Rpt_hdr,S
Job &JOBNAME. was run on Sysplex &SYSPLEX. on &LWDAY
If you used this symbol statement in a job named TEST2 that you ran on sysplex
MAS3 on a Thursday, DFSORT would transform it into the following symbol
statement:
Rpt_hdr,C
Job TEST2 was run on Sysplex MAS3 on THU
&JOBNAME. was replaced with 'TEST2', &SYSPLEX. was replaced with 'MAS3'
and &LWDAY was replaced with 'THU'.
If you used this same symbol statement in a job named BIGTEST that you ran
on sysplex MAS2 on a Monday, DFSORT would transform it into the following
symbol statement:
Rpt_hdr,C
Job BIGTEST was run on Sysplex MAS2 on MON
This time &JOBNAME. was replaced with 'BIGTEST', &SYSPLEX. was replaced
with 'MAS2' and &LWDAY was replaced with 'MON'.
You could use the Rpt_hdr symbol in an OUTFIL statement like this:
OUTFIL HEADER2=(Rpt_hdr,5X,Page ,PAGE=(EDIT=(TTT)),/)
and get the heading you needed based on the setting of the system symbol
values where and when the job was run.
The types of system symbols you can use in a system symbol string are:
Dynamic system symbols like &YYMMDD, &LYYMMDD, &HHMMSS,
&LHHMMSS, &DAY, &LDAY, &HR, &LHR, &JDAY, &LJDAY, &JOBNAME,
&MIN, &LMIN, &MON, &LMON, &SEC, &LSEC, &WDAY, &LWDAY, &YR2,
&LYR2, &YR4, and &LYR4.
System-defined static system symbols like &SYSCLONE, &SYSNAME,
&SYSPLEX, &SYSR1, and &SYSALVL.
Installation-defined static system symbols specified by your installation in an
IEASYMxx member of SYS1.PARMLIB.
Note: System symbols must be specified using all uppercase characters.
Lowercase characters are not recognized as system symbols. For example,
&JOBNAME is recognized as a system symbol, but &jobname and &Jobname are
not.
You can use all three types of system symbols separately or in combination
within a system symbol string. You can also use all of the system symbol string
737

building facilities such as substring notation, concatenation, embedded blanks,
and so on, in a system symbol string. Thus, system symbol strings can be very
simple like this one:
Sysname,s&SYSNAME
Or more complex like this one:

Dsn1,sACCT.&SYSNAME(3:2)..D&LJDAY
For more information on system symbols, see z/OS MVS Initialization and Tuning
Reference, SA23-1380.
Note: JCL symbols and IPCS symbols are not system symbols and will not be
recognized or replaced in a system symbol string.
s'original_string' will be treated like S'original_string'.
If you want to include a single apostrophe in the system symbol string, you
must specify it as two single apostrophes.
DFSORT uses the ASASYMBM service to transform S'original_string' into
C'result_string'. If your system symbol string has errors involving symbol
names, substring notation, and so on, ASASYMBM transforms the symbol
system string to a character string according to its rules for substitution. For
example, if you specify these symbol statements:
Ok,S&YR4(1:2).&SYSNAME.
Bad,S&YR4(1.2).&SYSNAM.
where the valid substring notation 1:2 and the known system symbol
&SYSNAME. (for example, 'ED53') are used for Ok, but the invalid substring
notation 1.2 and the unknown system symbol &SYSNAM. are used for Bad, the
resulting symbol statements are:
Ok,C20EDS3
Bad,C2006(1.2).&SYSNAM.
See z/OS MVS Programming: Assembler Services Reference ABE-HSP, SA23-1369 for
complete details on the ASASYMBM service.
After a symbol statement containing a system symbol string (S'original_string') is
transformed into a symbol statement containing the resulting character string
(C'result_string'), it will be error checked and processed like any other symbol
statement containing a character string. See the description of character string
previously in this section for details.
If a SYMNOUT data set is specified, it will show the symbol statement
containing the original system symbol string as well as the transformed symbol
statement containing the resulting character string. This can be helpful for
debugging "errors" or unexpected results in your system symbol strings.
v A hexadecimal string in the format X'yy...yy' or x'yy...yy'.
The value yy represents any pair of hexadecimal digits. Each hexadecimal digit
must be 0-9, A-F or a-f. You can specify up to 32 pairs of hexadecimal digits.
x'yy...yy' will be treated like X'yy...yy'. a-f will be treated like A-F.
Some examples of valid hexadecimal strings are: X'F2C3', x'2fa71e', and X'07'.
Some examples of invalid hexadecimal strings are: X'F2G301' (G is not a valid
hexadecimal digit), x'bf3' (unpaired hexadecimal digits) and X'' (no hexadecimal
digits).
v A bit string in the format B'bbbbbbbb...bbbbbbbb' or b'bbbbbbbb...bbbbbbbb'.
The value bbbbbbbb represents 8 bits that constitute a byte. Each bit must be 1,
0 or . (period). You can specify up to 8 groups of 8 bits. b'bbbbbbbb...bbbbbbbb'
will be treated like B'bbbbbbbb...bbbbbbbb'.
738

Some examples of valid bit strings are: B'01100100', b'11..00..000..111' and
B'11......'.
Some examples of invalid bit strings are: B'0101' (not a multiple of eight bits),
b'00..11....' (not a multiple of eight bits), b'00000002' (2 is not a valid bit) and B''
(no bits).
v A decimal number in the format n, +n or -n. You can specify from 1 to 31
significant digits.
Some examples of valid decimal numbers are: +270, 270, 000036, +0 and
-2000000.
Some examples of invalid decimal numbers are: ++15 (too many plus signs),
280- (sign in wrong place) and 2.8 (period is not allowed).
v A two-digit year date string in the format Y'string' or y'string'.
string can be:
yy, yyx,yyxx, yyxxx or yyxxxx where y is a hexadecimal year digit (0-9) and x
is a hexadecimal non-year digit (0-9).
Uppercase, lowercase or mixed case forms of LOW, BLANKS, or HIGH.
A two-digit year date string in the format Y'datex', Y'datex'+n or Y'datex'-n.
- datex can be uppercase, lowercase or mixed case forms of DATE1, DATE2
or DATE3.
- n is a decimal number representing the number of days for DATE1 and
DATE3, or the number of months for DATE2.
Some examples of valid two-digit year date strings are: Y'99', y'00123', y'date2',
Y'DATE1'+20, y'Date3'-1, and Y'Blanks'.
Some examples of invalid two-digit year date strings are: Y'9', y'AB123',
Y'DATE2-3', and Y'blank'.
Field: A field can be specified as p,m,f (position, length and format), p,m (position
and length) or p (position only).
A symbol for a field value must be used only where such a field is allowed and
has the desired result. Otherwise, substitution of the field for the symbol will result
in an error message or unintended processing. For example, if the following
SYMNAMES statement is specified:
Field1,15,2,CH
Field1 can be used in a SORT statement such as:

SORT FIELDS=(Field1,A)
because a character field is allowed in a SORT statement. However, Field1 will

result in an error message if used in a SUM statement such as:
SUM FIELDS=(Field1)
because a character field is not allowed in a SUM statement.

As another example, Field1: can be used in an INREC statement such as:
OUTFIL OVERLAY=(Field1:CAB)
to set the output column to 15, because column notation is allowed in the BUILD,
OVERLAY or PUSH operand of an INREC or OUTREC statement and in the
BUILD, OVERLAY, PUSH, TRLUPD, HEADERx, or TRAILERx operand of an
OUTFIL statement. However, Field1: will result in an error message if used in any
other operand, such as WHEN.
739

Make sure the fields that will be substituted for your symbols are appropriate. If in
doubt, check the rules for p, m, f, and c: given in the description of the relevant
operand.
You can specify p,m,f for your field symbols and then use them where p,m is
required because DFSORT or ICETOOL will substitute just p,m when appropriate.
You can specify p,m,f or p,m for your field symbols and then use them where c: is
required because DFSORT will substitute just p: when appropriate. For example, if
you specify the following in SYMNAMES:
First_Field,12,2,BI
Second_Field,18,6,CH
Third_Field,28,5,PD
Fourth_Field,36,3
Fifth_Field,52,4,PD
Max,200000
Outcol2,16
These DFSORT control statements:

OMIT COND=(Fifth_Field,GT,Max)
SORT FIELDS=(First_Field,A,Fourth_Field,A),FORMAT=CH
SUM FIELDS=(Second_Field,ZD)
OUTFIL OUTREC=(First_Field:First_Field,
Outcol2:Third_Field,M11,
Fourth_Field)

OMIT COND=(52,4,PD,GT,200000)
SUM FIELDS=(18,6,ZD)
OUTFIL BUILD=(12:12,2,16:28,5,PD,M11,36,3)
Note that DFSORT did the following substitutions:

v OMIT statement: p,m,f for Fifth_Field as required by COND without FORMAT.
v SORT statement: p,m for First_Field and Fourth_Field as required by FIELDS
with FORMAT.
v SUM statement: p,m for Second_Field as required for symbol,f (that is,
Second_Field,ZD).
v OUTFIL statement: p: for First_Field: and Outcol2: as required by the OUTREC
operand for an output column. p,m for First_Field and Fourth_Field as required
by the OUTREC operand for an unedited field. p,m,f for Third_Field as required
by the OUTREC operand for an edited field (that is, Third_Field,M11).
The general rules for using p, m and f in symbol statements are as follows:
v p can be a number, an asterisk (*) or an equal sign (=). A number from 1 to
32752 is allowed in p,m or p,m,f. Because p (position only) cannot be
distinguished from the constant n, 1 to 15 significant digits are allowed for p
(position only).
An asterisk (*) can be used to assign the next position to p. Each time a symbol
for p,m,f or p,m is read, the next position is set to p+m. Additionally, the next
position can be modified by keyword statements (see Keyword statements on
page 744). When * is specified for p, the next position is assigned to p. If the next
position has not been set when * is used for p (for example, * is used in the first
symbol), p is set to 1.
The symbol table printed in the SYMNOUT data set (if specified) will show you
the actual positions assigned when you specify * for p.
740

As an example of how * can be used, if you specify the following SYMNAMES
statements:
Sym1,*,5,ZD
Con1,27
Sym2,*,2,BI
Field1,8,13,CH
Field2,*,5,PD
Field3,*,2,FI
SYMNOUT will show the following symbol table:

Sym1,1,5,ZD
Con1,27
Sym2,6,2,BI
Field1,8,13,CH
Field2,21,5,PD
Field3,26,2,FI
By using * for p, you can map consecutive fields in your records without having
to compute their actual positions. You can also map fields added between other
fields without having to change the p values for the original or inserted fields. *
is also useful for creating mappings of contiguous fields using concatenated
symbol data sets. As a simple example, if you specify:
//SYMNAMES DD DSN=MY.SYMPDS(RDW),DISP=SHR
//
DD DSN=MY.SYMPDS(SECTION1),DISP=SHR
//
DD DSN=MY.SYMPDS(SECTION2),DISP=SHR
and the RDW member contains:

RDW,1,4,BI
the SECTION1 member contains:

Flag_Byte,*,1,BI
Error1,X80
Error2,X40
Count_of_Parts,*,5,ZD
and the SECTION2 member contains:

New_Parts,*,5,ZD
Old_Parts,*,5,ZD
Variable_Fields,*

RDW,1,4,BI
Flag_Byte,5,1,BI
Error1,X80
Error2,X40
Count_of_Parts,6,5,ZD
New_Parts,11,5,ZD
Old_Parts,16,5,ZD
Variable_Fields,21
You might use these symbols in the following statements:

OPTION COPY
OUTFIL FNAMES=ERR1,INCLUDE=(Flag_Byte,EQ,Error1),
OUTREC=(RDW,Count_of_Parts,Variable_Fields)
OUTFIL FNAMES=ERR2,INCLUDE=(Flag_Byte,EQ,Error2),
OUTREC=(RDW,New_Parts,Old_Parts,Variable_Fields)
An equal sign (=) can be used to assign the previous position to p. Each time a
symbol for p,m,f or p,m is read, the previous position is set to p. Additionally, the
previous position can be modified by a POSITION keyword statement (see later in
741

this section). When = is specified for p, the previous position is assigned to p. If
the previous position has not been set when = is used for p, an error message is
issued.
the actual positions assigned when you specify = for p.
As an example of how = can be used for p, if you specify the following
SYMNAMES statements:
Sym1,5,4,CH
Sym2,=,2,CH
Sym3,*,2,CH

Sym1,5,4,CH
Sym2,5,2,CH
Sym3,7,2,CH
By using = and * for p, you can easily map fields onto other fields.
Whenever you use = for p, you must ensure that the previous position is the one
you want. In particular, if you insert a new field symbol with the wrong position
before a symbol that uses = for p, you will need to change = to the actual
position you want.
v m can be an equal sign (=) or a number from 1 to 32752. An equal sign (=) can
be used to assign the previous length to m. Each time a symbol for p,m,f or p,m is
read, the previous length is set to m. When = is specified for m, the previous length
is assigned to m. If the previous length has not been set when = is used for m, an
error message is issued.
the actual lengths assigned when you specify = for m.
As an example of how = can be used for m, if you specify the following
Flags1,5,1,BI
Error1,X08
Flags2,15,=,BI
Error2,X04
Flags3,22,=,BI
Error3,X23

Flags1,5,1,BI
Error1,X08
Flags2,15,1,BI
Error2,X04
Flags3,22,1,BI
Error3,X23
Whenever you use = for m, you must ensure that the previous length is the one
you want. In particular, if you insert a new field symbol with the wrong length
before a symbol that uses = for m, you will need to change = to the actual
length you want.
v f can be an equal sign (=) or one of the following formats: AC, AQ, ASL, AST,
BI, CH, CLO, CSF, CSL, CST, CTO, DC1, DC2, DC3, DE1, DE2, DE3, DT1, DT2,
DT3, D1, D2, FI, FL, FS, LS, OL, OT, PD, PD0, SFF, SS, TC1, TC2, TC3, TC4, TE1,
TE2, TE3, TE4, TM1, TM2, TM3, TM4, TS, UFF, Y2B, Y2C, Y2D, Y2DP, Y2P, Y2PP,
Y2S, Y2T, Y2TP, Y2U, Y2UP, Y2V, Y2VP, Y2W, Y2WP, Y2X, Y2XP, Y2Y, Y2YP, Y2Z,
Y4T, Y4U, Y4V, Y4W, Y4X, Y4Y or ZD.
742

You can specify f using uppercase letters (for example, CH), lowercase letters
(for example, ch) or mixed case letters (for example, Ch). f specified in any case
will be treated like uppercase.
An equal sign (=) can be used to assign the previous format to f. Each time a
symbol for p,m,f is read, the previous format is set to f. When = is specified for f,
the previous format is assigned to f. If the previous format has not been set when =
is used for f, an error message is issued.
the actual formats assigned when you specify = for f.
As an example of how = can be used for f, if you specify the following
Field1,5,8,CH
Field1a,=,3
Field2,*,12,=
Field3,*,20,=

Field1,5,8,CH
Field1a,5,3
Field2,8,12,CH
Field3,20,20,CH
Whenever you use = for f, you must ensure that the previous format is the one
you want. In particular, if you insert a new field symbol with the wrong format
before a symbol that uses = for f, you will need to change = to the actual format
you want.
Parsed field: A parsed field can be specified as %nnn with nnn as 000-999, %nn
with nn as 00-99 or as %n with n as 0-9. %0n will be substituted for %n and %00n.
%nn will be substitued for %0nn. A symbol for a parsed field must be used only
where such a field is allowed and has the desired result. Otherwise, substitution of
%nn for the symbol will result in an error message. For example, if the following
SYMNAMES statement is specified:
Revenue,%03
Revenue can be used in an OUTREC statement such as:

OUTREC PARSE=(Revenue=(STARTAT=C+,FIXLEN=9)),
OVERLAY=(31:Revenue,UFF,EDIT=(III,IIT,TTT))
because a parsed field is allowed in the PARSE operand and the OVERLAY
operand. However, Revenue will result in an error message if used in a SORT
statement such as:
SORT FIELDS=(Revenue,UFF,A)
because a parsed field is not allowed in a SORT statement.

Make sure the parsed fields that will be substituted for your symbols are
appropriate. If in doubt, check the rules for %nn given in the description of the
relevant operand.
As an example of how parsed fields can be used, if you specify the following
branch,1,7
tab,X05
comma,,
first_name,%00
last_name,%1
743

date,%02
company,%03
city,%004
state,%305

branch,1,7
tab,X05
comma,C,
first_name,%00
last_name,%01
date,%02
company,%03
city,%04
state,%305
This DFSORT control statement:

INREC PARSE=(first_name=(ABSPOS=9,FIXLEN=12,ENDBEFR=comma),
last_name=(FIXLEN=15,ENDBEFR=comma),
%=(ENDBEFR=tab),
date=(FIXLEN=10,ENDBEFR=tab),
company=(FIXLEN=31,ENDBEFR=tab),
city=(FIXLEN=14,ENDBEFR=tab),
state=(FIXLEN=12)),
BUILD=(branch,9:first_name,22:last_name,
38:date,JFY=(SHIFT=RIGHT),
49:company,83:city,98:state)

INREC PARSE=(%00=(ABSPOS=9,FIXLEN=12,ENDBEFR=C,),%01=(FIXLEN=15,ENDB*
EFR=C,),%=(ENDBEFR=X05),%02=(FIXLEN=10,ENDBEFR=X05*
),%03=(FIXLEN=31,ENDBEFR=X05),%04=(FIXLEN=14,ENDBEFR=X*
05),%05=(FIXLEN=12)),BUILD=(1,7,9:%00,22:%01,38:%02,JF*
Y=(SHIFT=RIGHT),49:%03,83:%04,98:%305)
Keyword statements
The general format for a keyword statement is:
keyword,value remark
The general coding rules are as follows:

v Columns 1 through 80 are scanned.
v The keyword can start in column 1 or in any column after 1.
v The keyword must be specified in all uppercase letters. Otherwise, it will be
treated as a symbol.
v A remark is optional, but if specified, must be separated from the value by at
least one blank. A remark is printed in SYMNOUT (if specified), but otherwise
not processed.
v A semicolon (;) can be used instead of a comma (,) between the keyword and the
value.
v Continuation is not allowed. Each keyword and value must be completely
specified on one line.
The specific syntax for keyword statements is:
744

POSITION,q
POSITION,symbol
SKIP,n
ALIGN,H
ALIGN,F
ALIGN,D
Keyword statements can help you map the fields in your records by letting you set
a starting position, skip unused bytes and align fields on specific boundaries.
v POSITION,q can be used to set the next position and the previous position to q.
As discussed under p previously, the next position is used when an asterisk (*) is
specified for p in a symbol statement, and the previous position is used when an
equal sign (=) is specified for p in a symbol statement. q can be a number from 1
to 32752. When you use POSITION,q you can use either * or = interchangably
for p of the next symbol.
As an example of how POSITION,q can be used, if you specify the following
POSITION,27
Account_Balance,*,5,PD
Account_Id,*,8,CH
POSITION,84
New_Balance,=,20

Account_Balance,27,5,PD
Account_Id,32,8,CH
New_Balance,84,20
v POSITION,symbol can be used to set the next position and the previous position
to the position established for the indicated symbol. As discussed under p
previously, the next position is used when an asterisk (*) is specified for p in a
symbol statement, and the previous position is used when an equal sign (=) is
specified for p in a symbol statement. When you use POSITION,symbol you can
use either * or = interchangeably for p of the next symbol.
symbol can be any previously defined field symbol. Thus, POSITION,symbol
can be used like the Assembler ORG instruction to map different fields onto the
same area.
As an example of how POSITION,symbol can be used, if you specify the
following SYMNAMES statements:
Workarea,21,100
volser1,=,6,CH
volser2,*,6,CH
POSITION,Workarea
status,=,1,BI
dsname,*,44,CH
Use workarea for volsers

Reuse workarea for status and dsname

Workarea,21,100
volser1,21,6,CH
volser2,27,6,CH
status,21,1,BI
dsname,22,44,CH
v SKIP,n can be used to add n bytes to the next position. As discussed under p
previously, the next position is used when an asterisk (*) is specified for p in a
symbol statement. n can be a number from 1 to 32752.
745

As an example of how SKIP,n can be used, if you specify the following
Field#1,15,6,FS
SKIP,4
Unused bytes
Field#2,*,5,=
SKIP,2
Unused bytes
Field#3,*,8,CH

Field#1,15,6,FS
Field#2,25,5,FS
Field#3,32,8,CH
v ALIGN,H can be used to align the next position on a halfword boundary, that is,
1, 3, 5 and so on. As discussed under p previously, the next position is used when
an asterisk (*) is specified for p in a symbol statement. ALIGN,h will be treated
like ALIGN,H.
As an example of how ALIGN,H can be used, if you specify the following
A1,7,3,CH
ALIGN,H
A2,*,2,BI

A1,7,3,CH
A2,11,2,BI
v ALIGN,F can be used to align the next position on a fullword boundary, that is,
1, 5, 9 and so on. As discussed under p previously, the next position is used when
an asterisk (*) is specified for p in a symbol statement. ALIGN,f will be treated
like ALIGN,F.
As an example of how ALIGN,F can be used, if you specify the following
B1,7,3,CH
ALIGN,f
B2,*,4,BI

B1,7,3,CH
B2,13,4,BI
v ALIGN,D can be used to align the next position on a doubleword boundary, that
is, 1, 9, 17 and so on. As discussed under p previously, the next position is used
when an asterisk (*) is specified for p in a symbol statement. ALIGN,d will be
treated like ALIGN,D.
As an example of how ALIGN,D can be used, if you specify the following
C1,7,3,CH
ALIGN,D
C2,*,8,BI

C1,7,3,CH
C2,17,8,BI
746
Using SYMNOUT to check your SYMNAMES statements

To avoid surprises, it's a good idea to check for errors and incorrect positions,
lengths and formats in any SYMNAMES statements you create before you use
them in DFSORT or ICETOOL statements.
The following simple job will cause DFSORT to issue messages in SYSOUT for any
errors it detects in your SYMNAMES statements, allowing you to correct these
errors before proceeding. Once your SYMNAMES statements are free of errors, the
job will cause DFSORT to show the Symbol Table in SYMNOUT, allowing you to
correct any incorrect positions, lengths or formats for your symbols (for example,
those caused by incorrect use of *, =, SKIP, and so on).
//CHECK JOB A402,PROGRAMMER
//DOIT EXEC PGM=ICEMAN
//SYMNAMES DD ...
SYMNAMES statements to be checked
//SORTIN DD *
//SORTOUT DD DUMMY
//SYSIN DD *
OPTION COPY
/*
Once you've "debugged" your SYMNAMES statements, you can use them in
DFSORT and ICETOOL statements.
Using symbols in DFSORT statements

You can use symbols in the following DFSORT control statements: INCLUDE,
INREC, JOINKEYS, MERGE, OMIT, OPTION, OUTFIL, OUTREC, REFORMAT,
SORT and SUM. In general, you can use symbols in these DFSORT statements
where you can use constants ('string', C'string', X'string', B'string', n, +n or -n),
fields (p,m,f or p,m or p), parsed fields (%nn), and output columns (c:). See the
discussion of each control statement in Chapter 3, Using DFSORT program control
statements, on page 81 for a description of its syntax.
You can use symbols in these control statements in any source (that is, DFSPARM,
SYSIN, SORTCNTL, and parameter lists).
When DFSORT transforms these control statements, it removes labels and remarks,
and continues statements by placing an asterisk in column 72 and beginning the
next line in column 16. DFSORT will list the original control statements as
specified (with labels, remarks, comment statements and blank statements) by
source, as well as the transformed statements.
Details and examples of the use of symbols for each applicable DFSORT control
statement is given later in this section. The examples are meant to illustrate
variations in how symbols can be used and how they will be transformed.
Therefore, the examples do not necessarily correspond to how symbols would be
used in real applications.
The examples use these SYMNAMES statements:
C_Field1,6,5,CH
Filler,*
Any_Format,12,3
Z_Field1,22,8,ZD
P_Field1,30,4,PD
C_Field2,4,2,ch
SubString,16,3,SS
Start_col,11
747

Dept_col,55
LIMIT,+12500
Sysplex,S&SYSPLEX.
Depts,J82,L92,M72
Code_1,c86A4Z
QCON,CCarries Constant
Stopper,XFFFFFF
Flags,35,1,BI
Error,B11010000
Empty,B......01
Full,XFF
Lookup,52,1,BI
Entry1,X05
Value1,Read
Entry2,X20
Value2,Update
RDW,1,4
Record Descriptor Word
Variable_Fields,451 Variable fields at end of variable-length record
* Constants for report
Div_Title,Division:
BO_Title,Branch Office
BO_Hyphens,-------------
BO_Equals,=============
PL_Title,
Profit/(Loss)
PL_Hyphens,--------------------
PL_Equals,====================
Total,Total
Lowest,Lowest
* Fields for report
Division,3,10,CH
Branch_Office,16,13,CH
Profit_or_Loss,31,10,ZD
Class_value,%00
Students_value,%01
Class_constant,CLASS=(
Students_constant,STUDENTS=(
End_constant,)
n5,5
n6,6
len80,80
at81,81
len3,3
SORT and MERGE

FIELDS operand: You can use symbols where you can use fields (p,m,f and p,m).
A symbol for p,m,f results in substitution of p,m if FORMAT=f or symbol,f is
specified.
KEYWORD=n operands: You can use a symbol where you can use a number (n)
with the SKIPREC and STOPAFT operands.
Example 1
SORT FIELDS=(C_Field1,A,Z_Field1,D,
C_Field2,ZD,A),EQUALS,SKIPREC=n5

SORT FIELDS=(6,5,CH,A,22,8,ZD,D,4,2,ZD,A),EQUALS,SKIPREC=5
Example 2
MERGE FIELDS=(Any_Format,A,C_Field1,A),FORMAT=CH
748

MERGE FIELDS=(12,3,A,6,5,A),FORMAT=CH
SUM
FIELDS operand: You can use symbols where you can use fields (p,m,f and p,m).
A symbol for p,m,f results in substitution of p,m if FORMAT=f or symbol,f is
specified.
Example 1
SUM FIELDS=(Z_Field1,C_Field1,ZD)

SUM FIELDS=(22,8,ZD,6,5,ZD)
Example 2
SUM FORMAT=ZD,FIELDS=(C_Field1,Any_Format)

SUM FORMAT=ZD,FIELDS=(6,5,12,3)
INCLUDE and OMIT

COND operand: You can use symbols where you can use fields (p1,m1,f1 and
p1,m1 and p2,m2,f2 and p2,m2) and constants (n, +n, -n, C'xx...x', X'yy...yy',
Y'yyx...x' and B'bbbbbbbb...bbbbbbbb'). A symbol for p,m,f results in substitution of
p,m if FORMAT=f or symbol,f is specified. A symbol for 'string' always results in
substitution of C'string'.
Example 1
INCLUDE COND=((Z_Field1,GT,LIMIT,AND,Any_Format,CH,EQ,C_Field2),OR,
(SubString,NE,Depts),OR,
(Flags,ALL,Error,AND,Flags,NE,Empty))

INCLUDE COND=((22,8,ZD,GT,+12500,AND,12,3,CH,EQ,4,2,CH),OR,(16,3,SS,NE*
,CJ82,L92,M72),OR,(35,1,BI,ALL,B11010000,AND,35,1,BI*
,NE,B......01))
Example 2
OMIT FORMAT=BI,COND=(C_Field1,EQ,Code_1,OR,
Any_Format,EQ,Stopper,OR,
Flags,EQ,Full)

OMIT FORMAT=BI,COND=(6,5,EQ,C86A4Z,OR,12,3,EQ,XFFFFFF,OR,35,1,EQ,X*
FF)
Example 3
INCLUDE COND=(25,8,CH,EQ,Sysplex)
If the value for the system symbol &SYSPLEX. is 'MAS3', the INCLUDE statement
INCLUDE COND=(25,8,CH,EQ,CMAS3)
749

Note: You can use a symbol for Y'DATEx', Y'DATEx'+n or Y'DATEx'-n (where x is
1, 2 or 3) in the COND operand, but you cannot use symbol+n or symbol-n to
substitute Y'DATEn'+n or Y'DATEn'-n in the COND operand. For example, if you
have the symbols:
YD1,YDATE1
YD1P5,YDATE1+5
The INCLUDE statement:

INCLUDE COND=(1,6,Y2T,EQ,YD1)

INCLUDE COND=(1,6,Y2T,EQ,YDATE1)

INCLUDE COND=(1,6,Y2T,EQ,YD1P5)

INCLUDE COND=(1,6,Y2T,EQ,YDATE1+5)

INCLUDE COND=(1,6,Y2T,EQ,YD1+5)
will not be transformed because YD1+5 is not valid.
INREC and OUTREC

PARSE operand: You can use symbols where you can use parsed fields (%nn) and
constants (C'xx...x' and X'yy...yy').
FIELDS, BUILD, OVERLAY, FINDREP, IFTHEN BUILD, IFTHEN OVERLAY,
IFTHEN FINDREP and IFTHEN PUSH operands: You can use symbols where
you can use output columns (c:), fields (p,m,f and p,m and p), parsed fields (%nn),
non-repeated constants (C'xx...x' and X'yy...yy', but not nC'xx...x' or nX'yy...yy'), and
decimal constants (+n and -n, but not n). You cannot use symbols for edit patterns
('pattern').
In the JFY and SQZ suboperands, you can use symbols where you can use
In the CHANGE and NOMATCH suboperands, you can use symbols where you
can use a number (n), fields (q,n), parsed fields (%nn), and constants (C'xx...x',
X'yy...yy' and B'bbbbbbbb').
In the RESTART suboperand for SEQNUM, you can use a symbol where you can
use a number (n), a field (p,m) or a parsed field (%nn).
A symbol for p or p,m or p,m,f or p,m,Y2x or p,m,Y2xP or p,m,Y4x results in
substitution of p: for symbol: (output column).
A symbol for p,m,f results in substitution of p,m,f if (symbol),fo or symbol,TO=fo
or symbol,ZDF or symbol,ZDC or symbol,PDC or symbol,PDF is specified, but
results in substitution of p,m if symbol,fo is specified (when fo is not ZDF, ZDC,
PDC, or PDF) because fo cannot be distinguished from f (except for ZDF, ZDC,
PDC, or PDF). For example, if SYM1 is defined as 5,4,ZD, (SYM1),PD is
transformed to (5,4,ZD),PD and SYM1,TO=PD is transformed to 5,4,ZD,TO=PD,
750

whereas SYM1,PD is transformed to 5,4,PD. Thus, you should always use
(symbol),fo or symbol,TO=fo rather than symbol,fo
A symbol for p,m,f results in substitution of p,m unless symbol,edit or symbol,to
or (symbol) is specified or the symbol is part of an arithmetic expression. For
example, if SYM1 is defined as 5,4,ZD, SYM1,X is transformed to 5,4,X, whereas
SYM1,M12 is transformed to 5,4,ZD,M12 and SYM1,ADD,+1 is transformed to
5,4,ZD,ADD,+1.
A symbol for p,m,Y2x results in substitution of p,m,Y2x if symbol,TO=fo or
symbol,todate or symbol,dateop or symbol,ZDF or symbol,ZDC or symbol,PDC or
symbol,PDF is specified, but results in substitution of p,m if symbol,fo is specified
(when fo is not ZDF, ZDC, PDC, or PDF) because fo cannot be distinguished from
f (except for ZDF, ZDC, PDC, or PDF). For example, if SYM1 is defined as 5,4,Y2T,
SYM1,TO=PD is transformed to 5,4,Y2T,TO=PD, whereas SYM1,PD is transformed
to 5,4,PD. Thus, you should always use symbol,TO=fo rather than symbol,fo when
dealing with symbols for p,m,Y2x fields
A symbol for p,m,Y2x results in substitution of p,m,Y2x unless symbol,f or
symbol,HEX is specified. A symbol for p,m,Y2xP results in substitution of
p,m,Y2xP unless symbol,f or symbol,HEX is specified.
symbol,todate or symbol,dateop or symbol,ZDF or symbol,ZDC or symbol,PDC or
dealing with symbols for p,m,Y4x fields.
symbol,HEX is specified.
A symbol for 'string' always results in substitution of C'string'.
Example 1
INREC FIELDS=(Start_col:C_Field2,2X,C_Field1,F,Stopper,5C*,
Z_Field1,Dept_col:Depts,X,P_Field1,TO=FS,X,Z_Field1,M10,
Z_Field1,MUL,(P_Field1,SUB,LIMIT),EDIT=(SIIIIIT.TT),SIGNS=(+,-))

INREC FIELDS=(11:4,2,2X,6,5,F,XFFFFFF,5C*,22,8,55:CJ82,L92,M72,X*
,30,4,PD,TO=FS,X,22,8,ZD,M10,22,8,ZD,MUL,(30,4,PD,SUB,+1*
2500),EDIT=(SIIIIIT.TT),SIGNS=(+,-))
Example 2
OUTREC FIELDS=(RDW, ** Record Descriptor Word **
Z_Field1,2Z,
3CSymbol cannot be used for a repeated constant,
Code_1,Flags,
Variable_Fields) ** Variable part of input record

OUTREC FIELDS=(1,4,22,8,2Z,3CSymbol cannot be used for a repeated con*
stant,C86A4Z,35,1,451)
751

Example 3
INREC PARSE=(Class_value=(STARTAFT=Class_constant,
ENDBEFR=End_constant,FIXLEN=8),
Students_value=(STARTAFT=Students_constant,
ENDBEFR=End_constant,FIXLEN=n5)),
BUILD=(Class_value,3X,Students_value,UFF,M11,LENGTH=n5)

INREC PARSE=(%00=(STARTAFT=CCLASS=(,ENDBEFR=C),FIXLEN=8),%01=(STAR*
TAFT=CSTUDENTS=(,ENDBEFR=C),FIXLEN=5)),BUILD=(%00,3X*
,%01,UFF,M11,LENGTH=5)
IFTHEN WHEN=(logexp), IFTHEN BEGIN=(logexp) and IFTHEN END=(logexp)

operands: You can use symbols where you can use fields (p1,m1,f1 and p1,m1 and
p2,m2,f2 and p2,m2) and constants (n, +n, -n, C'xx...x', X'yy...yy', Y'yyx...x' and
B'bbbbbbbb...bbbbbbbb'). A symbol for p,m,f results in substitution of p,m if
symbol,f is specified. A symbol for 'string' always results in substitution of
C'string'.
Example 1
INREC IFTHEN=(WHEN=(Lookup,EQ,Entry1),OVERLAY=(75:Value1)),
IFTHEN=(WHEN=(Lookup,EQ,Entry2),OVERLAY=(75:Value2))

INREC IFTHEN=(WHEN=(52,1,BI,EQ,X05),OVERLAY=(75:CRead)),IFTHEN=(WH*
EN=(52,1,BI,EQ,X20),OVERLAY=(75:CUpdate))
1, 2 or 3) in the WHEN, BEGIN or END operand, but you cannot use symbol+n or
symbol-n to substitute Y'DATEn'+n or Y'DATEn'-n in the WHEN, BEGIN or END
operand. See the discussion of "INCLUDE and OMIT" for further details.
with the IFOUTLEN, RECORDS, ABSPOS, SUBPOS, ADDPOS, FIXLEN, REPEAT,
STARTPOS, ENDPOS, MAXLEN, DO, ID, SEQ, START, INCR and LENGTH
operands.
Example 1
OUTREC IFOUTLEN=len80,
IFTHEN=(WHEN=GROUP,RECORDS=n5,PUSH=(at81:ID=n6))

OUTREC IFOUTLEN=80,IFTHEN=(WHEN=GROUP,RECORDS=5,PUSH=(81:ID=6))
OUTFIL
INCLUDE, OMIT, IFTRAIL TRLID=(logexp), IFTHEN WHEN=(logexp), IFTHEN
BEGIN=(logexp) and IFTHEN END=(logexp) operands: You can use symbols
where you can use fields (p1,m1,f1 and p1,m1 and p2,m2,f2 and p2,m2) and
constants (n, +n, -n, C'xx...x', X'yy...yy', Y'yyx...x' and B'bbbbbbbb...bbbbbbbb'). A
symbol for p,m,f results in substitution of p,m if symbol,f is specified. A symbol for
'string' always results in substitution of C'string'.
1, 2 or 3) in the INCLUDE, OMIT, TRLID, WHEN, BEGIN or END operand, but
you cannot use symbol+n or symbol-n to substitute Y'DATEn'+n or Y'DATEn'-n in
752

the INCLUDE, OMIT, TRLID, WHEN, BEGIN or END operand. See the discussion
of "INCLUDE and OMIT" for further details.
PARSE operand: You can use symbols where you can use parsed fields (%nn) and
OUTREC, BUILD, OVERLAY, FINDREP, IFTRAIL TRLUPD, IFTHEN BUILD,
IFTHEN OVERLAY, IFTHEN FINDREP and IFTHEN PUSH operands: You can
use symbols where you can use output columns (c:), fields (p,m,f and p,m and p),
parsed fields (%nn), non-repeated constants (C'xx...x' and X'yy...yy', but not
nC'xx...x' or nX'yy...yy'), and decimal constants (+n and -n, but not n). You cannot
use symbols for edit patterns ('pattern').
In the JFY and SQZ suboperands, you can use symbols where you can use
In the CHANGE and NOMATCH suboperands, you can use symbols where you
can use fields (q,n), parsed fields (%nn), and constants (C'xx...x', X'yy...yy' and
B'bbbbbbbb').
In the RESTART suboperand for SEQNUM, you can use a symbol where you can
use a field (p,m) or a parsed field (%nn).
In the IFTHEN KEYBEGIN operand, you can use a symbol where you can use a
field (p,m).
A symbol for p or p,m or p,m,f or p,m,Y2x or p,m,Y2xP results in substitution of p:
for symbol: (output column).
A symbol for p,m,f results in substitution of p,m,f if (symbol),fo or symbol,TO=fo
or symbol,ZDF or symbol,ZDC or symbol,PDC or symbol,PDF is specified, but
results in substitution of p,m if symbol,fo is specified (when fo is not ZDF or ZDC
ZDF, ZDC, PDC, or PDF) because fo cannot be distinguished from f (except for
ZDF, ZDC, PDC, and PDF). For example, if SYM1 is defined as 5,4,ZD, (SYM1),PD
is transformed to (5,4,ZD),PD and SYM1,TO=PD is transformed to 5,4,ZD,TO=PD,
whereas SYM1,PD is transformed to 5,4,PD. Thus, you should always use
(symbol),fo or symbol,TO=fo rather than symbol,fo.
A symbol for p,m,f results in substitution of p,m unless symbol,edit or symbol,to
or (symbol) is specified or the symbol is part of an arithmetic expression. For
example, if SYM1 is defined as 5,4,ZD, SYM1,X is transformed to 5,4,X, whereas
SYM1,M12 is transformed to 5,4,ZD,M12 and SYM1,ADD,+1 is transformed to
5,4,ZD,ADD,+1.
symbol,todate or symbol, dateop or symbol,ZDF or symbol,ZDC or symbol,PDC or
f (except for ZDF, ZDC, PDC, and PDF). For example, if SYM1 is defined as
5,4,Y2T, SYM1,TO=PD is transformed to 5,4,Y2T,TO=PD, whereas SYM1,PD is
transformed to 5,4,PD. Thus, you should always use symbol,TO=fo rather than
symbol,fo when dealing with symbols for p,m,Y2x fields
symbol,HEX is specified. A symbol for p,m,Y2xP results in substitution of
p,m,Y2xP unless symbol,f or symbol,HEX is specified.
753

symbol,todate or symbol, dateop or symbol,ZDF or symbol,ZDC or symbol,PDC or
dealing with symbols for p,m,Y4x fields.
symbol,HEX is specified.
VLFILL operand: You can use symbols where you can use constants (C'x' and
X'yy'). A symbol for 'string' always results in substitution of C'string'.
VLTRIM operand: You can use symbols where you can use constants (C'x' and
X'yy'). A symbol for 'string' always results in substitution of C'string'.
VLTRAIL operand: You can use symbols where you can use constants (C'xx...x'
and X'yy...yy'). A symbol for 'string' always results in substitution of C'string').
HEADER1 and HEADER2 operands: You can use symbols where you can use
output columns (c:), fields (p,m) and non-repeated constants ('xx...x', C'xx...x', and
X'yy...yy', but not n'xx...x', nC'xx...x', or nX'yy...yy'). A symbol for p or p,m or p,m,f
results in substitution of p: for symbol: (output column).A symbol for p,m,f always
results in substitution of p,m. A symbol for 'string' always results in substitution of
C'string'.
TRAILER1 and TRAILER2 operands: Outside of the suboperands TOTAL, TOT,
MIN, MAX, AVG, SUBTOTAL, SUBTOT, SUB, SUBMIN, SUBMAX and SUBAVG:
You can use symbols where you can use output columns (c:), fields (p,m) and
non-repeated constants ('xx...x', C'xx...x' and X'yy...yy', but not n'xx...x', nC'xx...x' or
nX'yy...yy'). A symbol for p or p,m or p,m,f results in substitution of p: for symbol:
(output column). A symbol for p,m,f always results in substitution of p,m. A
symbol for 'string' always results in substitution of C'string'.
Inside of the suboperands TOTAL, TOT, MIN, MAX, AVG, SUBTOTAL, SUBTOT,
SUB, SUBMIN, SUBMAX and SUBAVG: You can use symbols where you can use
fields (p,m,f). A symbol for p,m,f results in substitution of p,m,f if symbol,TO=fo or
symbol,ZDF or symbol,ZDC or symbol,PDC or symbol,PDF is specified, but results
in substitution of p,m if symbol,fo is specified (when fo is not ZDF, ZDC, PDC, or
PDF) because fo cannot be distinguished from f (except for ZDF, ZDC, PDC, and
PDF). For example, if SYM1 is defined as 5,4,ZD, MIN=(SYM1,TO=PD) is
transformed to MIN=(5,4,ZD,TO=PD) whereas MIN=(SYM1,PD) is transformed to
MIN=(5,4,PD). Thus, you should always use symbol,TO=fo rather than symbol,fo.
SECTIONS operand: The "HEADER1 and HEADER2 operands" discussion shown
previously also applies to the HEADER3 suboperand of SECTIONS. The
"TRAILER1 and TRAILER2 operands" discussion also applies to the TRAILER3
suboperand of SECTIONS.
Outside of the HEADER3 and TRAILER3 suboperands, you can use symbols
where you can use fields (p,m). A symbol for p,m,f always results in substitution
of p,m.
754

with the IFOUTLEN, RECORDS, ABSPOS, SUBPOS, ADDPOS, FIXLEN, REPEAT,
STARTPOS, ENDPOS, MAXLEN, DO, ID, SEQ, START, INCR, LENGTH,
STARTREC, ENDREC, SAMPLE, ACCEPT, SPLITBY, SPLIT1R and LINES
operands.
Example 1
OUTFIL FNAMES=OUT1,
INCLUDE=(Z_Field1,GT,LIMIT,AND,Any_Format,CH,EQ,C_Field2),
OUTREC=(Any_Format:P_Field1,M0,2X,Any_Format,BI,LENGTH=len3,2X,
QCON,2X,
C_Field2,HEX,2X,Z_Field1,EDIT=(I III IIT.T),2X,
* Lookup Table
Lookup,CHANGE=(n6,Entry1,Value1,Entry2,Value2),
NOMATCH=(Lookup))

OUTFIL FNAMES=OUT1,INCLUDE=(22,8,ZD,GT,+12500,AND,12,3,CH,EQ,4,2,CH),O*
UTREC=(12:30,4,PD,M0,2X,12,3,BI,LENGTH=3,2X,CCarries *
Constant,2X,4,2,HEX,2X,22,8,ZD,EDIT=(I III IIT.T),2X,*
52,1,CHANGE=(6,X05,CRead,X20,CUpdate),NOMATCH=(5*
2,1))
Example 2
OUTFIL FNAMES=REPORT,
OUTREC=(6:Branch_Office,24:Profit_or_Loss,M5,LENGTH=20,75:X),
SECTIONS=(Division,SKIP=P,
HEADER3=(2:Div_Title,Division,5X,Page:,&PAGE,2/,
6:BO_Title,24:PL_Title,/,
6:BO_Hyphens,24:PL_Hyphens),
TRAILER3=(6:BO_Equals,24:PL_Equals,/,
6:Total,24:TOTAL=(Profit_or_Loss,M5,LENGTH=20),/,
6:Lowest,24:MIN=(Profit_or_Loss,M5,LENGTH=20)))

OUTFIL FNAMES=REPORT,OUTREC=(6:16,13,24:31,10,ZD,M5,LENGTH=20,75:X),SE*
CTIONS=(3,10,SKIP=P,HEADER3=(2:CDivision: ,3,10,5X,P*
age:,&PAGE,2/,6:CBranch Office,24:C
Profit/(Lo*
ss),/,6:C-------------,24:C--------------------),TR*
AILER3=(6:C=============,24:C====================,/,*
6:CTotal,24:TOTAL=(31,10,ZD,M5,LENGTH=20),/,6:CLowest*
,24:MIN=(31,10,ZD,M5,LENGTH=20)))
Example 3
OUTFIL IFTHEN=(WHEN=INIT,
PARSE=(Class_value=(STARTAFT=Class_constant,
ENDBEFR=End_constant,FIXLEN=8),
Students_value=(STARTAFT=Students_constant,
ENDBEFR=End_constant,FIXLEN=n5)),
BUILD=(1:Class_value)),
IFTHEN=(WHEN=(1,8,SS,EQ,CBiology ,Algebra ,Geometry),
BUILD=(CThere are ,Students_value,UFF,M10,X,
Class_value,C Students)),
IFTHEN=(WHEN=NONE,
BUILD=(C*Not relevant*))
755

OUTFIL IFTHEN=(WHEN=INIT,PARSE=(%00=(STARTAFT=CCLASS=(,ENDBEFR=C),*
FIXLEN=8),%01=(STARTAFT=CSTUDENTS=(,ENDBEFR=C),FIXLE*
N=5)),BUILD=(1:%00)),IFTHEN=(WHEN=(1,8,SS,EQ,CBiology ,*
Algebra ,Geometry),BUILD=(CThere are ,%01,UFF,M10,X,%*
00,C Students)),IFTHEN=(WHEN=NONE,BUILD=(C*Not releva*
nt*))
JOINKEYS
FIELDS operand: You can use symbols where you can use fields (p,m). A symbol
for p,m,f results in substitution of p,m.
INCLUDE and OMIT operands: You can use symbols where you can use fields
(p1,m1,f1 and p1,m1 and p2,m2,f2 and p2,m2) and constants (n, +n, -n, C'xx...x',
X'yy...yy', Y'yyx...x' and B'bbbbbbbb...bbbbbbbb'). A symbol for p,m,f results in
substitution of p,m if symbol,f is specified. A symbol for 'string' always results in
substitution of C'string'.
1, 2 or 3) in the INCLUDE or OMIT operand, but you cannot use symbol+n or
symbol-n to substitute Y'DATEn'+n or Y'DATEn'-n in the INCLUDE or OMIT
operand. See the discussion of "INCLUDE and OMIT" for further details.
STOPAFT=n operand: You can use a symbol where you can use a number (n)
with the STOPAFT operand.
Example 1
JOINKEYS FILE=F1,FIELDS=(Division,A),
INCLUDE=(Substring,EQ,Depts)

JOINKEYS FILE=F1,FIELDS=(3,10,A),INCLUDE=(16,3,SS,EQ,CJ82,L92,M72)
REFORMAT
FIELDS operand: You can use symbols where you can use fields (p,m and p). A
symbol for p,m,f results in substitution of p,m.
FILL operand: You can use symbols where you can use constants (C'x' and X'yy').
Example 1
REFORMAT FIELDS=(F1:RDW,Profit_or_Loss,F2:Any_Format,
F1:Variable_Fields),FILL=filler

REFORMAT FIELDS=(F1:1,4,31,10,F2:12,3,F1:451),FILL=C*
OPTION
with the SKIPREC and STOPAFT operands.
Example 1
OPTION SKIPREC=n5,MAINSIZE=MAX,STOPAFT=n6

OPTION SKIPREC=5,MAINSIZE=MAX,STOPAFT=6
756
Using symbols in ICETOOL operators

You can use symbols in the following ICETOOL operators: COUNT, DATASORT,
DISPLAY, OCCUR, RANGE, SELECT, SPLICE, STATS, SUBSET, UNIQUE, and
VERIFY. In general, you can use symbols in these ICETOOL operators where you
can use constants ('string', n, +n or -n) and fields (p,m,f or p,m). See the discussion
of each operator in Chapter 7, Using ICETOOL, on page 563 for a description of
its syntax.
ICETOOL reads the SYMNAMES data set once and uses it for all operators and
DFSORT control statements for the run. You can use symbols in operators from the
TOOLIN data set or your parameter list. You can also use symbols in DFSORT
control statements in xxxxCNTL data sets or in the DFSPARM data set (see Using
symbols in DFSORT statements on page 747 for details).
ICETOOL will list the original operator statements as well as the transformed
operator statements.
Details of the use of symbols for each applicable ICETOOL operator is given later
in this section followed by a complete ICETOOL example. The example is meant to
illustrate variations in how symbols can be used and how they will be
transformed. Therefore, the example does not necessarily correspond to how
symbols would be used in real applications.
COUNT
HIGHER, LOWER, EQUAL and NOTEQUAL operands: You can use symbols
where you can use constants (x, y, v, and w).
SUB and ADD operands: You can use symbols where you can use constants (q
and r).
TEXT operand: You can use symbols where you can use constants ('string'). A
symbol for C'string' always results in substitution of 'string'.
DATASORT
HEADER, FIRST, TRAILER and LAST operands: You can use symbols where you
can use constants (u and v).
DISPLAY
ON operand: You can use symbols where you can use fields (p,m,f and p,m). A
symbol for p,m,f results in substitution of p,m if symbol,f or symbol,HEX is
specified.
BREAK operand: You can use symbols where you can use fields (p,m,f and p,m).
A symbol for p,m,f results in substitution of p,m if symbol,f is specified.
TITLE, HEADER, TOTAL, MAXIMUM, MINIMUM, AVERAGE, COUNT,
BTITLE, BTOTAL, BMAXIMUM, BMINIMUM, BAVERAGE and BCOUNT
operands: You can use symbols where you can use constants ('string'). A symbol
for C'string' always results in substitution of 'string'.
757
OCCUR
symbol for p,m,f results in substitution of p,m if symbol,f or symbol,HEX is
specified.
TITLE and HEADER operands: You can use symbols where you can use constants
('string'). A symbol for C'string' always results in substitution of 'string'.
HIGHER, LOWER and EQUAL operands: You can use symbols where you can
use constants (x, y and v).
RANGE
symbol for p,m,f results in substitution of p,m if symbol,f is specified.
HIGHER, LOWER, EQUAL and NOTEQUAL operands: You can use symbols
where you can use constants (x, y, v and w).
SELECT
FIRST, FIRSTDUP, HIGHER, LOWER and EQUAL operands: You can use
symbols where you can use constants (u, v, w, x, and y).
SPLICE
WITH operand: You can use symbols where you can use fields (p,m). A symbol for
p,m,f results in substitution of p,m.
STATS, UNIQUE and VERIFY

SUBSET
HEADER, FIRST, RRN, TRAILER and LAST operands: You can use symbols
where you can use constants (q, r, u and w).
ICETOOL Example
//TOOLSYM JOB A402,PROGRAMMER
//DOIT EXEC PGM=ICETOOL
//SYMNAMES DD *
Rdw,1,4,BI
Account_Code,12,1,CH
Dept_Code,*,=,=
Customer_Name,*,20,CH
SKIP,2
Customer_Balance,*,10,ZD
Customer_Flags,*,1,BI
* Department Codes
758

Research,R
Marketing,M
* Balance Cutoffs
Cancel,+10000
100.00
Gift,+1000000
10,000.00
Stop_Check,-500
-5.00
* Headings and Titles
Title,SCustomer Report for &LWDAY.
Head1,Customer Name
Head2,Customer Balance
Head3,Customer Flags
/*
//IN DD DSN=MY.CUSTOMER.INPUT,DISP=SHR
//OUT DD DSN=&&O,UNIT=SYSDA,SPACE=(CYL,(5,5),RLSE),
// DISP=(,PASS)
//LIST1 DD SYSOUT=*
//TOOLIN DD *
RANGE FROM(IN) ON(Customer_Balance) LOWER(Stop_Check)
SORT FROM(IN) TO(OUT) USING(CTL1)
DISPLAY FROM(OUT) LIST(LIST1) BLANK WIDTH(133) TITLE(Title) DATE(4MD/) PAGE HEADER(Head1) ON(Customer_Name) HEADER(Head2) ON(Customer_Balance,C1) HEADER(Head3) ON(Customer_Flags,HEX)
/*
//CTL1CNTL DD *
SORT FIELDS=(Customer_Balance,D,Customer_Name,A)
INCLUDE COND=((Dept_Code,EQ,Research,OR,Dept_Code,EQ,Marketing),
AND,Customer_Balance,GT,Gift)
/*
If the value for the system symbol &LWDAY. is 'FRI', SYMNOUT will show the
following:
------- ORIGINAL STATEMENTS FROM SYMNAMES ------Rdw,1,4,BI
Dept_Code,*,=,=
Customer_Name,*,20,CH
SKIP,2
Customer_Balance,*,10,ZD
Customer_Flags,*,1,BI
* Department Codes
Research,R
Marketing,M
* Balance Cutoffs
Cancel,+10000
100.00
Gift,+1000000
10,000.00
Stop_Check,-500
-5.00
* Headings and Titles
Title,SCustomer Report for &LWDAY.
Head1,Customer Name
Head2,Customer Balance
Head3,Customer Flags
------------------ SYMBOL TABLE ----------------Rdw,1,4,BI
Dept_Code,13,1,CH
Customer_Name,14,20,CH
Customer_Balance,36,10,ZD
Customer_Flags,46,1,BI
Research,CR
Marketing,CM
Cancel,+10000
Gift,+1000000
Stop_Check,-500
759

Title,CCustomer
Head1,CCustomer
Head2,CCustomer
Head3,CCustomer
Report for FRI

Name
Balance
Flags
The ICETOOL operators will be transformed to:

RANGE FROM(IN) ON(36,10,ZD) LOWER(-500)
SORT FROM(IN) TO(OUT) USING(CTL1)
DISPLAY FROM(OUT) LIST(LIST1) BLANK WIDTH(133)TITLE(Customer Report for FRI) DATE(4MD/) PAGEHEADER(Customer Name) ON(14,20,CH)HEADER(Customer Balance) ON(36,10,ZD,C1)HEADER(Customer Flags) ON(46,1,HEX)
The DFSORT control statements in CTL1CNTL will be transformed to:

SORT FIELDS=(36,10,ZD,D,14,20,CH,A)
INCLUDE COND=((13,1,CH,EQ,CR,OR,13,1,CH,EQ,CM),AND,36,10,ZD,GT,+10*
00000)
Using SET and PROC symbols in DFSORT and ICETOOL statements

For jobs that directly invoke DFSORT (PGM=SORT or PGM=ICEMAN) or
ICETOOL (PGM=ICETOOL), you can construct DFSORT symbols that incorporate
JCL PROC or SET symbols as well as text and system symbols. You can then use
those constructed DFSORT symbols in DFSORT and ICETOOL control statements
in the same way you use regular DFSORT symbols.
As a simple example, if you wanted to use JCL SET symbols named XDSN and
YDSN in a DFSORT statement, you could code the following:
// SET XDSN=X.DATA,YDSN=Y.DATA
//S1 EXEC PGM=SORT,PARM=MSGDDN=MYOUT,JP1"&XDSN",JP2"&YDSN",LIST
//MYOUT DD SYSOUT=*
...
//SYSIN DD *
OPTION COPY
OMIT COND=(1,44,CH,EQ,JP1,OR,1,44,CH,EQ,JP2)
/*
For each JPn"string" parameter found in EXEC PARM, a DFSORT symbol in the
following form is constructed, and treated as if it was specified in the SYMNAMES
data set (even if a SYMNAMES data set is not actually present):
JPn,Sstring
Up to ten symbols, JP0-JP9, can be set up this way.

For the previous example, the following DFSORT symbols are constructed and
treated as if they were specified in SYMNAMES:
JP1,SX.DATA
JP2,SY.DATA
These symbols are placed in the Symbol Table as:

JP1,CX.DATA
JP2,CY.DATA
When JP1 and JP2 are used in the OMIT statement, the statement is transformed
to:
760

OMIT COND=(1,44,CH,EQ,CX.DATA,OR,1,44,CH,EQ,CY.DATA)
Here is another example using ICETOOL:

// SET VLSR1=339001
// SET VLSR2=339002
//S2 EXEC PGM=ICETOOL,
// PARM=JP0"Report for disks &VLSR1 and &VLSR2 on &WDAY"
...
//TOOLIN DD *
DISPLAY TITLE(JP0) FROM(IN) LIST(RPT) ON(5,4,CH)
The following DFSORT symbol is constructed:

JP0,SReport for disks 339001 and 339002 on &WDAY
If the job is run on Friday, 'FRI' is substituted for the &WDAY system symbol and
the following is placed in the Symbol Table:
JP0,CReport for disks 339001 and 339002 on FRI
When JP0 is used in the DISPLAY statement, the statement is transformed to:
DISPLAY TITLE(Report for disks 339001 and 339002 on FRI)FROM(IN) LIST(OUT) ON(1,5,CH)
Note: For a JOINKEYS application, you can use JCL SET and PROC symbols (JPn)
in DFSORT control statements for the main task, but you cannot use JCL SET and
PROC symbols (JPn) in DFSORT control statements for the subtasks.
Using JPn parameters in EXEC PARM for DFSORT

//stepname EXEC PGM=SORT,PARM=option<,option>...
option can be JPn"string" or one of the other valid DFSORT options (for example,
VLSCMP or DSA=8). JPn"string" options are used to construct DFSORT symbols.
Other DFSORT options are passed to DFSORT as EXEC PARM options in the
normal way.
PGM=ICEMAN or one of the other aliases can be used instead of PGM=SORT.
Normal system JCL rules for the EXEC PARM operand apply. In addition:
v JPn"string" parameters can only be used when DFSORT is invoked directly (for
example, with PGM=SORT), not when DFSORT is invoked from a program (for
example, with LINK EP=SORT).
v If the length of the EXEC PARM options is greater than the JCL limit of 100
bytes, JPn"string" options will not be processed.
Using JPn parameters in EXEC PARM for ICETOOL

//stepname EXEC PGM=ICETOOL,PARM=option<,option>...
option can be JPn"string". JPn"string" options are used to construct DFSORT

symbols. Other options are ignored by ICETOOL.
Normal system JCL rules for the EXEC PARM operand apply. In addition:
v JPn"string" parameters can only be used when ICETOOL is invoked directly (for
example, with PGM=ICETOOL), not when ICETOOL is invoked from a program
(for example, with LINK EP=ICETOOL).
761

v If the length of the EXEC PARM options is greater than the JCL limit of 100
bytes, JPn"string" options will not be processed.
Description of JPn"string"
Only JP0 through JP9 can be used for a symbol name where 'JP' must be uppercase
EBCDIC (X'D1D7') and n must be '0'-'9' (X'F0'-X'F9').
A quote must appear before and after the string. A quote must not appear within
the string (it would be interpreted as the ending quote). PARM='...,JPn"string' (a
missing ending quote) will be interpreted as PARM='...,JPn"string"'.
If the keyword does not appear as JPn"string", it will not be used as a JPn symbol.
For example, JP1"ABC" would be used as a JP1 symbol, but JP1="ABC",
JP1("ABC"), JP1=("ABC") or jp1"ABC" would not.
Any or all of the following can appear in the string:
v any EBCDIC character (except quote). If you want to include a single apostrophe
in the character string, you must specify it as two single apostrophes
v a JCL SET or PROC symbol (&SYMBOL). Normal system rules for SET and
PROC symbols apply.
v a system symbol (for example, &JOBNAME, &SYSPLEX, etc). Normal system
rules for system symbols apply.
The symbol will be constructed as:
JPn,Sstring
and must conform to the rules for a system symbol string as documented earlier in
this Chapter.
If you specify:
the JPn symbols will be listed along with any other DFSORT symbols you specify
in SYMNAMES.
If you specify a SYMNAMES data set, the JPn symbols will be processed before
the first SYMNAMES symbol.
Note: Here is an example of one suggested way of specifying multiple options in
EXEC PARM:
// SET X1=ANY-STRING
// SET X2=ANOTHER-STRING
//S1 EXEC PGM=SORT,
// PARM=(RESALL=12K,
//
VLLONG,
//
JP1"&X1",
//
JP2"&X2")
Example 1
MYPROC procedure
//MYPROC PROC DAY=
//STEP01 EXEC PGM=SORT,PARM=JP1"&DAY"
762

//SORTIN DD DSN=MYCNTL.PDS(SEL),DISP=SHR
//SYSIN DD DSN=MYCNTL.PDS(OVLY),DISP=SHR
// PEND
Execution of MYPROC
//S1 EXEC MYPROC,DAY=2
Record in SEL member

SELECT * FROM MYTABLE WHERE TM_RECEIPT >= (CURRENT DATE - ? DAYS)
Records in OVLY member

OPTION COPY
INREC OVERLAY=(59:JP1)
This example illustrates how you can use a JCL PROC symbol in a DFSORT
control statement.
The JCL PROC symbol is DAY which will be set to a 1 character numeric value (2
in this example), and used to overlay the number of days in the SELECT statement
(in the SEL member).
PARM='JP1"&DAY"' is used to create a DFSORT symbol named JP1 containing a
string with the DAY value.
The SYMNOUT listing will show the following:
------ SYMNAMES STATEMENTS FROM EXEC PARM ------JP1,S2
------------------ SYMBOL TABLE ----------------JP1,C2
The DFSORT INREC statement in the OVLY member uses JP1. It will be
transformed to:
INREC OVERLAY=(59:C2)
and used to change the SELECT statement to the following in SORTOUT:

SELECT * FROM MYTABLE WHERE TM_RECEIPT >= (CURRENT DATE - 2 DAYS)
Example 2
// SET JOBNM=FRANK2
//S1
EXEC PGM=ICETOOL,
// PARM=(JP1"&JOBNM",
//
JP2"&JOBNM STEP",
//
JP3"&JOBNM EXCPS")
//SYMNAMES DD *
JOBN,5,8,CH
STEPN,*,8,CH
EXCPS,*,5,ZD
//IN DD *
FRANK2 S1
00008
FRANK2 S2
00123
FRANK2 S3
00023
FRANK3 S1
00016
FRANK3 S2
00152
/*
//T1 DD DSN=&&T1,UNIT=SYSDA,SPACE=(CYL,(5,5)),DISP=(,PASS)
763

//RPT DD SYSOUT=*
//TOOLIN DD *
COPY FROM(IN) TO(T1) USING(CTL1)
DISPLAY FROM(T1) LIST(RPT) BLANK HEADER(JP2) ON(STEPN) HEADER(JP3) ON(EXCPS)
/*
//CTL1CNTL DD *
INCLUDE COND=(JOBN,EQ,JP1)
/*
This example illustrates how you can use a JCL SET symbol in ICETOOL and
DFSORT control statements.
The JCL SET symbol is JOBNM which is set to a constant ('FRANK2' for this
example).
//
//
//
PARM=(JP1"&JOBNM",
JP2"&JOBNM STEP",
JP3"&JOBNM EXCPS")
is used to create DFSORT symbols named JP1, JP2 and JP3 containing strings
including the JOBNM value. ICETOOL treats these JPn symbols as if they were
specified before the Symbols in SYMNAMES.
The SYMNOUT listing will show the following:
------ SYMNAMES STATEMENTS FROM EXEC PARM ------JP1,SFRANK2
JP2,SFRANK2 STEP
JP3,SFRANK2 EXCPS
------- ORIGINAL STATEMENTS FROM SYMNAMES ------JOBN,5,8,CH
STEPN,*,8,CH
EXCPS,*,5,ZD
------------------ SYMBOL TABLE ----------------JP1,CFRANK2
JP2,CFRANK2 STEP
JP3,CFRANK2 EXCPS
JOBN,5,8,CH
STEPN,13,8,CH
EXCPS,21,5,ZD
The DFSORT INCLUDE statement in CTL1CNTL uses the JOBN and JP1 symbols.
It will be transformed to:
INCLUDE COND=(5,8,CH,EQ,CFRANK2)
The ICETOOL DISPLAY statement uses the JP2, JP3, STEPN and EXCPS symbols.
It will be transformed to:
DISPLAY FROM(T1) LIST(RPT) BLANKHEADER(FRANK2 STEP) ON(13,8,CH)HEADER(FRANK2 EXCPS) ON(21,5,ZD)
RPT will contain the following report:

FRANK2 STEP
----------S1
S2
S3
764
FRANK2 EXCPS
-----------8
123
23
Notes for symbols

v EFS programs cannot be used with symbol processing.
v DFSORT or ICETOOL scans each SYMNAMES statement for errors, and prints
an error message for the first error detected. A marker ($) is printed directly
below the SYMNAMES statement near the error, if appropriate.
Scanning stops at the first error, and then continues with the next SYMNAMES
statement. However, once an error is detected, positions generated by using an
asterisk (*) for p or POSITION,symbol in subsequent statements will not be
checked for errors. DFSORT and ICETOOL terminate after all SYMNAMES
statements are scanned if an error is detected in any statement.
v If DFSORT or ICETOOL detects an error in a control statement or operator
statement during the substitution phase (that is, while attempting to substitute
values for symbols), it may either:
print the original statement in error followed by a $ marker (if appropriate)
and an error message, continue the substitution phase with the next statement
and terminate when the substitution phase is complete, or
stop performing substitution for the statement in error, continue with the next
statement and let processing after the substitution phase handle the error. It is
possible in this case for a symbol, rather than a substituted value, to appear
in a transformed statement.
v If the substitution phase is successful, DFSORT and ICETOOL will substitute
values for symbols wherever symbols are allowed. Substituted values which are
invalid for a particular statement or operand will be detected after the
substitution phase. This makes it easier to determine the cause of the error. For
example, if SYMNAMES contains:
Sym1,5,4,ZD
Con1,1234
Con2,1234
the statement:
INCLUDE COND=(Sym1,EQ,Con1)
will be transformed to the following during the substitution phase:

INCLUDE COND=(5,4,ZD,EQ,C1234)
An ICE114A message with a $ marker under C'1234' will then be issued for the
statement because a ZD field cannot be compared to a character constant. In this
example, the error could be fixed by using Con2 (a decimal constant) in the
statement instead of Con1 or by redefining Con1 as a decimal constant.
v If you use a temporary or permanent message data set, it is best to specify a
disposition of MOD to ensure you see all messages and control statements in the
message data set. In particular, if you use symbols processing and do not use
MOD, you will not see the original control statements unless Blockset is selected.
v If you rearrange your records in any way (for example, using E15, E35, INREC,
OUTREC or OUTFIL) and want to use symbols for the rearranged records, be
sure to use symbols that map to the new positions of your fields. For example, if
you use a SYMNAMES data set with the following statements:
Field1,1,5,ZD
Field2,*,6,ZD
Field3,*,3,ZD
Field4,*,4,ZD
for this INREC statement:

765

INREC FIELDS=(Field2,Field4)
the resulting records will only contain Field2 and Field4. If you want to use
symbols for the rearranged records (for example, in a SORT statement), you will
need to use a SYMNAMES data set with symbols that map to the rearranged
records, such as:
New_Field2,1,6,ZD
New_Field4,*,4,ZD
If you use unique symbols for the rearranged fields, as in the previous example,
you can concatenate the old and new symbol data sets together and use the old
and new symbols where appropriate, as in this example:
INREC FIELDS=(Field2,Field4)
SORT FIELDS=(New_Field2,A,New_Field4,A)
766
Chapter 9. Using extended function support

Using EFS
Like the user exits described in Chapter 5, Using your own user exit routines, on
page 487, the DFSORT Extended Function Support (EFS) interface is a means by
which you can pass run-time control to an EFS program you write yourself. An
EFS program is essential if you want to process double-byte character sets (such as
Japanese characters) with DFSORT.
To process Japanese data types with DFSORT, you can use the IBM Double Byte
Character Set Ordering Support Program (DBCS Ordering), Licensed Program
5665-360, Release 2.0, or you can use locale processing with the appropriate locale.
Using an EFS program and EFS program exit routines, you can:
v Sort or merge user-defined data types (such as double-byte character sets) with
user-defined collating sequences
v Include or omit records based on the user-defined data types
v Provide user-written messages to DFSORT for printing to the message data set
v Examine, alter, or ignore control statements or EXEC PARM options prior to
processing by DFSORT.
The EFS program can also perform routine tasks, such as opening and initializing
data sets, terminating DFSORT, and closing data sets.
You can write your EFS program in any language that uses standard register and
linkage conventions, and can:
v Pass a parameter list and a record (if you provide the EFS01 and EFS02 exit
routines in the EFS program) in register 1
v Pass a return code in general register 15.
Note:
1. DFSORT does not support EFS programs for Conventional merge or tape work
data set sort applications.
2. VLSHRT is not allowed if EFS processing is in effect and an EFS01 or EFS02
exit routine is provided by the EFS program.
3. If you use locale processing for SORT, MERGE, INCLUDE, or OMIT fields, you
must not use an EFS program. DFSORT's locale processing may eliminate the
need for an EFS program. See OPTION control statement on page 173 for
information related to locale processing.
4. If you use symbol processing, you must not use an EFS program.
The DFSORT target library, SICEUSER, contains a mapping macro called ICEDEFS,
which provides a separate Assembler DSECT for the EFS parameter list.
767
Addressing and Residence Mode of the EFS Program
Addressing and residence mode of the EFS program

You can design the EFS program to reside and run above or below 16MB virtual.
Residency and addressing mode can be any valid combination of 24-bit, 31-bit, or
ANY. If your EFS program is designed to reside and run below 16MB virtual, the
EFS program must determine the proper return mode.
How EFS works

The EFS interface consists of a variable-length parameter list used to communicate
between DFSORT and your EFS program. DFSORT activates the EFS program you
write at specific points during run-time, and communicates information back and
forth across the interface as your EFS program runs.
You can activate your EFS program with the installation or run-time option
EFS=name (name is the name of your EFS program):
See Appendix B, Specification/override of DFSORT options, on page 863 for
override information. Figure 35 illustrates the role of the EFS interface in linking
DFSORT's processing capabilities to the EFS program you write.
DFSORT and Non-DFSORT

Control Statements
and EXEC PARM Options
DFSORT
EFS Program
EFS
Interface
Processing
Input
Data
Set
EFS
Interface
EFS
Interface
Processing
Output
Data
Set
Figure 35. Relationship Between DFSORT and an EFS Program
DFSORT program phases

A DFSORT program phase is a large DFSORT component designed to perform a
specific task such as writing the output file. An EFS program is called at various
points during run-time of DFSORT program phases in performing the variety of
tasks capable with an EFS program. When the termination phase is completed,
DFSORT returns control to the operating system or invoking program.
EFS processing can be invoked during the initialization, input, and termination
phases of DFSORT. DFSORT always calls the EFS program during the initialization
phase.
768
How EFS Works

During the input phase, DFSORT reads input records, and performs any INCLUDE
or OMIT statement logic on the records. If the EFS program generates exit routines
(EFS01 and EFS02), DFSORT calls them during the input phase.
During the termination phase, DFSORT closes data sets, releases storage, and
returns control to the calling program or system. DFSORT always calls the EFS
program from the termination phase.
DFSORT calls to your EFS program

DFSORT makes five functional calls (Major Calls 1 through 5) at various phases to
transfer information across the EFS interface, between DFSORT and your EFS
program. DFSORT can make multiple calls at Major Calls 2 and 3. Refer to
Figure 36 and Figure 37 on page 770 as you read this section for illustrations of the
relationships between program phases and calls during run-time.
Initialization Phase
DFSORT
Major Call 1
EFS
Processing
Major Call 2
EFS
Program
Major Call 3
DFSORT
Input Phase
EFS01 and
EFS02
Parameter
List
Processing
Call for each

Input Record
EFS01
One or more calls for each

EFS02
Input Record
DFSORT
Termination Phase
Major Call 4
EFS
Processing
Major Call 5
EFS
Program
Figure 36. EFS Program Calls for a Sort. The figure also shows the calls to the EFS program
EFS01 and EFS02 exit routines.
769
How EFS Works
DFSORT
Major Call 1
EFS
Processing
Major Call 2
EFS
Program
Major Call 3
DFSORT
Input and
Output Phase
EFS01 and
EFS02
Parameter
List
Processing
Call for each

Input Record
EFS01
(Merge
Only)
One or more calls for each

EFS02
Input Record
DFSORT
Termination Phase
Major Call 4
EFS
Processing
Major Call 5
EFS
Program
Figure 37. EFS Program Calls for a Merge or Copy. The figure also shows the calls to the
EFS program EFS01 and EFS02 exit routines.
Initialization phase
DFSORT runs Major Calls 1 through 3 during the initialization phase.
Major call 1: The EFS program can perform initialization processing such as
opening data sets and obtaining storage.
Information is passed in both directions between DFSORT and the EFS program
across the EFS interface.
At Major Call 1, DFSORT supplies your EFS program with fields in the EFS
interface containing:
v An action code indicating that Major Call 1 is in effect
v Informational flags that describe current processing.
770
How EFS Works

When the EFS program returns control to DFSORT, it can supply fields in the EFS
v A control statement request list, with a list of DFSORT and non-DFSORT control
statement operation definers, or EXEC PARM options
Note: OUTFIL statements cannot be requested by an EFS program.
v An EFS Program Context area (a private communication area for the EFS
program)
v A list containing messages for printing to the message data set
v A return code (in general register 15).
Major call 2: At this call, your EFS program can examine, alter, or ignore control
statements before DFSORT processes them, and provide user-written messages to
the message data set. DFSORT calls your EFS program once for each control
statement or EXEC PARM you request.
v The original control statement or EXEC PARM option requested by the EFS
program
v The length of the original control statement or EXEC PARM option
v Informational flags that describe current processing
v An EFS Program Context area (a private communication area for the EFS
program).
v A modified version of the control statement or EXEC PARM option sent by
DFSORT to the EFS program. If you plan to sort or merge user-defined data
types, or include or omit user-defined data types, your EFS program must return
new formats for the SORT/MERGE or INCLUDE/OMIT control statements.
These new formats (D1 and D2) signal DFSORT to call the EFS01 and EFS02 exit
routines you included with your EFS program.
v
v
v
v
Note: OUTFIL statements cannot be passed to an EFS program or returned

from an EFS program to be parsed.
The length of the altered control statement or EXEC PARM option.
Informational flags signaling DFSORT whether to parse or ignore the control
statement or EXEC PARM option.
A list of messages for DFSORT to print to the message data set.
A return code (in general register 15).
Major call 3: At Major Call 3, your EFS program can provide DFSORT with
user-written messages to print to the message data set. DFSORT can call the EFS
program once for the Blockset technique and once for the Peerage/Vale techniques.
DFSORT obtains more information at this call from the EFS program to process the
EFS01 and EFS02 exit routines.
771
How EFS Works

v
v
v
v
An extract buffer offsets list needed by the EFS01 exit routine

A record lengths list of input and output records
Informational flags that describe current processing
An EFS Program Context area (a private communication area for the EFS
program).
v An EFS01 exit routine address
v An EFS02 exit routine address
v A list of messages for printing to the message data set
v A return code in general register 15.
Input phase
DFSORT runs the two exit routines, EFS01 and EFS02, during the input phase. The
EFS01 routine supports sorting or merging user-defined data types with
user-defined collating sequences and is called once for each record. The EFS02
routine provides logic to include or omit records on user-defined data types and is
called one or more times for each record, according to the logic.
Information is passed in both directions between DFSORT and the exit routines
across the EFS01 and EFS02 parameter lists.
DFSORT supplies the EFS01 routine with fields in the parameter list containing:
v An Extract Buffer Area to which the EFS01 routine must move all EFS control
fields. See EFS01 user exit routine on page 789 for more information.
v The input data record.
v An EFS Program Context Area (a private communication area for the EFS
program).
When the EFS01 routine returns control to DFSORT, it must return a return code in
general register 15.
DFSORT supplies the EFS02 routine with fields in the parameter list containing:
v A Correlator Identifier, which identifies a relational condition containing EFS
fields. See EFS02 user exit routine on page 790 for more information.
v The input data record.
When the EFS02 routine returns control to DFSORT, it must return a return code in
Termination phase
DFSORT runs Major Calls 4 and 5 during the termination phase. Only one call is
made at each of these Major Calls.
Note: If a system abend occurs while DFSORT's ESTAE recovery routine is in
effect, and Major Calls 4 and 5 have not already been run, the ESTAE routine runs
them. If an EFS abend occurs during Major Call 1, the ESTAE routine does not run
Major Calls 4 and 5. See Appendix E, DFSORT abend processing, on page 907 for
more information about ESTAE.
Major call 4: The EFS program provides any final user-written messages for
printing to the message data set.
772
How EFS Works

v An action code indicating that Major Call 4 is in effect.
program).
v A message list containing messages for printing to the message data set.
v A return code (in general register 15).
Major call 5: The EFS program performs any termination processing, such as
closing data sets and releasing storage.
v An action code indicating that Major Call 5 is in effect.
program).
When the EFS program returns control to DFSORT, it must supply a return code in
What you can do with EFS

You can design your EFS program to perform seven basic tasks at the initialization,
input, and termination phases of DFSORT. Some of the tasks require using the EFS
program-generated user exit routines EFS01 and EFS02.
Table 87. Functions of an Extended Function Support (EFS) Program
Opening and
initializing
Input Phase
Termination Phase
EFS Program
Examining, altering,
EFS Program
or ignoring DFSORT
and non-DFSORT
control statements
prior to processing by
DFSORT
Sorting or merging
user-defined data
types with
user-defined collating
sequences
EFS01
Providing the logic to

include or omit
records based on
user-defined data
types
EFS02
Supplying messages
to DFSORT for
printing to the
message data set
EFS Program
Terminating DFSORT
EFS Program
EFS Program
EFS01, EFS02
EFS Program
773
What You Can Do with EFS

Table 87. Functions of an Extended Function Support (EFS) Program (continued)
Closing data sets and
housekeeping
Input Phase
Termination Phase
EFS Program
Opening and initializing data sets

Your EFS program can open data sets, obtain necessary storage, and perform other
forms of initialization needed during a run.
Examining, altering, or ignoring control statements

At Major Call 1, your EFS program can send a control statement request list to
indicate the control statements and EXEC PARM options you want DFSORT to
send to your EFS program at Major Call 2. OUTFIL statements cannot be requested
by an EFS program.
At Major Call 2, your EFS program can examine, alter, or ignore control statements
and EXEC PARM options that DFSORT reads from the EXEC statement, SYSIN,
SORTCNTL, DFSPARM, or a parameter list passed from an invoking program.
OUTFIL statements cannot be passed to an EFS program or returned from an EFS
program to be parsed.
Refer to Figure 38 on page 775 for an illustration of the control statement
processing sequence used when an EFS program is activated.
The same override rules apply to control statements and parameters returned from
an EFS program as apply to the original control statements and parameters.
For example, a STOPAFT parameter added to the SORT statement by an EFS
program is overridden by a STOPAFT parameter in an OPTION statement in the
same way as if the SORT statement originally contained the STOPAFT parameter.
See Appendix B, Specification/override of DFSORT options, on page 863 for full
override details.
774
Direct Invocation of DFSORT
Product Invocation of DFSORT
DFSPARM
DFSORT processing of DFSPARM

parameters not requested by
the EFS program
EFS program processing of

requested DFSPARM parameters
at Major Call 2
DFSORT processing of DFSPARM

parameters returned by the
EFS program
JCL EXEC statement

PARM options
SYSIN
SORTCNTL
DFSORT processing of
SYSIN control statements
and JCL EXEC statement
PARM options not requested
by the EFS program
Same processing as for

DFSPARM
EFS program processing of

requested SYSIN control
statements and EXEC
PARM options at Major
Call 2
Invokers parameter list
DFSORT processing of
SYSIN control statements
and JCL EXEC statement
PARM options returned by
the EFS program
Same processing as for

DFSPARM
SYSOUT
Figure 38. Control Statement Processing Sequence
775
Processing user-defined data types with EFS program user

exit routines
You can write your EFS program to provide two user exit routines to perform
various tasks during run-time.
Your EFS program user exit routines can:
v Process user-defined data types. Your EFS program can provide an EFS01
routine to alter any control field of an input record.
v Include or omit records based on user-defined data types. Your EFS program can
provide an exit routine to examine any input field of an input record to
determine whether or not to include that record for processing.
Supplying messages for printing to the message data set

You can use an EFS program to tailor messages for several purposes:
v To describe new types of operations
v To describe extended field parameters
v To customize the message data set to your site
v To display statistical information about control statements or EXEC PARM
options.
You can control whether to print the control statements returned by an EFS
program to the message data set with:
v
v The LISTX installation option (see Installation defaults on page 17)
v The LISTX or NOLISTX operators in the PARM field of the JCL EXEC statement
(see Specifying EXEC/DFSPARM PARM options on page 32)
v The LIST or NOLIST operators of the OPTION program control statement.
Terminating DFSORT
Your EFS program can terminate DFSORT at any of the five Major Calls and also
from either of the two EFS program exit routines during the input phase.
Closing data sets and housekeeping

At Major Call 5, your EFS program can close data sets, free storage and perform
any other necessary housekeeping.
Structure of the EFS interface parameter list

The EFS interface consists of a variable-length parameter list and is used to
communicate between DFSORT and your EFS program. DFSORT initializes the
parameter list to zeros during the initialization phase, except that the list end
indicator is set to X'FFFFFFFF'.
The parameter list resides below 16MB virtual, and remains accessible while the
EFS program is active, although DFSORT might change its storage location during
run-time to optimize use of storage. The actual address in register 1 (used to pass
the interface parameter list address) can therefore change while DFSORT is
running.
Figure 39 on page 778 illustrates the structure of the EFS interface parameter list.
The illustrated portions of the list are explained in order in the following pages.
776
Structure of the EFS Interface Parameter List

EXEC PARMs are not described in the figure, but are included in processing.
777

R1
Action code
4 bytes
Address of
Control Statement list
4 bytes
Address of
original Control Statement
including all keywords and
corresponding subparameters
Control statement
request list
** bytes
Original control
statement string
* bytes
4 bytes
Address of
modified Control Statement
4 bytes
Length of
original Control Statement
4 bytes
Length of
modified Control Statement
4 bytes
Address of
EFS context area
4 bytes
778
Address of
Extract buffer offsets
(zeros if no EFS fields exist)
4 bytes
Address of
Record lengths list
4 bytes
RESERVED
4 bytes
RESERVED
4 bytes
RESERVED
4 bytes
Information flags
4 bytes
Address of
message list
(zeros if none)
4 bytes
RESERVED
4 bytes
RESERVED
4 bytes
RESERVED
4 bytes
Address of
f
EFS01 extract routine
(zeros if none)
4 bytes
Address of
EFS02 INCLUDE/OMIT
f
routine (zeros if none)
4 bytes
List end indicator (X'FFFFFFFF')
Modified control
statement string
* bytes
Record lengths
list
8 bytes
4 bytes
** - Length determined by length fields in the list

* - Length determined by corresponding length field
Action codes
DFSORT sets one of five action codes before a call to the EFS program:
0
Indicates Major Call 1 to the EFS program. DFSORT sends this action code
once.
Indicates Major Call 2 to the EFS program. DFSORT might send this action
code several times at Major Call 2 depending on how many control
statements are requested and found. For example, if the SORT, MERGE,
and INCLUDE control statements are all supplied in SYSIN and are
requested, the EFS program is called twice: once for the SORT control
statement (because SORT and MERGE are mutually exclusive, and
assuming the SORT statement is specified first, only the SORT statement is
taken) and once for the INCLUDE control statement.
Indicates Major Call 3 to the EFS program. DFSORT can send this action
code once for the Blockset technique and once for the Peerage/Vale
technique.
12
once.
16
once.
Control statement request list

The control statement request list describes the control statements and the PARM
options to be sent to the EFS program by DFSORT. The control statement request
list consists of control statement operation definers and PARM option names. The
maximum length allowed for an operation definer or PARM option name is eight
bytes. If the operation definer or PARM option name is longer, DFSORT will use
only the first eight bytes. Length field values must not include their own length.
OUTFIL statements cannot be requested by an EFS program.
Non-DFSORT operation definers and PARM options must be in EBCDIC format,
and the first character must be non-numeric. The format of the control statement
request list is:
Chain pointer to
next operation
definer or EXEC
PARM option name,
or zero for end
of list
4 bytes
Length of
operation definer
or EXEC PARM
option name
Operation definer
or EXEC PARM
option name
(variable-length)
2 bytes
* bytes
The asterisk (*) indicates that the length is determined by the corresponding length
field (maximum of 8 bytes).
Control statement string sent to the EFS program

DFSORT scans for the requested control statement from SYSIN, SORTCNTL,
DFSPARM, or the invoker's parameter list to create a contiguous control statement
string; DFSORT will handle any necessary continuation requirements for control
statements from SYSIN, SORTCNTL, or DFSPARM. DFSORT scans for the
requested PARM option to create a contiguous PARM option string.
779

DFSORT places a copy of the requested control statement or PARM option string
in a contiguous storage area for the EFS program. No labels are supplied with the
control statement; the address of the string always points to the first byte of the
appropriate operation definer or PARM option.
DFSORT will send the requested control statement(s) or PARM option(s) to the EFS
program as found by DFSORT; DFSORT will provide limited syntax checking of
control statements or PARM option(s) before sending them to the EFS program.
In addition to following the rules in General coding rules on page 83, you must
observe the following rules for non-DFSORT control statements:
v DFSORT will recognize a control statement with no operand(s) provided the
operation definer (1) is supplied in SYSIN, SORTCNTL, or DFSPARM and (2) is
the only operation definer contained on a line.
v Operation definers supplied through SYSIN, SORTCNTL, DFSPARM, or the
extended parameter list and requested by the EFS program will not be
recognized if they are longer than eight bytes.
In addition to observing the rules in z/OS MVS JCL User's Guide and z/OS MVS
JCL Reference you must observe the following rule for non-DFSORT PARM options:
v PARM options requested by the EFS program will not be recognized if they are
longer than eight bytes.
DFSORT will send the requested DFSORT or non-DFSORT control statements or
PARM options that remain after DFSORT override rules have been applied.
v
If duplicate DFSORT or non-DFSORT control statements or PARM options are
supplied through the same source (such as SYSIN), then DFSORT will send the
first occurrence of the control statement. The second occurrence of the DFSORT
or non-DFSORT control statement or PARM option will be ignored by DFSORT.
If duplicate DFSORT or non-DFSORT control statements are supplied through
different sources (such as extended parameter list, SORTCNTL, and DFSPARM),
then DFSORT will send the control statement remaining after different source
override rules have been applied, except for the DFSORT OPTION and DEBUG
control statements (see Special handling of OPTION and DEBUG control
statements on page 781).
If mutually exclusive DFSORT control statements (such as SORT/MERGE) are
supplied through the same source (such as SYSIN), then DFSORT will send the
first occurrence of the control statement. The second occurrence of the DFSORT
control statement will be ignored by DFSORT.
If mutually exclusive DFSORT control statements (such as SORT/MERGE) are
supplied through different sources (such as extended parameter list, SORTCNTL,
and DFSPARM), then DFSORT will send the control statement remaining after
different source override rules have been applied. The DFSORT control
statement not sent will be ignored by DFSORT.
Thus the EFS program will not be sent duplicate DFSORT or non-DFSORT control
statements (except for the DFSORT OPTION and DEBUG control statements as
explained in Special handling of OPTION and DEBUG control statements on
page 781), or duplicate PARM options.
780

If the EFS program supplies non-DFSORT operands on the DFSORT OPTION
control statement and the OPTION control statement is supplied in the extended
parameter list, the EFS program must specify the non-DFSORT operands after all
DFSORT operands.
DFSORT will free any storage it acquired for the control statement or PARM string.
Note: Blanks and quotes are very important to DFSORT in determining the
control statement to send to an EFS program. Do not supply unpaired quotes in
the INCLUDE/OMIT control statements, because DFSORT treats data within
quotes as a constant, and treats blanks outside of quotes as the major delimiter.
Special handling of OPTION and DEBUG control statements

The override features of both the DFSORT OPTION and DEBUG control
statements, when supplied through different sources, require special handling
when EFS processing is in effect and either or both control statements are
requested by the EFS program.
For example, DFSORT handles override for the OPTION and DEBUG control
statements as follows:
v The OPTION control statement supplied in SORTCNTL will selectively override
corresponding options on the OPTION control statement supplied in the
extended parameter list.
v The DEBUG control statement supplied in SORTCNTL will selectively override
corresponding options on the DEBUG control statement supplied in the 24-bit
parameter list or the extended parameter list.
Because of these override features, DFSORT cannot simply send the OPTION
control statement supplied in SORTCNTL and not send the OPTION control
statement supplied in the extended parameter list. For the EFS program to process
all possible operands on the OPTION control statements, DFSORT must send the
OPTION control statements supplied in both SORTCNTL and the extended
parameter list. DFSORT will send both the OPTION and DEBUG control
statements supplied through different sources. If duplicate OPTION or DEBUG
control statements are supplied in the same source and the OPTION or DEBUG
control statements are also supplied in different sources, DFSORT will send the
first occurrence of both the OPTION and DEBUG control statements supplied
through different sources.
Control statement string returned by the EFS program

Your EFS program can alter the control statement or PARM option string and
replace it in the original contiguous storage area. If the area is too small, your
program must allocate a new contiguous area. If the string is returned in a new
storage area, your EFS program will be responsible for freeing the acquired
storage.
Your EFS program must set an Informational flag to indicate whether the control
statement or PARM option in the string should be parsed or ignored by DFSORT
(see Information flags on page 786 for further details).
OUTFIL statements cannot be returned from an EFS program to be parsed.
Rules for parsing

The content and format of the altered control statement to be parsed must
correspond to valid DFSORT values as described in Chapter 3, Using DFSORT
781

program control statements, on page 81, except when using the FIELDS operand
with SORT or MERGE, or the COND operand with INCLUDE or OMIT (see EFS
formats for SORT, MERGE, INCLUDE, and OMIT control statements on page
783).
You must observe the following rules for control statements to be returned to
DFSORT for parsing:
v
v The operation definer and corresponding operands must be in uppercase
EBCDIC format.
v At least one blank must follow the operation definer (SORT, MERGE, RECORD,
and so on). A control statement can start with one or more blanks and can end
with one or more blanks. No other blanks are allowed unless the blanks are part
of a constant.
v Labels are not allowed; a leading blank, or blanks, before the control statement
name is optional.
v No continuation character is allowed.
v Comment statements, blank statements, and remarks are not allowed.
The content and format of the altered EXEC PARM option to be parsed must
correspond to valid DFSORT values as described in Specifying EXEC/DFSPARM
PARM options on page 32.
The following operands will be ignored by DFSORT if returned by an EFS program
on the OPTION control statement:
v EFS
v
v
v
v
v
v
v
v
v
LIST
NOLIST
LISTX
NOLISTX
LOCALE
MSGDDN
MSGDD
MSGPRT
SMF
v SORTDD
v SORTIN
v SORTOUT
v USEWKDD
The following EXEC PARM options will be ignored by DFSORT if returned by an
EFS program:
v EFS
v LIST
v
v
v
v
v
782
NOLIST
LISTX
NOLISTX
LOCALE
MSGDDN

v MSGDD
v MSGPRT
EFS formats for SORT, MERGE, INCLUDE, and OMIT control

statements
In addition to using the SORT, MERGE, INCLUDE, and OMIT control statements
as explained in Program Control Statements, you can also use two additional
formats on the FIELDS and COND parameters. The formats are termed D1 and D2
and apply as follows:
v D1 with the FIELDS parameter of the SORT or MERGE statement
v D2 with the COND parameter of the INCLUDE or OMIT statement.
Use D1 and D2 to reflect data types that require special processing by EFS
program exit routines EFS01 and EFS02, respectively. You cannot specify D2 format
with the INCLUDE or OMIT parameters of the OUTFIL statement.
D1 format on FIELDS operand

The syntax for the SORT and MERGE statements using the D1 format on the
FIELDS operand is as follows.
,
SORT
MERGE
FIELDS= ( mp,mm,mf,ms
Where Represents
mp
field position within the input record
mm
field length
mf
the D1 format that designates this field as an EFS control field
ms
must be either ascending (A) or descending (D); modification by an E61

exit (E) is not allowed.
Table 88 on page 784 gives an example of using the D1 format for a SORT control
statement returned to DFSORT by the EFS program.
You must adhere to the following requirements for the D1 format:
v The mp, mm, and ms values returned must be valid SORT or MERGE control
statement values, except:
The combined value of mp and mm may exceed the record length.
CHALT will have no effect on EFS fields and will not limit the length to 256.
Value E for ms will not be allowed; EFS fields may not be altered by an E61.
FORMAT=D1 will not be allowed.
783

Table 88. D1 Format Returned by an EFS Program
Original SORT control statement sent to EFSPGM

SORT FIELDS=(15,4,FF,A,20,4,CH,A,40,7,FF,D)
Altered SORT control statement returned by EFSPGM
SORT FIELDS=(15,4,D1,A,20,4,CH,A,40,7,D1,D)
where:
FF is a user-defined format that is modified to D1 by the EFS program before returning to
DFSORT
D2 format on COND operand

Following is the syntax for the INCLUDE or OMIT statements using the D2 format
on the COND operand.
,
,
(
INCLUDE COND= (
OMIT
AND ,
OR
AND ,
OR
mc,mm,mf, operator , mc, mm, mf,
constant
mask
Where Represents
mc
the correlator identifier, a numeric value used to identify each relational

condition
mm
field length
mf
the D2 format designating an EFS field within the relational condition
operator
a valid DFSORT comparison or bit logic operator
constant
a valid DFSORT decimal, character, hexadecimal or bit constant.
mask
a valid DFSORT hexadecimal or bit string
Table 89 on page 785 gives an example of using a correlator identifier and the D2
format for an INCLUDE control statement returned to DFSORT by the EFS
program.
Note: The values for the correlator identifiers assigned to each relational condition
by the EFS program can be in any chosen order. The example in Table 89 on page
785 shows a sequential ordering for the correlator identifiers.
You must adhere to the following requirements for the D2 format:
v The mc, mm, or constant values returned must be valid INCLUDE or OMIT
control statement values, except:
The combined value of mc and mm might exceed the record length.
784

Any valid DFSORT constant or mask is allowed.
If COND=(mc1,mm1,mf1,operator,mc2,mm2,mf2) is used, both mf1 and mf2
must be D2.
CHALT has no effect on EFS fields.
FORMAT=D2 is not allowed.
Table 89. Correlator Identifier and D2 Format Returned by an EFS Program
Original INCLUDE control statement sent to EFSPGM

INCLUDE COND=(15,4,FF,EQ,20,4,FF,AND,40,7,FF,NE,50,7,FF,OR, 30,2,FF,NE,35,2,FF)
Altered INCLUDE control statement returned by EFSPGM
INCLUDE COND=(1,4,D2,EQ,1,4,D2,AND,2,7,D2,NE,2,7,D2,OR,3,2,D2,NE,3,2,D2)
Where:
v FF is a user-defined format and modified to D2 by the EFS program before returning to
DFSORT.
v The first relational condition specified, (1,4,D2,EQ,1,4,D2), uses correlator identifier value
1 to identify this relational condition.
v The second relational condition specified, (2,7,D2,NE,2,7,D2), uses correlator identifier
value 2 to identify this relational condition.
v The third relational condition specified, (3,2,D2,NE,3,2,D2), uses correlator identifier
value 3 to identify this relational condition.
Length of original control statement

The control statement includes the first byte of the control statement through the
last operand of the control statement or, if only an operation definer is supplied,
the length of the operation definer. DFSORT does not send labels supplied with the
control statement.
Length of the altered control statement

The length includes the first byte of the control statement through the last operand
of the control statement. If leading blanks are provided, the length includes the
first leading blank.
EFS program context area

The EFS program context area is a private communication area that can be set up
and used by the EFS program as appropriate. DFSORT sends the context area
address to the EFS program at each Major Call and at each call to EFS01 and
EFS02.
The EFS program is responsible for obtaining (at Major Call 1) and releasing (at
Major Call 5) the necessary storage for the EFS program context area.
Extract buffer offsets list

A linked list of offsets into the extract buffer will be passed to your EFS program.
The offsets show the starting positions into the buffer area of any EFS control fields
specified on the SORT or MERGE FIELDS operand. The offsets are sent only for
EFS control fields and are sent in the same order as specified on the FIELDS
operand. If no EFS control fields exist, the address to the offsets is zero.
785

DFSORT frees any storage it acquired for the extract buffer offsets list. The format
Chain pointer to
the next offset or
zero for end of
list
of the extract buffer offsets list is:
Offset n
4 bytes
4 bytes
Record lengths list

The record lengths list is a linked list containing the input and output record
lengths. You must be aware of the possible change in record size during run-time
(for example, with an E15 user exit).
The input and output record lengths are sent to the EFS program for informational
use only. DFSORT ignores any changes to the values made to the record lengths
list returned by the EFS program.
DFSORT frees any storage it acquired for the record lengths list. The format of the
Input record
length
record lengths list is:
4 bytes
Output record
length
4 bytes
Information flags
The information flags are defined in the figure that follows:
786

bit
0 1 2 3 45 6 7
0 0 000 0 0 0
0 0000000
00000000
00000000
Reserved
0 = Inform DFSORT to ignore parsing

the verb the EFS program returns
to DFSORT
1 = Inform DFSORT to parse the verb
the EFS program returns to DFSORT
0 = Fixed-length records
1 = Variable-length records
0 = PARM option/Control statement not from DFSPARM
1 = PARM option/Control statement from DFSPARM
0 0 = No application in effect
0 1 = SORT application in effect
1 0 = MERGE application in effect
1 1 = COPY application in effect
0 = SORTDIAG not being used
1 = SORTDIAG being used
0 = Directly invoked
1 = Program invoked
0 0 = Option from EXEC PARM
0 1 = Control statement from SYSIN
1 0 = Control statement from SORTCNTL
1 1 = Control statement from invoking parameter list
Figure 40. Information Flags
Bit
Description
Bits 0 and 1
Indicate the source of the control statement being processed. Information
flags 0 and 1 are set by DFSORT before a call to the EFS program at Major
Call 2 (multiple calls are possible at Major Call 2).
Bit 2
Indicates how DFSORT was invoked. Information flag 2 is set by DFSORT

before Major Call 1 to the EFS program.
Bit 3
Indicates whether diagnostic messages are to be printed. Information flag 3

is set by DFSORT before Major Call 1 to the EFS program.
Bits 4 and 5
Indicate the DFSORT function being run. Information flags 4 and 5 are set
by DFSORT before each call at Major Call 2 and Major Call 3 to the EFS
program (multiple calls are possible at Major Call 2 and Major Call 3).
Bit 6
Indicates the source of PARM options and control statements from

DFSPARM. Information flag 6 is set by DFSORT before each call at Major
Call 2 to the EFS program (multiple calls are possible at Major Call 2).
Bit 7
Indicates whether fixed-length records or variable-length records are to be

processed. Information flag 7 is set by DFSORT before each call at Major
Call 3 to the EFS program (multiple calls are possible at Major Call 3).
Bit 8
Set by the EFS program to inform DFSORT whether to parse or ignore the
control statement returned by the EFS program. Printing of the control
statement is managed by the LISTX/NOLISTX parameters (see OPTION
787

control statement on page 173 for further details). Information flag 8 is set
by the EFS program before returning to DFSORT from each call at Major
Call 2 (multiple calls are possible at Major Call 2).
Message list
Your EFS program can return informational or critical messages. A return code of 0
in general register 15 indicates an informational message while a return code of 16
indicates a critical message. If the EFS program has no messages to send after a
Major Call, it must zero the message list address in the EFS interface parameter
list.
At Major Call 2, if the EFS program finds a syntax error in a control statement, it
can return an offset relative to the start of the string to indicate the location of the
error. DFSORT first prints the control statement in error and then prints another
line containing a dollar symbol ($) at the location indicated by the offset.
Because DFSORT associates the relative offset with a critical message, the EFS
program must return with a return code of 16 in general register 15. If a relative
offset is returned for an EXEC PARM, the relative offset will be ignored. The EFS
program must free any storage it acquired for its messages.
The length field values must not include their own length.
The message list format follows:
Pointer to
next
message or
zero for
list end
4 bytes
Relative offset
(to syntax
error) or
zero
2 bytes
Length of
the message
text
2 bytes
Message
text
(variable
length)
* bytes
An asterisk (*) indicates that the length is determined by the corresponding length
field.
DFSORT imposes no restrictions on the format of the messages returned by an EFS
program. If you wish, you can use the DFSORT message format so that messages
in the message data set are consistent in appearance. For a description of the
message format used by DFSORT, see z/OS DFSORT Messages, Codes and Diagnosis
Guide
EFS program exit routines

If you specify EFS control fields (D1 format) or EFS fields (D2 format), DFSORT
calls the EFS01 or EFS02 exit routines, respectively, to process those fields. The
routines are generated by your EFS program, which can return the following
information about them at Major Call 3:
v The address of an extract routine, EFS01, which is used to extract the control
fields of an input record to a buffer area before a sort or merge takes place;
EFS01 is not applicable to a copy application.
v The address of an INCLUDE or OMIT routine, EFS02, which is used to process
comparison logic for including or omitting records.
During the termination phase, the EFS program must free any storage used by
these routines.
788
EFS Program Exit Routines
EFS01 and EFS02 function description

Each DFSORT control statement describes to DFSORT the type of operation to be
performed on input data. Through the EFS interface, DFSORT enables an EFS
program to provide user exit routines to perform functions beyond the capabilities
of DFSORT control statements.
The EFS program can provide user exit routine EFS01 to supplement the function
of the DFSORT SORT/MERGE control statements and can provide user exit
routine EFS02 to perform the function of the DFSORT INCLUDE/OMIT control
statements.
When preparing your EFS program exit routines, remember:
v The routines must follow standard linkage conventions.
v The general registers used by DFSORT for linkage and communication of
parameters follow operating system conventions (see Figure 41).
v The routines must use the described interfaces (see EFS01 parameter list on
page 790 and EFS02 parameter list on page 792).
Register
Use
1
DFSORT places the address of a parameter list in this register.
13
DFSORT places the address of a standard save area in this register. The area can
be used to save contents of registers used by the EFS program exit routine. The
first word of the area contains the characters SM1 in its three low-order bytes.
14
Contains the address of DFSORT return point.
15
Contains the address of the EFS program exit routine. This register can be used
as the base register for EFS program exit routine. This register is also used by
the EFS program exit routine to pass return codes to DFSORT.
Figure 41. DFSORT Register Convention
EFS01 user exit routine

Processing of user-defined data types with the EFS01 exit routine requires using
the function that alters control statements. EFS program requirements at Major
Calls 1 and 2 are:
v At Major Call 1, the EFS program must provide the control statement request list
with the SORT or MERGE operation definer. See Control statement request list
on page 779 or further details.
v At Major Call 2, the EFS program must return a new format, D1, on the SORT or
MERGE control statement that informs DFSORT to call the EFS01 routine, (the
control fields defined with the D1 format are also known as EFS control fields).
See EFS formats for SORT, MERGE, INCLUDE, and OMIT control statements
on page 783 for further details. The EFS program must also return the final
position, length, and sequence order. DFSORT uses the final position and length
to create a list of offsets.
At Major Call 3, DFSORT sends the EFS program a list of offsets into a buffer.
These offsets indicate where in the buffer the EFS program must have the EFS01
routine move the data indicated by the EFS control fields. See Extract buffer
offsets list on page 785 for further details. At Major Call 3, the EFS program must
return the address of the EFS01 routine to DFSORT.
789

During the input phase, DFSORT calls the EFS01 routine for each input record. The
EFS01 exit routine must move all data indicated by the EFS control fields, specified
in the SORT or MERGE FIELDS operand, from the input record to the extract
buffer area as specified by the offsets in the extract buffer offsets list. For each EFS
control field, the total number of bytes moved by EFS01 into the buffer area is
equal to the total number of bytes specified in the mm parameter of the altered
SORT or MERGE operand. Records are ordered according to the altered ms
parameter.
The EFS01 routine is called to extract all EFS control fields to the extract buffer
area each time a new record is brought into the input phase.
DFSORT will do sort or merge processing using the data in the extract buffer, and
will treat the data as binary data.
EFS01 parameter list

DFSORT sends three words to the EFS01 user exit routine each time it is entered:
v The address of the extract buffer area
v The address of the input record
v The address of the EFS program context area.
DFSORT places the address of a parameter list in register 1. The list begins on a
fullword boundary and is three fullwords long. The format of the parameter list is:
Address of the extract buffer area

Address of the input record
Address of the EFS program context area
The EFS01 routine must return one of the following return codes in general register
15:
0
The extraction of the EFS control field was successful.
16
The extraction of the EFS control field was unsuccessful; terminate

DFSORT.
EFS02 user exit routine

Including or omitting records based on user-defined data types with the EFS02
user exit routine requires using the function of altering control statements. EFS
program requirements at Major Calls 1 and 2 are:
v At Major Call 1, the EFS program must provide the control statement request list
with the INCLUDE or OMIT operation definer. See Control statement request
list on page 779 for further details.
v At Major Call 2, the EFS program must return a new format, D2, on the
INCLUDE or OMIT control statement that informs DFSORT to call the EFS02
routine (the fields defined with the D2 format are also known as EFS compare
fields). See EFS formats for SORT, MERGE, INCLUDE, and OMIT control
statements on page 783 for further details. The EFS program must also return
the final length and, in place of the position value, must send an identifier
(known as a correlator identifier) that identifies a specific relational condition.
For each relational condition containing EFS fields, there must be a unique
790

correlator identifier to identify the particular relational condition. See EFS
formats for SORT, MERGE, INCLUDE, and OMIT control statements on page
783 for further details.
At Major Call 3, the EFS program must return the address of the EFS02 routine to
DFSORT.
The EFS02 routine is called to perform the INCLUDE or OMIT comparison logic
for each relational condition containing an EFS field. During the input phase,
DFSORT will call the EFS02 exit routine one or more times for each input record
according to the evaluation defined by the AND, OR, or parentheses. The EFS02
exit routine must use the correlator identifier to determine the current relational
condition being performed. EFS02 must perform the comparison logic for the
current relational condition as identified by the correlator identifier. Figure 42
repeats Table 89 on page 785 to illustrate an example of the calling sequences to an
EFS02 by DFSORT.
Original INCLUDE control statement sent to EFSPGM
INCLUDE COND=(15,4,FF,EQ,20,4,FF,AND,40,7,FF,NE,50,7,FF,OR, 30,2,FF,NE,35,2,FF)
Altered INCLUDE control statement returned by EFSPGM
INCLUDE COND=(1,4,D2,EQ,1,4,D2,AND,2,7,D2,NE,2,7,D2,OR, 3,2,D2,NE,3,2,D2)
Where: the calling sequence to EFS02 may be summarized with the following tables:
Relational
condition
with
Correlator
Identifier
1
Relational
condition
with
Correlator
Identifier
2
Relational
condition
with
Correlator
Identifier
3
EFS02 returns
a return code
of 0=True or
4=False
DFSORT action when the next

logical operator is:
AND
True
Call EFS02 with Correlator Id 2
False
EFS02 returns
a return code
of 0=True or
4=False
True
False
EFS02 returns
a return code
of 0=True or
4=False

OR
Include the record

None
True
Include the record
False
Omit the record
Figure 42. Calling Sequence to EFS02 by DFSORT
791
EFS02 parameter list

DFSORT sends three words to the EFS02 exit routine each time it is entered:
v The address of the correlator identifier
v The address of the input record
v The address of the EFS program context area.
DFSORT places the address of a parameter list in register 1. The list begins on a
fullword boundary and is three fullwords long. The format of the parameter list is:
Byte 1
Byte 2
Byte 3
Byte 4
00
00
00
Correlator identifier
Address of the input record

Address of the EFS program context area
The EFS02 exit routine must return one of the following return codes in general
register 15:
0
True
The record passed the INCLUDE or OMIT test for the relational condition
of an EFS field. If applicable, processing continues with the next relational
condition. Otherwise, DFSORT accepts the record if INCLUDE is specified
or omits the record if OMIT is specified.
False
The record did not pass the INCLUDE or OMIT test for the relational
condition of an EFS field. If applicable, processing continues with the next
relational condition. Otherwise, DFSORT omits the record if INCLUDE is
specified or includes the record if OMIT is specified.
16
Terminate
An error occurred in processing the INCLUDE or OMIT logic; terminate
DFSORT.
Addressing and residence mode of EFS program user exit

routines
DFSORT supplies the following features to allow residence above or below 16MB
virtual and use of either 24-bit or 31-bit addressing:
f (bit 0 of EFS program exit routine address)
0
Enter the EFS program exit routine with 24-bit addressing in effect.
Enter the EFS program exit routine with 31-bit addressing in effect.
The EFS program user exit routine can return to DFSORT with either 24-bit or
31-bit addressing in effect. The return address that DFSORT placed in register 14
must be used.
Except for the EFS program context area address (which DFSORT sends to the EFS
program user exit routine unchanged), DFSORT handles the EFS program exit
routine parameter list addresses (that is, the pointer to the EFS program exit
routine parameter list and the addresses in the parameter list) as follows:
792
Addressing and Residence Mode

v If the EFS program exit routine is entered with 24-bit addressing in effect,
DFSORT can pass clean (zeros in the first 8 bits) 24-bit addresses or 31-bit
addresses to the EFS program exit routine. The EFS program exit routine must
return clean 24-bit addresses if the EFS program exit routine returns to DFSORT
with 31-bit addressing in effect.
v If the EFS program exit routine is entered with 31-bit addressing in effect,
DFSORT can pass clean 24-bit addresses or 31-bit addresses to the EFS program
exit routine. The EFS program exit routine must return 31-bit addresses or clean
24-bit addresses.
EFS program return codes you must supply

Your EFS program must pass one of two return codes to DFSORT:
0
Continue Processing
If you want DFSORT to continue processing for this Major Call, return
with a return code of zero in general register 15.
16
Terminate DFSORT
If you want DFSORT to terminate processing for this Major Call, return
with a return code of 16 in general register 15.
If the EFS program returns a return code of 16 from a Major Call prior to
Major Call 4 or one of its generated user exit routines returns a return code
of 16, DFSORT will skip interim Major Calls, where applicable, to the EFS
program or user exit routine, and will call the EFS program at Major Call 4
and at Major Call 5.
Multiple calls are possible at Major Call 2 and Major Call 3. If the EFS
program returns with a return code of 16 from one of the multiple calls at
Major Call 2, subsequent calls at Major Call 2, if applicable, will be
completed. If the EFS program returns with a return code of 16 from one
of the multiple calls at Major Call 3, subsequent calls at Major Call 3, if
applicable, will not be completed.
If the EFS program returns a return code of 16 at Major Call 4, DFSORT
will still call the EFS program at Major Call 5.
Record processing order

The order of record processing when using EFS is similar to processing without it.
Figure 43 on page 794 illustrates the record processing sequence for a sort or merge
and Figure 44 on page 795 illustrates the record processing sequence for a copy
when EFS processing is in effect.
The figures illustrate the same points as described in Figure 2 on page 9 with the
following exceptions:
v When record processing is done for an INCLUDE or OMIT control statement, an
EFS02 user exit routine is called to perform the comparison logic for the
relational conditions with EFS fields.
v When record processing is done for a SORT or MERGE control statement, an
EFS01 user exit routine is called to perform the extraction process for EFS
control fields.
793
Record Processing Order

Sort Application
Merge Application
SORTIN
SORTINnn
SKIPREC
E15 or
COBOL E15
E15 or
COBOL E15
INCLUDE
OMIT
EFS02
E32
INCLUDE
OMIT
EFS02
STOPAFT
INREC
INREC
SORT
SUM
EFS01
OUTREC
MERGE
SUM
OUTREC
E35 or
COBOL E35
E35 or
COBOL E35
E35 or
COBOL E35
E35 or
COBOL E35
SORTOUT
OUTFIL
SORTOUT
OUTFIL
Figure 43. EFS Record Processing Sequence for a Sort or Merge
794
EFS01
How to Request a SNAP Dump

Copy Application
SORTIN
SKIPREC
E15 or
COBOL E15
E15 or
COBOL E15
INCLUDE
OMIT
EFS02
STOPAFT
INREC
COPY
OUTREC
E35 or
COBOL E35
E35 or
COBOL E35
SORTOUT
OUTFIL
Figure 44. EFS Record Processing Sequence for a Copy
How to request a SNAP dump

You can request a SNAP dump for diagnostic purposes before or after any Major
Call except Major Call 1. Use either the EFSDPBFR parameter or the EFSDPAFT
parameter on the DEBUG statement.
See DEBUG control statement on page 91 for the correct syntax.
795
EFS Program Example
EFS program example

The following example shows how an EFS program can be used to change control
statements at run-time.
The following information is assumed for this example DFSORT run:
v The EFS program EFSPGM resides in the same library as the DFSORT
modules.
v The JCL statements for the application are:
//EXAMPLE1 JOB A12345,J. SMITH
//S1 EXEC PGM=SORT,PARM=EFS=EFSPGM
//SYSOUT
DD SYSOUT=A
//SORTIN
DD DSNAME=SMITH.INPUT,DISP=SHR,
//
UNIT=3380,SPACE=(TRK,(15,2)),VOL=SER=XYZ003,
//
DCB=(LRECL=80,BLKSIZE=80,RECFM=F)
//SORTOUT DD DSNAME=SMITH.OUTPUT,DISP=(NEW,KEEP),
//
UNIT=3380,SPACE=(TRK,(15,2)),VOL=SER=XYZ003
//SYSIN
DD *
SORT FIELDS=(5,20,CH,A,13,5,BI,D)
OPTION STOPAFT=30,DYNALLOC=3390
/*
DFSORT initialization phase:

Major call 1
Prior to Major Call 1, DFSORT sets the following fields in the EFS interface
parameter list:
v Action code=0
Major Call 1 is in effect.
v Informational bit flag 2=0
The DFSORT run is JCL-invoked.
SORTDIAG is not in effect.
DFSORT calls EFS program EFSPGM at Major Call 1, and EFSPGM sets the
following fields in the EFS interface parameter list:
v Control statement request list
Contains the OPTION operation definer, which indicates to DFSORT that the
OPTION control statement is requested by EFSPGM.
v EFSPGM program context area
EFSPGM will be using the context area.
v Message list=0
EFSPGM has no messages for DFSORT to print to the message data set. General
register 15 is set to zero.
Major call 2
parameter list:
v Action code=4
v Informational bit flag 4=0 and informational bit flag 5=0
No application is in effect.
796
EFS Program Example

EFSPGM requested the OPTION control statement. DFSORT makes a call to EFS
program EFSPGM for each control statement requested; in this case, one. DFSORT
also sets the following fields in the EFS interface parameter list:
The requested control statement came from SYSIN.
v The original OPTION control statement, including all operands and
v The length of the original OPTION control statement, including all operands and
The original control statement string is 31 bytes long.
DFSORT must parse the control statement returned by EFSPGM.
v The altered OPTION control statement, including all operands and
subparameters
OPTION STOPAFT=30,DYNALLOC=3380,EQUALS
v The length of the altered OPTION control statement, including all operands and
subparameters
The altered control statement string is 38 bytes long.
v Message list=0
Table 90 shows the original control statement sent to EFS program EFSPGM and
the altered control statement returned by EFS program EFSPGM.
Table 90. Original and Altered Control Statements
Original OPTION control statement sent to EFSPGM

Altered OPTION control statement returned by EFSPGM
OPTION STOPAFT=30,DYNALLOC=3380,EQUALS
Where:
STOPAFT=30 is the original operand and value

DYNALLOC=3380 is the original operand with a new value
EQUALS option has been added
Major call 3
parameter list:
v Action code=8
797
EFS Program Example

A sort application is in effect.
Fixed-length records are being processed.
v Record lengths list values=80
The LRECL of the input and output data sets is 80. Because the SORTOUT
LRECL was not specified, DFSORT defaulted to the SORTIN LRECL for the
SORTOUT LRECL.
v Extract buffer offsets list=0
No EFS control fields were specified on the SORT control statement.
v EFS01 address=0
Because the SORT control statement has no EFS control fields, the EFS01 user
exit routine is not used.
Because no INCLUDE control statement was supplied (with EFS fields), the
EFS02 user exit routine is not used.
v Message list=0
DFSORT termination phase

Major call 4
parameter list:
v Action code=12
v Message list=0
EFSPGM has no messages for DFSORT to print to the message data set.
And general register 15 is set to zero.
Major call 5
parameter list:
v Action Code=16
DFSORT calls EFS program EFSPGM at Major Call 5, and EFSPGM does not set
any fields in the EFS interface parameter list but sets general register 15 to zero.
798
Chapter 10. Improving efficiency

Improving performance
DFSORT is designed to optimize performance automatically. It sets optimization
variables (such as buffer sizes) and selects the most efficient of several sorting and
merging techniques.
You can improve DFSORT performance in several ways:
v Design your applications to maximize performance:
Directly invoke DFSORT processing
Specify efficient sort/merge techniques
Specify input/output data set characteristics accurately

Use extended format data sets
Specify devices that improve elapsed time
Use options that enhance performance
Use DFSORT's fast, efficient productivity features
Avoid options that degrade performance.
v Use main storage efficiently

v Allocate temporary work space efficiently
v Use Hipersorting
v
v
v
v
v
Sort with data space

Use memory object sorting
Use ICEGENER instead of IEBGENER
Use DFSORT's Performance Booster for The SAS System
Use DFSORT's BLDINDEX support.
The DFSORT z/OS DFSORT Tuning Guide provides additional information related
to many of the topics covered in this chapter.
Design your applications to maximize performance

Even though DFSORT automatically optimizes performance when your application
is run, you can still improve efficiency by using specifications and options that
permit DFSORT to make the best possible use of available resources.
Directly invoke DFSORT processing

You can enhance performance by invoking DFSORT with JCL instead of invoking
it from a COBOL or a PL/I program. Generally, COBOL or PL/I is used for
convenience. However, the trade-off can be degraded performance. You can
improve efficiency by taking advantage of the way DFSORT installation defaults
and run-time options can be fine-tuned for optimum performance, especially to
make use of control statements that work together, such as INCLUDE/OMIT,
INREC/OUTREC, SUM, and OUTFIL. You can eliminate records from input files,
reformat records to eliminate unwanted fields, combine records arithmetically, and
create reports, without requiring routines from other programs.
799
Design Your Applications to Maximize Performance

You should consider several factors when designing new applications. Some of
these factors are discussed in this section.
Whenever possible:
v Use either EBCDIC character or binary control fields
v Place binary control fields so they start and end on byte boundaries
v Avoid using the alternative collating sequence character translation
v If you know that a fixed-point control field always contains positive values,
specify it as a binary field.
v If you know that a packed decimal or zoned decimal control field always
contains positive values with the same sign (for example, X'C'), specify it as a
binary field.
v Use packed decimal format rather than zoned decimal
v If several contiguous character or binary control fields in the correct order of
significance are to be sorted or merged in the same order (ascending or
descending), specify them as one control field
v Avoid overlapping control fields.
v Avoid using locale processing if your SORT, MERGE, INCLUDE, or OMIT
character fields can be processed using the binary encoding of the data.
Efficient blocking
You can improve the performance of DFSORT significantly by blocking your input
and output records efficiently. Whenever possible, use system-determined optimum
block sizes for your data sets.
For more information about letting DFSORT select system-determined optimum
block sizes for your output data sets, see the discussion of the SDB option in
Specify efficient sort/merge techniques

Depending on various conditions, DFSORT selects different techniques for sorting
and merging. Message ICE143I informs you which technique has been selected.
For copy applications, Blockset is the only technique used. If your program cannot
use Blockset, DFSORT issues error message ICE160A and stops processing.
Sorting techniques
One condition that affects which sorting technique DFSORT selects is the type of
device used for intermediate storage. If you use a tape device, the Conventional
technique is used, which is less efficient. For more information on using tape
devices for intermediate storage, see Tape work storage devices on page 811.
The Blockset and Peerage/Vale techniques can be used only with disk work data
sets. These techniques are discussed later in this section.
Blockset sorting techniques: DFSORT's most efficient techniques, FLR-Blockset
(for fixed-length records) and VLR-Blockset (for variable-length records), will be
used for most sorting applications.
800

Notes:
v The Blockset technique might require more intermediate work space than
Peerage/Vale. For more information, see Allocate temporary work space
efficiently on page 810.
v If Blockset is not selected, you can use a SORTDIAG DD statement to force
message ICE800I, which gives a code indicating why Blockset cannot be used.
Peerage/vale sorting techniques: When the conditions for use of the Blockset
sorting technique are not met, DFSORT uses Peerage/Vale.
Merging techniques
For merging applications, DFSORT uses the Blockset and Conventional techniques.
Blockset merging techniques: DFSORT's most efficient techniques, FLR-Blockset
(for fixed-length records) and VLR-Blockset (for variable-length records), will be
used for most merging applications.
Note: If Blockset is not selected, you can use a SORTDIAG DD statement to force
message ICE800I, which gives a code indicating why Blockset cannot be used.
Conventional merging technique: When the conditions for use of the Blockset
merging technique are not met, DFSORT uses the Conventional merge technique,
which is less efficient.
Specify input/output data set characteristics accurately

DFSORT uses the information given it (about the operation it is to perform) to
optimize for highest efficiency. When you supply incorrect information or do not
supply information such as data set size and record format, the program makes
assumptions that, if incorrect, can lead to inefficiency or program termination.
Input file size

When DFSORT has accurate information about the input file size, it can make the
most efficient use of both main storage and intermediate work storage. See File
size and dynamic allocation on page 857 for information about when and how to
specify the input file size.
Variable-length records
When the input data set consists of variable-length records and dynamic allocation
of intermediate data sets is used, specify the average record length as accurately as
possible using AVGRLEN=n in the OPTION statement.
Disk devices
System performance is improved if storage is specified in cylinders rather than
tracks or blocks. Storage on sort work data sets will be readjusted to cylinders if
possible. The number of tracks per cylinder for disk devices is shown in Table 91.
Table 91. Number of Tracks per Cylinder for Disk Devices
Tracks per Cylinder
3380
15
3390
15
9345
15
If WRKSEC is in effect and the work data set is not allocated to virtual I/O,
DFSORT allocates secondary extents as required, even if not requested in the JCL.
801

Allocating twice the space used by the input data sets is usually adequate for the
work data sets. Certain conditions can cause additional space requirements. These
include:
v
v
v
v
Long control words (more than 150 bytes)

Using different device types or work data sets
Using an alternative collating sequence
Low ratio of available storage to input file size.
Care should be taken to ensure that the LRECL parameter of the DCB corresponds
to the actual maximum record length contained in your data set.
Use extended format data sets

Extended format data sets have features that can:
v significantly reduce the elapsed time DFSORT spends reading and writing data
v assist in managing the space requirements of very large data sets.
The Sequential Data Striping feature can reduce the time required to read and
write your DFSORT input and output data sets. We recommend using sequential
data striping for your very large DFSORT input and output data sets as a way to
improve elapsed time.
Features such as compression, extended addressability (for VSAM), increased
extent and volume support, and space constraint relief options assist in managing
the space requirements of very large data sets.
If you decide to use compression, realize that there is a CPU cost which could
outweigh the I/O performance benefits. Although compression can be used to
improve disk storage capacity, the impact to CPU usage and elapsed time should
be evaluated in your environment.
Consult with your storage administrator to learn how these features can be
exploited in your environment.

The use of tapes managed by DFSMSrmm, or a tape management system that uses
ICETPEX, allows DFSORT to obtain accurate information about input file size and
data set characteristics. This can result in improved performance and more efficient
use of both main storage and intermediate work storage.
Specify devices that improve elapsed time

To get the best elapsed time improvement when using disk with DFSORT, you
should use high speed devices, such as IBM's Enterprise Storage Server (ESS) and
TotalStorage RS8000 Series subsystems for SORTIN, SORTOUT, SORTWKdd,
SORTOUT, and OUTFIL data sets. Throughput can be further improved by
leveraging high speed channels, such as IBM's FICON connectivity solutions.
Use options that enhance performance

To obtain optimum performance, you can fine-tune your installation and run-time
options. Several options that can enhance performance are described later in this
section.
802
CFW
Blockset sorting performance may be improved by using the Cache Fast Write
(CFW) capability of IBM's storage controllers. Because IBM's latest storage
subsystems have large cache sizes and high speed disk arrays, the use of CFW
may not produce a significant performance gain.
The CFW parameter specifies whether DFSORT can use CFW when processing
SORTWKdd data sets. The default is CFW=YES.
If you use the Hyperswap function in an environment that has a high availability
configuration, you should specify CFW=NO to prevent DFSORT from abending
when a Hyperswap is initiated.
DSA
Performance can be improved for Blockset sort applications by using Dynamic
Storage Adjustment (DSA).
The DSA parameter sets the maximum amount of storage available to DFSORT for
dynamic storage adjustment of a Blockset sort application when
SIZE/MAINSIZE=MAX is in effect. If you specify a DSA value greater than the
TMAXLIM value, you allow DFSORT to use more storage than the TMAXLIM
value if doing so should improve performance. DFSORT only tries to obtain as
much storage as needed to improve performance up to the DSA value.
DSPSIZE
Performance can be improved for sort applications that use the Blockset technique
by using dataspace sorting.
The DSPSIZE parameter sets the maximum size of a data space to be used during
a run. Specifying DSPSIZE=MAX allows DFSORT to optimize the maximum size of
a data space to be used during a run, subject to other system and concurrent
Hipersorting, memory object sorting and dataspace sorting activity throughout the
run. Total dataspace sorting activity on a system can be further limited by the
EXPMAX, EXPOLD, and EXPRES installation options. See the description of
DSPSIZE in OPTION control statement on page 173 for more information.
FASTSRT
By specifying the COBOL FASTSRT compiler option, you can significantly reduce
DFSORT processor time, EXCPs, and elapsed time. With FASTSRT, DFSORT
input/output operations are more efficient because DFSORT rather than COBOL
does the input/output (see Figure 45 on page 804). For more details, see the
COBOL publications.
The FASTSRT option does not take effect for input and output if input and output
procedures are used in the SORT statement. Many of the functions usually
performed in an input or output procedure are the same as those done by DFSORT
INREC, OUTFIL, OUTREC, INCLUDE or OMIT, STOPAFT, SKIPREC, and SUM
functions. You might be able to eliminate your input and output procedures by
coding the appropriate DFSORT program control statements and placing them in
either the DFSPARM (DFSORT), SORTCNTL (DFSORT), or IGZSRTCD (COBOL)
data set, thereby allowing your SORT statement to qualify for FASTSRT.
803
Figure 45. Faster Sorting with COBOL
SDB
To improve Blockset elapsed time, and disk and tape utilization, specify
installation option SDB=LARGE as your site's installation (SDB=INPUT is the
IBM-supplied default). SDB=LARGE allows DFSORT to select the
system-determined optimum block size for your disk and tape output data sets,
when appropriate.
HIPRMAX
Blockset sorting performance can be improved by using Hiperspace along with
disk for temporary storage.
The HIPRMAX parameter sets the maximum amount of Hiperspace to be used
during a run. Specifying HIPRMAX=OPTIMAL allows DFSORT to optimize the
maximum amount of Hiperspace to be used during a run, subject to other system
and concurrent Hipersorting, memory object sorting and dataspace sorting activity
throughout the run. Total Hipersorting activity on a system can be further limited
by the EXPMAX, EXPOLD, and EXPRES installation options. See the description of
HIPRMAX in OPTION control statement on page 173 for more information.
MOSIZE
Blockset sorting performance can be improved by using memory object sorting.
The MOSIZE parameter sets the maximum amount of memory object storage to be
used during a run. Specifying MOSIZE=MAX allows DFSORT to optimize the
maximum size of a memory object to be used during a run, subject to other system
and concurrent Hipersorting, memory object sorting and dataspace sorting activity
throughout the run. Total memory object sorting activity on a system can be
further limited by the EXPMAX, EXPOLD, and EXPRES installation options. See
the description of MOSIZE in OPTION control statement on page 173 for more
information.
804
Use DFSORT's fast, efficient productivity features

DFSORT offers a rich set of fast, efficient productivity features. These features can
eliminate the up-front costs of writing and debugging your own code to perform
various tasks, and will perform those tasks more efficiently. The functional
capabilities of each of the features listed later in this section is described in detail
elsewhere in this document. This section highlights the performance aspects of
these features.
INCLUDE or OMIT, STOPAFT, and SKIPREC

You can use the INCLUDE or OMIT statement and the STOPAFT and SKIPREC
options to reduce the number of records to be processed, which can reduce
processor and data transfer time.
v The INCLUDE and OMIT statements allow you to select records by comparing
fields with constants or other fields.
v The STOPAFT option allows you to specify the maximum number of records to
be accepted for sorting or copying.
v The SKIPREC option allows you to skip records at the beginning of the input
file and sort or copy only the remaining records.
OUTFIL
If you need to create multiple output data sets from the same input data set, you
can use OUTFIL to read the input data set only once, thus improving performance.
OUTFIL can be used for sort, merge, and copy applications to provide
sophisticated filtering, editing, conversion, lookup and replace, and report features.
If you are creating only a single output data set and do not need the features of
OUTFIL, use SORTOUT rather than OUTFIL for best performance.
LOCALE
You can use the LOCALE option to sort, merge, and compare character data based
on collating rules in an active locale; this enables you to obtain results with
DFSORT that were previously possible only through pre-processing or
post-processing of your data. By eliminating such costly processing, you can save
time and processing resources.
SUM
You can improve performance by using SUM to add the contents of fields. The
SUM statement adds the contents of specified SUM fields in records with identical
control fields. The result is placed in one record while the other record is deleted,
thus reducing the number of records to be output by DFSORT.
You can use installation option ZDPRINT=YES or run-time option ZDPRINT to
specify that positive zoned decimal fields that result from summing are to be
printable. That is, you can tell DFSORT to change the last digit of the zone from
hex C to hex F.
Eliminating duplicate records: You can eliminate records with duplicate keys by
specifying
SUM FIELDS=NONE
when using the SUM control statement.

For a diagram of the processing sequence for record handling statements, user
exits, and options, see Figure 2 on page 9.
805
ICETOOL
ICETOOL is a multi-purpose utility that allows you to use DFSORT's highly
efficient I/O and processing to perform multiple operations on one or more data
sets in a single job step. ICETOOL's 17 operators allow you to perform sort, copy,
statistical, and report operations quickly and efficiently.
Avoid options that degrade performance

Certain options can adversely affect performance, and should be used only when
necessary. For example, the CKPT option, which activates checkpoint/restart,
prevents use of the efficient Blockset techniques.
CKPT
The CKPT option might preclude the use of the more efficient Blockset technique.
Note: If installation option IGNCKPT=YES has been selected, DFSORT ignores the
checkpoint/restart request and selects the Blockset technique.
EQUALS
The EQUALS option increases the time needed for comparison of records and for
data transfer.
EQUCOUNT
The EQUCOUNT option takes additional time to count the number of records with
equal keys.
LOCALE
The LOCALE option may increase the time required to run an application.
NOCINV
The NOCINV option precludes the use of control interval access for more efficient
VSAM processing.
NOBLKSET
The NOBLKSET option precludes the use of the more efficient Blockset technique.
VERIFY
The VERIFY option degrades performance, because it involves extra processing.
Tape work data sets

Use of tape work data for intermediate storage precludes the use of the much more
efficient disk techniques.
User exit routines

When user exit routines are included in an application, the time required to run
the application is usually increased.
The run time required by most user exit routines is generally small, but the
routines at user exits E15, E32, and E35 are entered for each record of the data sets.
For large input data sets, the total run time of these routines can be relatively
large.
Dynamic binding or link-editing

Dynamic binding or link-editing of user exit routines degrades performance.
806
EFS programs
When EFS programs are included in an application, the time required to run the
application might increase.
Use main storage efficiently

In general, the more main storage you make available to DFSORT, the better the
performance for large applications. To prevent excessive paging, ensure that
sufficient real storage is available to back up the amount of main storage used.
This is especially important with main storage sizes greater than 32MB. The default
amount of main storage that will be made available to DFSORT is defined when it
is installed.
Although it is possible to run a "very minimal" sort application in 88KB of storage,
the actual minimum amount of storage required is generally 128KB to 440KB
depending on the application. However, 4MB or more of storage is recommended
for better performance. Improved performance will be most noticeable with large
input files. With MAINSIZE=MAX, the DSA value can be used to automatically
increase the amount of storage DFSORT uses for a sort application when doing so
should improve performance.
Note: When Blockset is selected, DFSORT can place selected buffers above 16MB
virtual. This frees more storage for DFSORT without having to increase the
REGION size. A REGION size of at least 440KB must be available to allow
DFSORT to use storage effectively.
Tuning main storage

Either the REGION value or the MAINSIZE/SIZE value can determine how much
storage is available to DFSORT. See z/OS DFSORT Installation and Customization for
details.
Generally, the most efficient way to allocate (virtual) main storage is to use
MAINSIZE/SIZE=MAX explicitly or by default. However, problems can arise if the
values for the TMAXLIM or MAXLIM installation options have been set
excessively high (or low). Guidelines for setting these values are given in z/OS
DFSORT Installation and Customization.
Note: Do not use SIZE/MAINSIZE=MAX with password-protected data sets if
passwords are to be entered through a routine at a user exit, because DFSORT
cannot then open the data sets during the initialization phase to make the
necessary calculations.
If you specify MAINSIZE/SIZE=n and give n a value less than that specified for
the MINLIM installation option, MINLIM is used.
When SIZE/MAINSIZE=MAX is in effect, DFSORT will use its Dynamic Storage
Adjustment (DSA) feature, when appropriate, to improve performance.
If the MINLIM value is greater than that specified for REGION on the EXEC
statement, DFSORT attempts to use the value specified for MINLIM; if it fails to
get the amount specified by MINLIM, DFSORT still tries to run, provided at least
88KB (below 16MB virtual) are available to DFSORT.
Storage used for OUTFIL processing will be adjusted automatically, depending
upon several factors, including:
v Total available storage
807
Use Main Storage Efficiently

v Non-OUTFIL processing storage requirements
v Number of OUTFIL data sets and their attributes (for example, block size).
OUTFIL processing is subject to the ODMAXBF limit and your system storage
limits (for example, IEFUSI) but not to DFSORT storage limits, that is,
SIZE/MAINSIZE, MAXLIM, and TMAXLIM. DFSORT attempts to use storage
above 16MB virtual for OUTFIL processing whenever possible.
Note:
1. In some cases, this release of DFSORT may use more storage than prior releases
for comparable applications. This might affect the operation of some
applications. For example, some applications that run as in-storage sorts (with
no SORTWKdd data sets) in previous releases might not run in-storage when
using this release. The amount of storage allocated is normally controlled by
TMAXLIM. A REGION size of at least 440KB must be available if DFSORT is to
achieve acceptable performance. The allocation of storage can be adversely
affected if you have a smaller region value or if DFSORT needs to allocate
buffers below 16MB virtual.
2. For extremely large sorts (for example, 2GB or more of data), make sure that
Hipersorting, memory object sorting and dataspace sorting are enabled, and
that 32MB or more of main storage is available to DFSORT.
The relationship between TMAXLIM, MAXLIM, MINLIM, and REGION might be
described as a series of checks and balances.
Your system programmer has set the default storage values according to your site's
major sorting requirements. If you have an overnight or batch time window that
must be met, increasing storage (using REGION or SIZE/MAINSIZE=n) can give
you some relief from the time constraint. If you are concerned with processor time,
decreasing storage (using REGION or SIZE/MAINSIZE=n) can reduce the
processor time associated with sorting small files.
In general, when you vary the amount of storage available to DFSORT, several
things occur:
1. If you increase the amount of storage:
v EXCPs are reduced.
v For larger files, processor time generally decreases; that is, overhead in
managing the extra storage is offset by DFSORT having to make fewer
passes over the data.
v For a very heavily loaded system, elapsed time might increase because
DFSORT can be swapped out more often.
v For very small sorts, processor time might remain stable or increase because
of the overhead in managing the extra storage. For larger files, processor
time will usually decrease because the overhead in managing the extra
storage would be less than the benefit gained by DFSORT making fewer
passes over the data.
2. If you decrease the amount of storage:
v EXCPs increase.
v Elapsed time increases for most sorts.
v Processor time decreases for very small files, but increases for larger files.
Changing the main storage allocation can affect system efficiency. By reducing the
amount of main storage allocated, you impair performance of DFSORT to allow
808

other programs to have the storage they need to operate simultaneously; by
increasing the allocation, you can run large DFSORT applications efficiently at the
risk of decreasing the efficiency of other applications sharing the
multiprogramming environment.
Releasing main storage

Under some circumstances, DFSORT uses all the available storage in your
REGION. This normally will not occur for storage above 16MB virtual (if it does,
use the ARESINV or ARESALL options or lower your SIZE/MAINSIZE value).
This section explains how to release storage within your REGION.
When SIZE/MAINSIZE=n is in effect and n is greater than the REGION parameter
or the default REGION value, or when SIZE/MAINSIZE=MAX and TMAXLIM is
greater than your REGION, specify the storage you need released in the following
way:
v For applications with user exits:
For directly invoked DFSORT, you can choose one of the following:
- Use the m parameter of the MODS control statement.
- If SIZE=MAX is in effect, you can use the RESALL option.
- Change your REGION so that REGION is greater than SIZE/MAINSIZE
(the difference is available).
- If the OVERRGN installation option is smaller than your system IEFUSI
value, this difference is available. (OVERRGN is an installation option that
can be modified only by your system programmer.)
For program invoked DFSORT, you can choose one of the following:
- If the user exit address is not passed in the parameter list (that is, it is
specified with a MODS statement), use the m parameter on the MODS
statement.
- If the user exit address is passed in the parameter list,and
SIZE/MAINSIZE=MAX is in effect, use the RESINV option.
- If the user exit address is passed in the parameter list, and
SIZE/MAINSIZE=n is in effect, change your REGION so that the REGION
is greater than SIZE/MAINSIZE (the difference is available).
- If many of your DFSORT applications pass the user exit address in the
parameter list and SIZE/MAINSIZE=n is in effect, then consider having the
OVERRGN value changed by your system programmer to less than your
IEFUSI value.
v For applications without user exits:
For directly invoked DFSORT, you can choose one of the following:
- If SIZE/MAINSIZE=MAX is in effect, use the RESALL option.
- If SIZE/MAINSIZE=n is in effect, change your REGION so that REGION is
greater than SIZE/MAINSIZE (the difference is available).
- Have the OVERRGN value changed by your system programmer to less
than your IEFUSI value.
For program invoked DFSORT, you can choose one of the following:
- If SIZE/MAINSIZE=MAX is in effect, use the RESINV option.
- If SIZE/MAINSIZE=n is in effect, change your REGION so that REGION is
greater than SIZE/MAINSIZE (the difference is available).
- Have the OVERRGN value changed by your system programmer to less
than your IEFUSI value.
809

When SIZE/MAINSIZE is less than REGION, make sure the difference between
SIZE/MAINSIZE and your REGION specification value or default provides
sufficient storage for system or user exit routine use.
Allocate temporary work space efficiently

Performance is enhanced when multiple channels are available. Performance is also
improved if the device is connected so that two channel paths exist between each
device and the processor that is running the program.
Disk work storage devices

Program performance is improved if you use devices, storage areas, and channels
efficiently. If you specify a particular device type with the UNIT parameter on the
DD statements that define intermediate storage data sets (for example,
UNIT=3390), DFSORT assigns areas, and some optimization occurs automatically.
You can get the best performance using disk intermediate storage devices when
you:
v Use two or more work data sets.
v Select the storage device with the fastest data transfer rate.
v Assign one work data set per volume if Parallel Access Volumes (PAV) are not
used. This will help minimize I/O queue time on the SORTWKdd data sets.
Use devices that are of the same type.
Use two channel paths to devices.
Make all work data sets the same size, or as nearly the same size as possible.
Make sure the SORTWKdd data sets do not share devices or channels with the
SORTIN, SORTOUT, or OUTFIL data sets.
v Specify contiguous space for work data sets, and make sure there is enough
primary space so that the automatic secondary allocation is not needed.
v
v
v
v
v Provide adequate virtual storage when work data sets are allocated on
non-synchronous devices, as explained in Non-synchronous storage
subsystems on page 855.
Elapsed time is decreased when DFSORT can both read input while writing to
SORTWKdd and write output while reading from SORTWKdd. If, for example,
you have two channels, the best allocation of them is to have SORTIN, SORTOUT,
and OUTFIL data sets on one and the SORTWKdd data sets on the other.
Storage requirements for different disk techniques can be estimated by using the
guidelines found in Appendix A, Using work space, on page 853.
Virtual I/O for work data sets

Although VIO data sets can provide performance improvements in many
applications, these work data sets are generally not as effective as disk work data
sets for DFSORT.
DFSORT temporary work data sets allocated to virtual devices (VIO) can provide
reduced elapsed time at the cost of increased CPU time for DFSORT applications.
In general, this is not a good trade-off and VIO should not be used for DFSORT
work data sets.
If work data sets are allocated to VIO, the VIO installation option tells DFSORT
how to handle allocation to VIOs:
v VIO=YES causes DFSORT to accept the use of VIOs for work data sets.
810
Allocate Temporary Work Space Efficiently

v VIO=NO causes DFSORT to reallocate work data sets from virtual devices to
real devices. Note that in order for re-allocation to be successful, a real device
with the same device type as the virtual device must be available.
Re-allocation of VIO data sets (VIO=NO) is recommended over no re-allocation
(VIO=YES). However, it is better to avoid using VIO work data sets in the first
place, since re-allocation wastes time and resources.
Tape work storage devices

The use of tape work storage devices prevents the use of the more efficient
Blockset technique. Best performance, using tape intermediate storage, is usually
obtained when you use six or more tape drives of the fastest type. As a general
rule, you should use as many tapes as you have available for intermediate storage.
A larger number of tapes increases the number of strings that can be merged in
one pass, and, therefore, decreases the number of passes required in the
intermediate merge phase. This then reduces elapsed time and, often, the number
of I/O operations.
Increasing the number of work units, however, also reduces the block size used for
intermediate storage; this can become a critical factor if you have relatively little
main storage available for buffers. For example, if DFSORT has only 88KB in
which to operate, you probably achieve no improvement (and might find
deterioration) if you use more than four tape work units. Therefore, apply the
general rule of using as many tapes as possible only when DFSORT has more than
100KB available.
For information on how to determine storage requirements when using different
tape techniques, see Appendix A, Using work space, on page 853.
Note:
1. The frequency with which the tape direction changes during DFSORT work file
operations has more of an impact on the effective data rate of the IBM
3480/3490/3590 Magnetic Tape Subsystems than on IBM 3420 Magnetic Tape
Units. Because of this characteristic, performance comparisons between these
tape units for intermediate storage cannot be reliably predicted and can vary
widely.
2. Devices using the Improved Data Recording Capability (IDRC) feature are not
recommended as intermediate storage devices, as they do not perform well
with the read backward command.
Use Hipersorting
Hipersorting uses Hiperspace to improve the performance of sort applications that
use DFSORT's Blockset Technique. A Hiperspace is a high-performance data space
that resides in central storage and is backed by auxiliary storage when necessary.
With Hipersorting, Hiperspace is used in place of and along with disk for
temporary storage of records during a Blockset sort. Hipersorting reduces I/O
processing, which in turn reduces elapsed time, EXCPs, and channel usage.
Hipersorting is recommended when the input or output is a compressed sequential
or VSAM data set.
You can control the maximum amount of Hiperspace for a Hipersorting application
with the HIPRMAX parameter. HIPRMAX can direct DFSORT to dynamically
determine the maximum amount of Hiperspace, subject to the available storage at
811
Use Hipersorting
the start of the run. You can also use HIPRMAX to suppress Hipersorting when
optimizing CPU time is your major concern because Hipersorting can slightly
degrade CPU time.
The actual amount of Hiperspace a Hipersorting application uses depends upon
several factors. See the HIPRMAX description in OPTION control statement on
page 173 for more details. Most important, throughout the run, DFSORT
determines the amount of available storage as well as the amount of storage
needed by other concurrent Hipersorting, memory object sorting, and datapace
sorting applications. Based on this information, DFSORT switches dynamically
from using Hiperspace to using disk work data sets when either a storage shortage
is predicted or the total Hipersorting, memory object sorting and dataspace sorting
activity on the system reaches the limits set by the EXPMAX, EXPOLD, and
EXPRES installation options. See z/OS DFSORT Installation and Customization for a
complete description of these installation options.
Sort with data space

dataspace sorting uses data space to improve the performance of sort applications
that use DFSORT's Blockset Technique.
dataspace sorting allows DFSORT to sort large pieces of data at a time. This helps
to reduce CPU time and elapsed time.
You can control the maximum size of a data space for a dataspace sorting
application with the DSPSIZE parameter. DSPSIZE can direct DFSORT to
dynamically determine the maximum size of a data space, subject to the available
central storage at the start of the run. DSPSIZE=0 means that DFSORT will not use
dataspace sorting.
The actual size of a data space that a dataspace sorting application uses depends
upon several factors. See the DSPSIZE description in OPTION control statement
on page 173 for more details.
The following functions and types of data sets are not supported for dataspace
sorting:
v Spool, dummy, or pipe data set, or z/OS UNIX file, as input.
v User exits
v INREC, OUTFIL, OUTREC, and SUM
v EQUCOUNT
Dataspace sorting is seldom used for very small data sets of a few MB or so
because it is more efficient to sort small amounts of data entirely in main storage.
The following are actions you can take that might increase the use of dataspace
sorting:
v Specify sufficient main storage. The default is 6MB, the recommended minimum
for dataspace sorting. If you increase the amount of main storage specified, more
dataspace sorting is possible, especially when sorting large amounts of data
(multiple hundred MBs). Specifying more than 12MB or so will have no
significant impact on DFSORT's decision to use dataspace sorting; it will,
however, improve the performance of large non-dataspace sorting applications.
v Specify generous extent sizes for work data sets, especially for secondary
extents. Dataspace sorting is frequently used in conjunction with disk work
space but never with Hiperspace or with tape work data sets.
812
Sort with Data Space

v Specify DSPSIZE=MAX.
v Verify that IEFUSI does not place any restrictions on the size of the data spaces a
single address space can create.
v Ensure that DFSORT has accurate information about the input file size. DFSORT
can automatically estimate the file size for disk input data sets and tape data
sets managed by DFSMSrmm or a tape management system that uses ICETPEX.
See File size and dynamic allocation on page 857 for information on situations
where DFSORT cannot determine the file size accurately, and what to do about
it.
Use memory object sorting

Memory object sorting uses a memory object in 64-bit virtual storage to improve
the performance of sort applications that use DFSORT's Blockset Technique. A
memory object is a data area in virtual storage that is allocated above the bar and
backed by central storage. With memory object sorting, a memory object is used in
place of and along with disk for temporary storage of records during a sort.
Memory object sorting reduces I/O processing, which in turn reduces elapsed
time, EXCPs, and channel usage. Memory object sorting is recommended for large
input data sets when a sufficient amount of central storage is available.
You can control the maximum size of a memory object for a memory object sorting
application with the MOSIZE parameter. MOSIZE can direct DFSORT to
dynamically determine the maximum size of a memory object, subject to the
available central storage at the start of the run. MOSIZE=0 means that DFSORT
will not use memory object sorting.
The actual size of a memory object that a memory object sorting application uses
depends upon several factors. See the MOSIZE description in OPTION control
statement on page 173 for more details.
The following are actions you can take that might increase the use of memory
object sorting:
v Verify that a sufficient size for memory objects is defined by the MEMLIMIT
parameter on the JOB or EXEC JCL statement.
v Specify MOSIZE=MAX.
v Specify generous extent sizes for work data sets, especially for secondary
extents.
v Verify that IEFUSI does not place any restrictions on the size of the memory
objects a single address space can create.
v Ensure that DFSORT has accurate information about the input file size. DFSORT
can automatically estimate the file size for disk input data sets and tape data
sets managed by DFSMSrmm or a tape management system that uses ICETPEX.
See File size and dynamic allocation on page 857 for information on situations
where DFSORT cannot determine the file size accurately, and what to do about
it.
Use ICEGENER instead of IEBGENER

You can achieve more efficient processing for applications set up to use the
IEBGENER system utility by using DFSORT's ICEGENER facility. Qualifying
IEBGENER jobs are processed by the equivalent (though not identical), but more
efficient, DFSORT copy function. If, for any reason, the DFSORT copy function
cannot be used (for example, if IEBGENER control statements are specified),
control is automatically transferred to the IEBGENER system utility.
813
Use ICEGENER Instead of IEBGENER

ICEGENER, like IEBGENER, will use an SDB=value parameter you supply using
PARM='SDB=value', when appropriate. The valid SDB=value parameters are
SDB=LARGE, SDB=YES, SDB=SMALL, SDB=INPUT and SDB=NO, as explained in
OPTION control statement on page 173. If you supply an invalid SDB=value
parameter, ICEGENER will transfer control to IEBGENER, which will terminate
due to the invalid parameter. If you do not supply an SDB=value parameter,
ICEGENER will use your site's DFSORT installation default for SDB, when
appropriate (the IBM-supplied default is SDB=INPUT). If ICEGENER transfers
control to IEBGENER, IEBGENER will use the SDB=value parameter you supply, if
any, or its normal default for SDB.
ICEGENER will also recognize DFSORT parameters other than SDB=value you
supply using PARM='parameter' that are valid on DFSORT's OPTION statement as
explained in OPTION control statement on page 173. However, IEBGENER does
not recognize any parameters other than the valid SDB=value forms, so if DFSORT
must transfer control to IEBGENER, IEBGENER will not recognize DFSORT's
parameters and will terminate. Likewise, if you supply a DFSORT parameter using
PARM='parameter' that is not valid on DFSORT's OPTION statement, DFSORT will
transfer control to IEBGENER and IEBGENER will terminate due to the invalid
parameter.
//S1 EXEC PGM=ICEGENER,PARM=SIZE=2000,MAINSIZE=2000K
ICEGENER will accept SIZE=2000 and MAINSIZE=2000K as valid DFSORT

OPTION parameters that specify an exact file size of 2000 records and a limit of
2000K bytes of storage, respectively. If DFSORT copy can be used, these parameters
will be used. But if DFSORT must transfer control to IEBGENER, IEBGENER will
terminate because it treats SIZE=2000 and MAINSIZE=2000K as invalid
parameters.
As another example, if you specify:
//S2 EXEC PGM=ICEGENER,PARM=SIZE=2000K
ICEGENER will treat SIZE=2000K as an invalid DFSORT OPTION parameter and

will transfer control to IEBGENER, which will terminate because it treats
SIZE=2000K as an invalid parameter.
Thus, you can pass PARM parameters to ICEGENER that are valid as DFSORT
OPTION parameters, but you must be aware that if ICEGENER transfers control to
IEBGENER, those parameters will cause IEBGENER to terminate. PARM
parameters that are not valid as DFSORT OPTION parameters (even if they are
valid as DFSORT PARM parameters) will cause ICEGENER to transfer control to
IEBGENER, which will terminate.
ICEGENER can transfer control to IEBGENER due to DFSPARM or SORTCNTL
statement errors or other errors detected by DFSORT. Therefore, we recommend
that ICEGENER not be used for any application for which IEBGENER cannot be
used, to avoid unwanted IEBGENER processing. For example, if ICEGENER is
used with an INCLUDE statement in DFSPARM, IEBGENER could be used and
complete successfully, but the INCLUDE statement would be ignored. Instead,
DFSORT copy should be used directly so that IEBGENER cannot be called.
However, if you know that ICEGENER will use DFSORT copy, you can use a
DFSPARM data set with ICEGENER to pass control statements and parameters to
DFSORT. For example, if you specify:
814

//DFSPARM DD *
OPTION SPANINC=RC0
/*
and ICEGENER uses DFSORT copy, any incomplete spanned records DFSORT
detects in a variable spanned input data set are eliminated.
If your site has installed ICEGENER to be invoked by the name IEBGENER, you
need not make any changes to your applications to use ICEGENER. If your site
has not chosen automatic use of ICEGENER, you can use ICEGENER by
substituting the name ICEGENER for IEBGENER on the EXEC statement (when
DFSORT is directly invoked) or LINK macro (when DFSORT is program-invoked)
in any applications you choose. Program-invoked applications must be recompiled.
Following is an example of how an IEBGENER application can be changed to use
ICEGENER by substituting the name ICEGENER for the name IEBGENER in the
EXEC statement.
//GENER JOB...
// EXEC PGM=ICEGENER
//SYSPRINT DD SYSOUT=*
//SYSUT1 DD DSN=CONTROL.MASTER,DISP=OLD,UNIT=3380,VOL=SER=MASTER
//SYSUT2 DD DSN=CONTROL.BACKUP,DISP=OLD,UNIT=3380,VOL=SER=BACKUP
//SYSIN DD DUMMY
The IEBGENER DD statements SYSUT1 (input), SYSUT2 (output), and SYSPRINT

(messages) are used by DFSORT for SORTIN, SORTOUT, and SYSOUT,
respectively. These DD statement names will be translated by using an extended
parameter list to invoke the copy function. If DFSORT cannot be used (for
example, because IEBGENER control statements are specified), control will be
transferred to IEBGENER.
Note:
1. The SYSUT2 data set should not be the same as the SYSUT1 data set because
this can cause lost or incorrect data or unpredictable results.
2. Whether ICEGENER is invoked from a program or not, DFSORT will be
invoked from ICEGENER using an extended parameter list. Therefore, the
installation defaults for the ICEAM2 or ICEAM4 environment, or for an
ICETDx environment if activated for ICEAM2 or ICEAM4, apply and
SORTCNTL or DFSPARM can be used to provide additional control statements
for the copy application; for example, OPTION. However, ICEGENER can
transfer control to IEBGENER due to DFSPARM or SORTCNTL statement
errors or other errors detected by DFSORT. Therefore, DFSORT copy should be
used directly rather than ICEGENER if DFSORT processing statements such as
INCLUDE, OUTREC, SUM and so on are required.
3. For most error conditions that prevent the use of DFSORT copy, control will be
transferred to the IEBGENER system utility. DFSORT messages will not be
printed unless a SORTDIAG DD statement is supplied. Use of the SORTDIAG
DD statement will allow you to determine why DFSORT copy could not be
used.
4. If DFSORT copy is used, its operation and messages will be equivalent to a
directly called DFSORT copy application. If an unrecoverable error is
encountered (for example, an I/O error), DFSORT's return code of 16 will be
changed by ICEGENER to a return code of 12 to emulate the return code from
a failing IEBGENER application.
5. DFSORT copy can perform some functions not provided by IEBGENER, such
as certain padding and truncation operations. ICEGENER processing is not
815

identical to IEBGENER processing in all cases, since DFSORT copy uses
methods to enhance performance (EXCP, for example) that are not used by
IEBGENER.
6. In some cases, IEBGENER terminates when the SYSUT2 LRECL is different
from the SYSUT1 LRECL. ICEGENER takes one of three actions depending on
the GNPAD (LRECL padding) or GNTRUNC (LRECL truncation) installation
option, as appropriate.
If you want ICEGENER to transfer control to IEBGENER when the SYSUT2
LRECL is larger than the SYSUT1 LRECL, use installation option GNPAD=IEB.
If you want ICEGENER to handle LRECL padding, use GNPAD=RC0 (the
supplied default) or GNPAD=RC4.
If you want ICEGENER to transfer control to IEBGENER when the SYSUT2
LRECL is smaller than the SYSUT1 LRECL, use installation option
GNTRUNC=IEB. If you want ICEGENER to handle LRECL truncation, use
GNTRUNC=RC0 (the supplied default) or GNTRUNC=RC4.
7. For a call to ICEGENER, or to IEBGENER as an alias for ICEGENER, register 1
must point to a valid parameter list consisting of three addresses as follows:
v Address1: The address of the Option List.
v Address2: The address of the Alternate DDname List.
v Address3: The address of the Page Number List.
Methods of calling ICEGENER that generate a valid parameter list will allow
ICEGENER to use DFSORT's copy feature, whereas methods of calling
ICEGENER that generate an invalid parameter list will cause ICEGENER to
transfer control to IEBGENER. For example:
call *(icegener)
on the TSO command line generates a valid parameter list, whereas:

icegener
on the TSO command line generates an invalid parameter list.
ICEGENER return codes

ICEGENER can use either IEBGENER or the DFSORT copy function. However, for
unsuccessful completion due to an unsupported operating system, ICEGENER
passes back a return code of 24 to the operating system or the invoking program,
without using either IEBGENER or DFSORT copy.
If ICEGENER transfers control to IEBGENER, IEBGENER passes back its return
code to the operating system or the invoking program.
If ICEGENER uses the DFSORT copy function:
v For successful completion, ICEGENER passes back a return code of 0 or 4 to the
operating system or the invoking program.
v For unsuccessful completion with NOABEND in effect, ICEGENER passes back
a return code of 12 (changed from 16) to the operating system or the invoking
program.
v For unsuccessful completion with ABEND in effect, DFSORT issues a user abend
with the appropriate code as specified by the ABCODE installation option (either
the error message number or a number between 1 and 99).
The meanings of the return codes that ICEGENER passes back (in register 15) are:
0
816
Successful completion. ICEGENER completed successfully.

4
Successful completion. ICEGENER completed successfully, and:

v Installation option GNPAD=RC4 was specified and the SYSUT2 LRECL
was larger than the SYSUT1 LRECL (LRECL padding) or
v Installation option GNTRUNC=RC4 was specified and the SYSUT2
LRECL was smaller than the SYSUT1 LRECL (LRECL truncation), or
v SPANINC=RC4 was in effect and one or more incomplete spanned
records was detected, or
v NULLOUT=RC4 was in effect and there were no records for the SYSUT2
data set, or
v NULLOFL=RC4 was in effect and there were no data records for an
OUTFIL data set.
12
Unsuccessful completion. DFSORT detected an error that prevented

ICEGENER from completing successfully.
24
Unsupported operating system. This operating system is not supported by

this release of DFSORT.
Use DFSORT's performance booster for The SAS System

DFSORT provides significant CPU time improvements for SAS applications. To
take advantage of this feature, contact SAS Institute Inc. for details of the support
they provide to enable this enhancement.
Use DFSORT's BLDINDEX support

DFSORT provides support that enables IDCAMS BLDINDEX to automatically use
DFSORT to improve the performance of most BLDINDEX jobs that require
BLDINDEX external sorting.
817
818
Chapter 11. Examples of DFSORT job streams

Summary of examples
The following table summarizes the examples provided in this chapter.
No.
Input
Output
Functions/Options
Sort
Disk
Tape
ALTSEQ
Sort
Disk
Disk
OMIT, SUM, OUTREC, DYNALLOC, ZDPRINT
Sort
Tape
Tape
ASCII Tapes
Sort
Tape
Disk
E15, E35, FILSZ, AVGRLEN, DYNALLOC
Sort
Disk
Disk
Program-invoked, SORTCNTL, CHALT,

DYNALLOC, FILSZ
Sort
Disk
Disk
VSAM Input/Output, DFSPARM, Option Override
Sort
Disk
Disk
COBOL E15, EXEC PARM, MSGDDN
Sort
Disk
Disk
Dynamic Link-editing of Exits
Sort
E15
Disk
Extended Parameter List Interface
Sort
10
Disk
Disk and
SYSOUT
OUTFIL
Sort
11
Pipe
Pipes
Pipes, OUTFIL SPLIT, FILSZ, DYNALLOC
Sort
12
Disk
Disk
INCLUDE, LOCALE
Sort
13
z/OS UNIX files z/OS UNIX file
Sort
14
Disk
Disk
IFTHEN
Sort
15
E15
Disk
64-bit parameter list interfaces
Merge
Disk
Disk
EQUALS
Merge
Disk
Disk
LOCALE, OUTFIL
Copy
Tape
Disk
EXEC PARMs, SKIPREC, MSGPRT, ABEND
Copy
Disk
Disk
INCLUDE, VLSHRT
Copy
Disk
Disk
OUTREC, PARSE, BUILD
ICEGENER
Disk
Disk
ICETOOL
Disk
Disk
OCCUR, COPY, SORT, MODE, VERIFY, STATS,

DISPLAY
Storage administrator examples

DFSORT provides a set of sample jobs that demonstrate techniques of interest to
Storage Administrators and others who analyze data collected from DFSMShsm,
DFSMSrmm, DCOLLECT and SMF. These sample jobs can be found in the
ICESTGEX member of the SICESAMP library (contact your System Programmer for
details). You can also download these sample jobs from the DFSORT FTP site.
These sample jobs show some of the many ways DFSORT features such as
ICETOOL and OUTFIL can be used to analyze data and generate reports:
DCOLEX1
DCOLLECT Example 1: VSAM report
819
Summary of Examples
DCOLEX2
DCOLLECT Example 2: Conversion reports
DCOLEX3
DCOLLECT Example 3: Capacity planning analysis and reports
DFHSMEX1
DFHSM Example 1: Deciphering Activity Logs
DFHSMEX2
DFHSM Example 2: Recover a DFHSM CDS with a broken index
RMMEX1
DFSMSrmm Example 1: SMF audit report
RMMEX2
DFSMSrmm Example 2: Create ADDVOLUME commands
REXX examples
Both DFSORT and ICETOOL can be called from REXX. The key is to specify
ALLOCATE statements for the data sets you need and then use an ADDRESS
ADDRESS LINKMVS name
which says to fetch the named program using the standard system search list.
Here is an example of a REXX CLIST to call DFSORT:
/* Simple REXX CLIST to call DFSORT */
"FREE FI(SYSOUT SORTIN SORTOUT SYSIN)"
"ALLOC FI(SYSOUT) DA(*)"
"ALLOC FI(SORTIN) DA(Y897797.INS1) REUSE"
"ALLOC FI(SORTOUT) DA(Y897797.OUTS1) REUSE"
"ALLOC FI(SYSIN)
DA(Y897797.SORT.STMTS) SHR REUSE"
ADDRESS LINKMVS ICEMAN
Here are the DFSORT control statements that might appear in the
Y897797.SORT.STMTS data set:
INCLUDE COND=(21,3,SS,EQ,CL92,J82,M72)
Here is an example of a REXX CLIST to call ICETOOL:

/* Simple REXX CLIST to call ICETOOL */
"FREE FI(TOOLMSG DFSMSG VLR LENDIST TOOLIN)"
"ALLOC FI(TOOLMSG)
DA(*)"
"ALLOC FI(DFSMSG) DUMMY"
"ALLOC FI(VLR)
DA(Y897797.VARIN) REUSE"
"ALLOC FI(LENDIST)
DA(*)"
"ALLOC FI(TOOLIN)
DA(Y897797.TOOLIN.STMTS) SHR REUSE"
ADDRESS LINKMVS ICETOOL
Here are the ICETOOL statements that might appear in the

Y897797.TOOLIN.STMTS data set:
OCCURS FROM(VLR) LIST(LENDIST) TITLE(LENGTH DISTRIBUTION REPORT) BLANK HEADER(LENGTH) HEADER(NUMBER OF RECORDS) ON(VLEN)
ON(VALCNT)
820
Summary of Examples
CLIST examples
Both DFSORT and ICETOOL can be called from a CLIST. They key is to specify
ALLOCATE statements for the data sets you need and then use a CALL statement
like this:
CALL *(name)
Here is an example of a CLIST to call DFSORT:

FREE FI(SYSOUT SORTIN SORTOUT SYSIN)
ALLOC FI(SYSOUT) DA(*)
ALLOC FI(SORTIN) DA(Y897797.INS1) REUSE
ALLOC FI(SORTOUT) DA(Y897797.OUTS1) REUSE
ALLOC FI(SYSIN) DA(Y897797.SORT.STMTS) SHR REUSE
CALL *(ICEMAN)
Here are the DFSORT control statements that might appear in the
Y897797.SORT.STMTS data set:
INCLUDE COND=(21,3,SS,EQ,CL92,J82,M72)
Here is an example of a CLIST to call ICETOOL:

FREE FI(TOOLMSG DFSMSG VLR LENDIST TOOLIN)
ALLOC FI(TOOLMSG) DA(*)
ALLOC FI(DFSMSG) DUMMY
ALLOC FI(VLR) DA(Y897797.VARIN) REUSE
ALLOC FI(LENDIST) DA(*)
ALLOC FI(TOOLIN) DA(Y897797.TOOLIN.STMTS) SHR REUSE
CALL *(ICETOOL)
Here are the ICETOOL statements that might appear in the

Y897797.TOOLIN.STMTS data set:
OCCURS FROM(VLR) LIST(LENDIST) TITLE(LENGTH DISTRIBUTION REPORT) BLANK HEADER(LENGTH) HEADER(NUMBER OF RECORDS) ON(VLEN)
ON(VALCNT)
Sort examples
This section includes 14 sort examples.
Example 1. Sort with ALTSEQ

INPUT
Blocked variable-length records on disk
OUTPUT
Blocked variable-length records on 3490
WORK DATA SETS
Two 3390 data sets
USER EXITS
None
FUNCTIONS/OPTIONS
ALTSEQ
//EXAMP
//S1
//SYSOUT
//SORTIN
JOB A400,PROGRAMMER
EXEC PGM=SORT
DD SYSOUT=A
DD DSN=A123456.IN5,DISP=SHR
01
02
03
04
821
Sort Examples
//SORTOUT DD DSN=OUT1,UNIT=3490,DISP=(,KEEP),VOL=SER=VOL001
//SORTWK01 DD UNIT=3390,SPACE=(CYL,(10,10))
//SORTWK02 DD UNIT=3390,SPACE=(CYL,(10,10))
//SYSIN
DD *
* COLLATE $, # and @ AFTER Z
ALTSEQ CODE=(5BEA,7BEB,7CEC)
05
06
07
08
09
10
11
Line
Explanation
01
JOB statement. Introduces this job to the operating system.
02
EXEC statement. Calls DFSORT directly by its alias SORT.
03
SYSOUT DD statement. Directs DFSORT messages and control statements

to system output class A.
04
SORTIN DD statement. The input data set is named A123456.IN5 and is

cataloged. DFSORT determines from the data set label that the RECFM is
VB, the maximum LRECL is 120, and the BLKSIZE is 2200.
05
SORTOUT DD statement. The output data set is named OUT1 and is to be

allocated on 3490 volume VOL001 and kept. DFSORT sets the RECFM and
LRECL from SORTIN and selects an appropriate BLKSIZE for this standard
labeled tape.
06
SORTWK01 DD statement. The first work data set is allocated on 3390.
07
SORTWK02 DD statement. The second work data set is allocated on 3390.
08
SYSIN DD statement. DFSORT control statements follow.
09
Comment statement. Printed but otherwise ignored.
10
SORT statement. FIELDS specifies an ascending 5-byte character control

field starting at position 7 (the third data byte, because the RDW occupies
the first 4 bytes). The control field is to be collated according to the
modified sequence described in the ALTSEQ statement.
11
ALTSEQ statement. CODE specifies that the three characters $, # and @ are
to collate in that order after Z.
Example 2. Sort with OMIT, SUM, OUTREC, DYNALLOC and

ZDPRINT
INPUT
Blocked fixed-length records on 3380 and 3390
OUTPUT
Blocked fixed-length records on 3390
WORK DATA SETS
Dynamically allocated
USER EXITS
None
FUNCTIONS/OPTIONS
OMIT, OUTREC, SUM, DYNALLOC, ZDPRINT
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC PGM=SORT
//SYSOUT
DD SYSOUT=H
//SORTIN
DD DSN=INP1,DISP=SHR,UNIT=3380,VOL=SER=SCR001
//
DD DSN=INP2,DISP=SHR,UNIT=3390,VOL=SER=SYS351
//SORTOUT DD DSN=&&OUTPUT,DISP=(,PASS),UNIT=3390,
//
SPACE=(CYL,(5,1)),DCB=(LRECL=22)
822
01
02
03
04
05
06
07
Sort Examples
//SYSIN
DD *
OMIT COND=(5,1,CH,EQ,CM)
SORT FIELDS=(20,8,CH,A,10,3,FI,D)
SUM FIELDS=(16,4,ZD)
OPTION DYNALLOC,ZDPRINT
OUTREC FIELDS=(10,3,20,8,16,4,2Z,5,1,C SUM)
08
09
10
11
12
13
Line
Explanation
01
02
03

to system output class H.
04-05
SORTIN DD statement. Consists of a concatenation of two data sets. The

first input data set is named INP1 and resides on 3380 volume SCR001.
The second input data set is named INP2 and resides on 3390 volume
SYS351. DFSORT determines from the data set labels that the record format
is FB, the LRECL is 80 and the largest BLKSIZE is 27920.
06-07
SORTOUT DD statement. The output data set is temporary and is to be

allocated on a 3390. Because the OUTREC statement results in a
reformatted output record length of 22 bytes, LRECL=22 must be specified.
DFSORT sets the RECFM from SORTIN and selects an appropriate
BLKSIZE.
08
09
OMIT statement. COND specifies that input records with a character M in

position 5 are to be omitted from the output data set.
10

field starting at position 20 and a descending 3-byte fixed-point control
field starting at position 10.
11
SUM statement. FIELDS specifies a 4-byte zoned-decimal summary field

starting at position 16. Whenever two records with the same control fields
(specified in the SORT statement) are found, their summary fields
(specified in the SUM statement) are to be added and placed in one of the
records, and the other record is to be deleted.
12
OPTION statement. DYNALLOC specifies that work data sets are to be

dynamically allocated using the installation defaults for the type of device
and number of devices. ZDPRINT specifies that positive ZD SUM fields
are to be printable.
13
OUTREC statement. FIELDS specifies how the records are to be

reformatted for output. The reformatted records are 22 bytes long and look
as follows:
Position
Content
1-3
4-11
12-15
16-17
Zeros
18
Input position 5
19-22
The character string ' SUM'

823
Sort Examples
Example 3. Sort with ASCII tapes

INPUT
Variable-length ASCII records on 3590
OUTPUT
Variable-length ASCII records on 3590
WORK DATA SETS
One SYSDA data set
USER EXITS
None
FUNCTIONS/OPTIONS
None
//EXAMP
JOB A400,PROGRAMMER
//RUNIT
EXEC SORTD
//SORTIN
DD DSN=SRTFIL,DISP=(OLD,DELETE),UNIT=3590,
// DCB=(RECFM=D,LRECL=400,BLKSIZE=404,OPTCD=Q,BUFOFF=L),
// VOL=SER=311500,LABEL=(1,AL)
//SORTOUT DD DSN=OUTFIL,UNIT=3590,LABEL=(,AL),DISP=(,KEEP),
// DCB=(BLKSIZE=404,OPTCD=Q,BUFOFF=L),VOL=SER=311501
//SORTWK01 DD UNIT=SYSDA,SPACE=(CYL,(4))
//SYSIN
DD *
SORT FIELDS=(10,8,AC,D)
RECORD TYPE=D,LENGTH=(,,,20,80)
824
01
02
03
04
05
06
07
08
09
10
11
Line
Explanation
01
02
EXEC statement. Uses the SORTD cataloged procedure to call DFSORT

directly.
03-05
SORTIN DD statement. The input data set is named SRTFIL and resides on
3590 volume 311500. It is to be deleted after this job step. It has a RECFM
of D (variable-length ASCII records), a maximum LRECL of 400, a
BLKSIZE of 404 and an ASCII label. For this job, the buffer offset is the
block length indicator. The records are to be translated from ASCII to
EBCDIC.
06-07
SORTOUT DD statement. The output data set is named OUTFIL and is to

be allocated on 3590 volume 311501 and kept. It is to be written with an
ASCII label. DFSORT sets the RECFM and LRECL from SORTIN and sets
the BLKSIZE to 404 as indicated in the DD statement. For this job, the
buffer offset is the block length indicator. The records are to be translated
from EBCDIC to ASCII.
08
SORTWK01 DD statement. The work data set is allocated on SYSDA.
09
10
SORT statement. FIELDS specifies a descending 8-byte ASCII control field

starting at position 10.
11
RECORD statement. TYPE specifies ASCII variable-length records.

LENGTH specifies that the minimum record length is 20 and the average
record length is 80.
Sort Examples
Example 4. Sort with E15, E35, FILSZ, AVGRLEN and

DYNALLOC
INPUT
Variable-length records on 3490
OUTPUT
Blocked variable-length records on SYSDA
WORK DATA SETS
USER EXITS
E15 and E35
FUNCTIONS/OPTIONS
FILSZ, AVGRLEN, DYNALLOC
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC PGM=ICEMAN
//SYSOUT
DD SYSOUT=A
//SORTIN
DD DSN=INPUT,VOL=SER=FLY123,
// UNIT=3490,DISP=OLD
//SORTOUT DD DSN=&&OUT,DISP=(,PASS),SPACE=(CYL,(10,12)),
// UNIT=SYSDA,DCB=(RECFM=VB)
//MODLIB
DD DSN=EXIT1.RTNS,DISP=SHR
//
DD DSN=EXIT2.RTNS,DISP=SHR
//SYSIN
DD *
SORT FIELDS=(23,4,PD,A,10,6,FS,A)
OPTION DYNALLOC=(3390,3),AVGRLEN=75,FILSZ=E50000
MODS E15=(MODREC,1024,MODLIB),E35=(ADDREC,1200,MODLIB)
01
02
03
04
05
06
07
08
09
10
11
12
13
Line
Explanation
01
02
EXEC statement. Calls DFSORT directly.
03

04-05
SORTIN DD statement. The input data set is named INPUT and resides on
3490 volume FLY123. DFSORT determines from the data set label of this
standard labeled tape that the RECFM is V, the LRECL is 120 and the
BLKSIZE is 124.
06-07

allocated on SYSDA. Because the input is unblocked and the output is to
be blocked, RECFM=VB must be specified. DFSORT sets the LRECL from
SORTIN and selects an appropriate BLKSIZE.
08-09
MODLIB DD statement. Specifies the load libraries that contain the exit
routines. When exit routines reside in more than one library, the libraries
must be concatenated using a single DD statement.
10
11
SORT statement. FIELDS specifies an ascending 4-byte packed-decimal

control field starting at position 23 and an ascending 6-byte floating-sign
control field starting at position 10.
12
OPTION statement. DYNALLOC=(3390,3) specifies that three 3390 work

data sets are to be allocated. AVGRLEN=75 specifies an average record
length of 75. AVGRLEN helps DFSORT optimize work space for
variable-length record input. FILSZ=E50000 specifies an estimate of 50000
records. Because the 3490 input data set is compacted, DFSORT might not
825
Sort Examples
be able to determine the file size accurately unless the data set is managed
by DFSMSrmm or a tape management system that uses ICETPEX.
Specification of FILSZ can make a significant difference in work space
optimization when tape input data sets are not managed.
13
MODS statement. E15 specifies a user exit routine named MODREC.

Approximately 1024 bytes are required for MODREC and the system
services (for example, GETMAIN and OPEN) it performs. E35 specifies a
user exit routine named ADDREC. Approximately 1200 bytes are required
for ADDREC and the system services it performs. MODREC and ADDREC
reside in the libraries defined by the MODLIB DD statement.
Example 5. Called sort with SORTCNTL, CHALT, DYNALLOC

and FILSZ
INPUT
Blocked fixed-length records on disk
OUTPUT
WORK DATA SETS
USER EXITS
None
FUNCTIONS/OPTIONS
CHALT, DYNALLOC, FILSZ
//EXAMP
//RUNSORT
//STEPLIB
//SYSOUT
//SYSPRINT
//SORTIN
//SORTOUT
//SORTCNTL
OPTION
826
JOB A400,PROGRAMMER
EXEC PGM=MYPGM
DD DSN=M999999.LOAD,DISP=SHR
DD SYSOUT=A
DD SYSOUT=A
DD DSN=M999999.INPUT(MASTER),DISP=OLD
DD DSN=M999999.OUTPUT.FILE,DISP=OLD
DD *
CHALT,DYNALLOC=(,3),FILSZ=U25000
01
02
03
04
05
06
07
08
09
Line
Explanation
01
02
EXEC statement. Calls a program named MYPGM that in turn calls

DFSORT.
03
STEPLIB DD statement. Specifies the load library that contains MYPGM.
04

05
SYSPRINT DD statement. Directs MYPGM output to system output class

A.
06
SORTIN DD statement. The input data set is member MASTER in the

cataloged partitioned data set M999999.INPUT. DFSORT determines the
RECFM, LRECL and BLKSIZE from the data set label.
07
SORTOUT DD statement. The output data set is named

M999999.OUTPUT.FILE and is cataloged. DFSORT determines the RECFM,
LRECL and BLKSIZE from the data set label.
Sort Examples
08
SORTCNTL DD statement. DFSORT control statements follow. Statements

in SORTCNTL override or supplement statements passed by MYPGM in
the DFSORT parameter list it uses.
09
OPTION statement. CHALT specifies that character format control fields

(specified in the SORT statement passed by MYPGM) are to be sorted
using the ALTSEQ installation option. DYNALLOC=(,3) specifies that three
work data sets are to be dynamically allocated using the installation
default for the type of device. FILSZ=U25000 specifies a file size of 25000
records is to be used by DFSORT to determine the amount of work space
needed. Because the input data set is a member of a PDS, specifying FILSZ
helps DFSORT optimize work data set space.
Example 6. Sort with VSAM input/output, DFSPARM and

option override
INPUT
VSAM TYPE=V records
OUTPUT
VSAM TYPE=V records
WORK DATA SETS
USER EXITS
None
FUNCTIONS/OPTIONS
Override of Various Options
//EXAMP
JOB A400,PROGRAMMER
//S1
EXEC PGM=SORT
//SYSOUT
DD SYSOUT=A
//SORTIN
DD DSN=TEST.SORTIN.FILE,DISP=SHR
//SORTOUT DD DSN=TEST.SORTOUT.FILE,DISP=SHR
//DFSPARM DD *
RECORD TYPE=V
OPTION HIPRMAX=10,DYNALLOC=3390,MAINSIZE=3M,
MSGPRT=CRITICAL,NOLIST
01
02
03
04
05
06
07
08
09
10
For purposes of illustration, assume that none of the standard installation defaults
for batch direct invocation of DFSORT have been changed by the site.
Line
Explanation
01
02
03

04
SORTIN DD statement. The input data set is TEST.SORTIN.FILE. DFSORT

determines that it is a VSAM data set and obtains its attributes from the
catalog.
05
SORTOUT DD statement. The output data set is TEST.SORTOUT.FILE.

DFSORT determines that it is a VSAM data set and obtains its attributes
from the catalog.
06
DFSPARM DD statement. DFSORT control statements follow. DFSPARM

can be used for both direct-invocation and program-invocation of DFSORT
827
Sort Examples
and overrides options and statements from all other sources. Certain
operands, such as MSGPRT and LIST/NOLIST, are used if supplied in
DFSPARM, the EXEC PARM or the invocation parameter list, but not used
if supplied in SYSIN or SORTCNTL.
07
RECORD statement. TYPE=V specifies that DFSORT is to treat the VSAM

records as variable-length. In this case, the RECORD statement could be
omitted, because DFSORT would automatically set a record type of V due
to the use of VSAM data sets for SORTIN and SORTOUT.
08
SORT statement. FIELDS specifies an ascending 4-byte binary control field

starting at position 30. This position corresponds to a specification of
KEYS(4 25) for the VSAM CLUSTER (4 bytes at offset 25, which is
equivalent to position 26 with 4 bytes added for the RDW that DFSORT
supplies at input and removes at output for VSAM TYPE=V records).
09-10
OPTION statement. HIPRMAX=10 specifies that up to 10 MBs of

Hiperspace can be used for Hipersorting, overriding the standard
installation default of HIPRMAX=OPTIMAL. DYNALLOC=3390 specifies
that work data sets are to be allocated on 3390s, overriding the standard
installation default of SYSDA. The standard installation default of four
work data sets is not overridden. MAINSIZE=3M specifies that up to 3
MBs of storage can be used, overriding the standard installation default of
MAINSIZE=MAX. MSGPRT=CRITICAL specifies that only error messages
are to be printed, overriding the standard installation default of
MSGPRT=ALL. NOLIST specifies that control statements are not to be
printed, overriding the standard installation default of LIST=YES.
Example 7. Sort with COBOL E15, EXEC PARM and MSGDDN

INPUT
Fixed-length records on disk
OUTPUT
Fixed-length records on SYSDA
WORK DATA SETS
None
USER EXITS
COBOL E15
FUNCTIONS/OPTIONS
MSGDDN
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC PGM=SORT,PARM=MSGDDN=DFSOUT
//STEPLIB DD DSN=SYS1.SCEERUN,DISP=SHR
//SYSOUT
DD SYSOUT=A
//DFSOUT
DD SYSOUT=A
//EXITC
DD DSN=COBEXITS.LOADLIB,DISP=SHR
//SORTIN
DD DSN=SORT1.IN,DISP=SHR
//SORTOUT DD DSN=&&OUT,DISP=(,PASS),SPACE=(CYL,(5,5)),
// UNIT=SYSDA,DCB=(LRECL=120)
//SYSIN
DD *
SORT FIELDS=(5,4,A,22,2,A),FORMAT=BI
MODS E15=(COBOLE15,37000,EXITC,C)
RECORD LENGTH=(,120)
828
01
02
03
04
05
06
07
08
09
10
11
12
13
Line
Explanation
01
02
EXEC statement. Calls DFSORT directly by its alias name SORT.
Sort Examples
MSGDDN=DFSOUT specifies an alternate message data set for DFSORT
messages and control statements to prevent the COBOL messages in
SYSOUT from being interleaved with the DFSORT messages and control
statements.
03
STEPLIB statement. Specifies the Language Environment library.
04
SYSOUT statement. Directs COBOL messages to system output class A.
05
DFSOUT statement. Directs DFSORT messages and control statements to

system output class A (this is the alternate message data set specified by
MSGDDN in the PARM field of the EXEC statement).
06
EXITC statement. Specifies the load library that contains the exit routine.
07
SORTIN DD statement. The input data set is named SORT1.IN and is

cataloged. DFSORT determines from the data set label that the RECFM is F,
the LRECL is 100 and the BLKSIZE is 100.
08-09

allocated on SYSDA. Because the E15 exit changes the length of the records
from 100 bytes to 120 bytes, LRECL=120 must be specified. DFSORT sets
the RECFM from SORTIN and sets the BLKSIZE to the LRECL (unblocked
records).
10
11
SORT statement. FIELDS specifies an ascending 4-byte control field starting

at position 5 and an ascending 2-byte control field starting at position 22.
FORMAT specifies that the control fields are binary.
12
MODS statement. E15 specifies a user exit routine named COBOLE15

written in COBOL. Approximately 37000 bytes are required for the exit, the
system services (for example, GETMAIN and OPEN) it performs, and the
COBOL library subroutines. COBOLE15 resides in the library defined by
the EXITC DD statement.
13
RECORD statement. LENGTH specifies that the COBOL E15 routine

changes the length of the records to 120 bytes.
Example 8. Sort with dynamic link-editing of exits

INPUT
OUTPUT
WORK DATA SETS
One SYSDA data set
USER EXITS
E11, E15, E17, E18, E19, E31, E35, E38, E39
FUNCTIONS/OPTIONS
None
//EXAMP
JOB A400,PROGRAMMER
//STEPA
EXEC SORT
//SORTIN
DD DSN=SMITH.INPUT,DISP=SHR
//SORTOUT DD DSN=SMITH.OUTPUT,DISP=(NEW,CATLG),
// UNIT=3380,SPACE=(TRK,(10,2)),VOL=SER=XYZ003
//SORTWK01 DD UNIT=SYSDA,SPACE=(CYL,(1,1))
//EXIT
DD DSN=SMITH.EXIT.OBJ,DISP=SHR
//EXIT2
DD DSN=SMITH.EXIT2.OBJ,DISP=SHR
01
02
03
04
05
06
07
08
829
Sort Examples
//SORTMODS DD UNIT=SYSDA,SPACE=(TRK,(10,,3))
//SYSIN
DD *
SORT FIELDS=(1,8,CH,A,20,4,BI,D)
MODS E11=(EXIT11,1024,EXIT,S),
E15=(E15,1024,SYSIN,T),
E17=(EXIT17,1024,EXIT2,T),
E18=(EXIT18,1024,EXIT,T),
E19=(E19,1024,SYSIN,T),
E31=(PH3EXIT,1024,EXIT,T),
E39=(E39,1024,SYSIN,T)
END
<object deck for E15 exit here>
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Line
Explanation
01
02
EXEC statement. Uses the SORT cataloged procedure to call DFSORT

directly and supply the DD statements (not shown) required by the linkage
editor.
03
SORTIN DD statement. The input data set is named SMITH.INPUT and is

cataloged. DFSORT determines the RECFM, LRECL and BLKSIZE from the
data set label.
04-05
SORTOUT DD statement. The output data set is named SMITH.OUTPUT

and is to be allocated on 3380 volume XYZ003 and cataloged. DFSORT sets
the RECFM and LRECL from SORTIN and selects an appropriate BLKSIZE.
06
SORTWK01 DD statement. The work data set is allocated on SYSDA.
07
EXIT DD statement. Specifies the partitioned data set containing the object
decks for the E11, E18, E31, E35 and E38 exit routines.
08
EXIT2 DD statement. Specifies the partitioned data set containing the

object deck for the E17 exit routine.
09
SORTMODS DD statement. The partitioned data set to hold exit routine

object decks from SYSIN for input to the linkage editor is to be allocated
on SYSDA.
10
SYSIN DD statement. DFSORT control statements, and object decks to be

used by the linkage editor, follow.
11

field starting at position 1 and a descending 4-byte binary control field
12-20
MODS statement. Specifies the exit routines to be used, the approximate

number of bytes required for each exit and that:
v The EXIT11 routine in the EXIT library is to be link-edited separately
from other input phase exit routines and associated with user exit E11.
v The E15 and E19 routines in SYSIN, the EXIT17 routine in EXIT2, and
the EXIT18 routine in EXIT are to be link-edited together and associated
with user exits E15, E19, E17, and E18, respectively.
v The E31, E35, and E38 routines in the PH3EXIT object deck and the E39
routine in SYSIN are to be link-edited together and associated with user
exits E31, E35, E38, and E39, respectively.
830
Sort Examples
21
END statement. Marks the end of the DFSORT control statements and the
beginning of the exit routine object decks.
22-24
Object decks. The three object decks for the E15, E19, and E39 exit routines
follow the END statement.
Example 9. Sort with the extended parameter list interface

INPUT
Fixed-length records from E15
OUTPUT
Blocked fixed-length records on SYSDA
WORK DATA SETS
USER EXITS
E15
FUNCTIONS/OPTIONS
OMIT, FILSZ, RESINV, DYNALLOC
The JCL for running program MYSORT, and highlights of the code used by
MYSORT to invoke DFSORT with the extended parameter list, are shown below.
For purposes of illustration, assume that none of the standard installation defaults
for batch program invocation of DFSORT have been changed by the site.
//EXAMP
JOB A400,PROGRAMMER
01
//STEP1
EXEC PGM=MYSORT
02
//SYSOUT
DD SYSOUT=C
03
//MSGOUT
DD SYSOUT=C
04
//STEPLIB DD DSN=A123456.LOAD,DISP=SHR
05
//SORTOUT DD DSN=&&OUTPUT,DISP=(,PASS),UNIT=SYSDA,
06
// SPACE=(CYL,(8,4))
07
//SORTCNTL DD *
08
* Update file size estimate
09
OPTION FILSZ=E30000
10
--------------------------------------------------------------------------------------------------------------------------------------------MYSORT CSECT
11
.
.
.
LA
R1,PL1
SET ADDRESS OF PARAMETER LIST
12
*
TO BE PASSED TO DFSORT
13
ST
R2,PL4
SET ADDRESS OF GETMAINED AREA
14
*
TO BE PASSED TO E15
15
LINK EP=SORT
INVOKE DFSORT
16
.
.
.
PL1
DC
A(CTLST)
ADDRESS OF CONTROL STATEMENTS
17
PL2
DC
A(E15)
ADDRESS OF E15 ROUTINE
18
PL3
DC
A(0)
NO E35 ROUTINE
19
PL4
DS
A
USER EXIT ADDRESS CONSTANT
20
PL5
DC
F-1
INDICATE END OF LIST
21
CTLST DS
0H
CONTROL STATEMENTS AREA
22
DC
AL2(CTL2-CTL1)
LENGTH OF CHARACTER STRING
23
CTL1
DC
C SORT FIELDS=(5,8,CSF,A)
24
DC
C RECORD TYPE=F,LENGTH=80
25
DC
C OPTION FILSZ=E25000,DYNALLOC,
26
DC
CRESINV=8000
27
DC
C OMIT FORMAT=CSF,COND=(5,8,EQ,13,8)
28
CTL2
EQU *
29
831
Sort Examples
OUT
E15
832
DCB DDNAME=MSGOUT,...
DS
0H
E15 ROUTINE
.
.
.
L
R2,4(,R1)
GET ADDRESS OF GETMAINED AREA
.
.
.
BR
R14
RETURN TO DFSORT
.
.
.
30
31
32
33
Line
Explanation
01
02
EXEC statement. Calls a program named MYSORT that in turn calls

DFSORT.
03

to SYSOUT class C.
04
MSGOUT DD statement. Directs MYSORT messages to SYSOUT class C.
05
STEPLIB DD statement. Specifies the load library that contains MYSORT.
06-07

allocated on SYSDA. Because SORTIN is not used, DFSORT sets the
RECFM and LRECL from the RECORD statement and sets the BLKSIZE to
the LRECL (unblocked records).
08
SORTCNTL DD statement. DFSORT control statements follow. Statements

in SORTCNTL override or supplement statements passed by MYSORT in
the extended parameter list it uses.
09
10
OPTION statement. FILSZ=E30000 specifies an estimate of 30000 records,

overriding FILSZ=E25000 in the OPTION statement of the extended
parameter list. Because the E15 routine supplies all of the input records,
DFSORT will not be able to determine the file size accurately; therefore,
specifying FILSZ can make a significant difference in work space
optimization when an E15 routine supplies all of the input records. It's
important to change the FILSZ value whenever the number of input
records increases significantly.
11
This is the start of program MYSORT. Assume that it GETMAINs a work

area, saves its address in register 2, and initializes it with information to be
used by the E15 routine.
12-13
MYSORT places the address of the extended parameter list to be passed to

DFSORT in register 1.
14-15
MYSORT places the address of the GETMAINed work area in the user exit
address constant field in the extended parameter list. DFSORT will pass
this address to the E15 routine (in the second word of the E15 parameter
list) when it is entered.
16
MYSORT calls DFSORT by it's alias SORT.
17-21
The extended parameter list specifies: the address of the control statements
area, the address of the E15 routine, that no E35 routine is present, and the
address of the GETMAINed work area. F'-1' indicates the end of the
Sort Examples
extended parameter list. Subsequent parameter list fields, such as the
address of an ALTSEQ table, are not used in this application.
Because the address of the E15 routine is passed in the parameter list,
SORTIN cannot be used; if a SORTIN DD statement were present, it would
be ignored.
22-23
This is the start of the control statements area. The total length of the
control statements is specified.
24
SORT statement. FIELDS specifies an ascending 8-byte floating-sign control

25
RECORD statement. TYPE=F and LENGTH=80 specify that the E15 inserts
fixed-length records of 80 bytes. In this case, TYPE=F could be omitted,
because DFSORT would automatically set a record type of F. However,
LENGTH must be specified when an E15 supplies all of the input records.
26-27
OPTION statement. FILSZ=E25000 specifies an estimate of 25000 records,

which is overridden by FILSZ=E30000 in SORTCNTL's OPTION statement.
DYNALLOC specifies that work data sets are to be dynamically allocated
using the installation defaults for the type of device and number of
devices. RESINV=8000 specifies that approximately 8000 bytes are required
for the system services (for example, GETMAIN and OPEN) that
MYSORT's E15 exit routine performs.
28
OMIT statement. FORMAT specifies that the compare fields are

floating-sign. COND specifies that input records with equal 8-byte
floating-sign compare fields starting in position 5 (also the control field)
and position 13 are to be omitted from the output data set.
29
This is the end of the control statements area.
30
This is the DCB for MYSORT's MSGOUT output.
31-33
This is MYSORT's E15 routine. The E15 routine loads the address of the
GETMAINed work area from the second word of the E15 parameter list.
The E15 routine must supply each input record by placing its address in
register 1 and placing a 12 (insert) in register 15. When all the records have
been passed, the E15 routine must place an 8 (do not return) in register
15.
Example 10. Sort with OUTFIL

INPUT
Fixed-length record data set
OUTPUT
Multiple fixed-length record data sets
WORK DATA SETS
Dynamically allocated (by default)
USER EXITS
None
FUNCTIONS/OPTIONS
OUTFIL
//EXAMP
//OUTFIL
//SYSOUT
//SORTIN
//ALLGPS
JOB A400,PROGRAMMER
EXEC PGM=SORT
DD SYSOUT=A
DD DSN=GRP.RECORDS,DISP=SHR
DD DSN=GRP.ALLGRPS,DISP=OLD
01
02
03
04
05
833
Sort Examples
//ALLBU
DD DSN=GRP.BU,DISP=(NEW,CATLG,DELETE),
//
UNIT=3390,SPACE=(TRK,(10,10))
//G1STATS DD SYSOUT=A
//G2STATS DD SYSOUT=A
//SYSIN
DD
*
834
06
07
08
09
10
11
OUTFIL FNAMES=(ALLGPS,ALLBU)
12
OUTFIL FNAMES=G1STATS,
INCLUDE=(1,3,CH,EQ,CG01),
HEADER2=(1:GROUP 1 STATUS REPORT FOR ,&DATE,
- PAGE ,&PAGE,2/,
6:ITEM ,16:STATUS
,31:PARTS,/,
6:-----,16:------------,31:-----),
OUTREC=(6:6,5,
16:14,1,CHANGE=(12,
C1,CSHIP,
C2,CHOLD,
C3,CTRANSFER),
31:39,1,BI,M10,LENGTH=5,
120:X)
13
14
15
16
17
18
19
20
21
22
23
24
25
26
OUTFIL FNAMES=G2STATS,
INCLUDE=(1,3,CH,EQ,CG02),
HEADER2=(1:GROUP 2 STATUS REPORT FOR ,&DATE,
- PAGE ,&PAGE,2/,
6:ITEM ,16:STATUS
,31:PARTS,/,
6:-----,16:------------,31:-----),
OUTREC=(6:6,5,
16:14,1,CHANGE=(12,
C1,CSHIP,
C2,CHOLD,
C3,CTRANSFER),
31:39,1,BI,M10,LENGTH=5,
120:X)
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Line
Explanation
01
02
03

to sysout class A.
04
SORTIN DD statement. The input data set is named GRP.RECORDS and is

FB, the LRECL is 80 and the BLKSIZE is 23440.
05
ALLGPS DD statement. The first OUTFIL output data set is named

GRP.ALLGRPS and is catalogued. DFSORT determines the RECFM, LRECL
and BLKSIZE from the data set label.
06-07
ALLBU DD statement. The second OUTFIL output data set is named

GRP.BU and is to be allocated on a 3390 and catalogued. DFSORT sets the
RECFM and LRECL from SORTIN and selects an appropriate BLKSIZE.
08
G1STATS DD statement. The third OUTFIL output data set is directed to

sysout class A. Because this is an OUTFIL report data set, DFSORT sets the
RECFM to FBA (FB from SORTIN and A for ANSI control characters) and
the LRECL to 121 (1 byte for the ANSI control character and 120 bytes for
the data). DFSORT sets an appropriate BLKSIZE.
09
G2STATS DD statement. The fourth OUTFIL output data set is directed to
Sort Examples
sysout class A. Because this is an OUTFIL report data set, DFSORT sets the
RECFM to FBA (FB from SORTIN and A for ANSI control characters) and
the LRECL to 121 (1 byte for the ANSI control character and 120 bytes for
the data). DFSORT sets an appropriate BLKSIZE.
10
11

12
OUTFIL statement. The sorted input records are written to the ALLGPS
and ALLBU data sets.
13-26
OUTFIL statement. The subset of sorted input records containing 'G01' in

positions 1 through 3 are used to produce a report, which is written to the
G1STATS data set.
27-40
OUTFIL statement. The subset of sorted input records containing 'G02' in

positions 1 through 3 are used to produce a report, which is written to the
G2STATS data set.
Example 11. Sort with Pipes and OUTFIL SPLIT

INPUT
Pipes
OUTPUT
Pipes
WORK DATA SETS
USER EXITS
None
FUNCTIONS/OPTIONS
FILSZ, OUTFIL, DYNALLOC
//EXAMP
JOB A400,PROGRAMMER
//RUNSORT EXEC PGM=ICEMAN
//SYSOUT DD SYSOUT=H
//SORTIN DD DSN=INPUT.PIPE,SUBSYS=PIPE,
//
DCB=(LRECL=60,RECFM=FB,BLKSIZE=32760)
//OUT1
DD DSN=OUTPUT.PIPE1,SUBSYS=PIPE,
//
//OUT2
DD DSN=OUTPUT.PIPE2,SUBSYS=PIPE,
//
//SYSIN
DD *
OPTION DYNALLOC,FILSZ=U1000000
SORT FIELDS=(1,20,CH,A,25,4,BI,A)
OUTFIL FNAMES=(OUT1,OUT2),SPLIT
01
02
03
04
05
06
07
08
09
10
11
12
13
Line
Explanation
01
Job statement. Introduces this job to the operating system.
02
03

to system output class H.
04-05
SORTIN DD statement. The SUBSYS=PIPE parameter directs the allocation

to the 'PIPE' subsystem for the pipe named INPUT.PIPE. The DCB
statement describes the data set characteristics to subsystem PIPE.
06-07
OUT1 DD statement. The SUBSYS=PIPE parameter directs the allocation to

835
Sort Examples
the 'PIPE' subsystem for the pipe named OUTPUT.PIPE1. The DCB
08-09
OUT2 DD statement. The SUBSYS=PIPE parameter directs the allocation to

the 'PIPE' subsystem for the pipe named OUTPUT.PIPE2. The DCB
10
11
OPTION statement. DYNALLOC specifies that work data sets are to be

dynamically allocated using the installation defaults for type of device and
number of devices. FILSZ=U1000000 specifies an estimate of one million
input records.
12

field starting at position 1 and an ascending 4 byte binary control field
13
OUTFIL statement. The records from the SORTIN pipe are sorted and
written alternatively to the OUT1 and OUT2 pipes (that is, the sorted
records are split evenly between the two output pipes).
Example 12. Sort with INCLUDE and LOCALE

INPUT
OUTPUT
WORK DATA SETS
USER EXITS
None
FUNCTIONS/OPTIONS
INCLUDE, LOCALE
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC PGM=SORT,PARM=LOCALE=FR_CA
//SYSOUT
DD SYSOUT=A
//SORTIN
DD DSN=INPUT.FRENCH.CANADA,DISP=SHR
//SORTOUT DD DSN=OUTPUT.FRENCH.CANADA,DISP=OLD
//SYSIN
DD *
SORT FIELDS=(1,20,CH,A,25,1,BI,D,30,10,CH,A)
INCLUDE COND=(40,6,CH,EQ,50,6,CH)
836
01
02
03
04
05
06
07
08
09
Line
Explanation
01
02
EXEC statement. Calls DFSORT directly by its alias name SORT. LOCALE
specified in EXEC PARM overrides the installation default for LOCALE.
The locale for the French language and the cultural conventions of Canada
will be active.
03
STEPLIB DD statement. Specifies the Language Environment run-time

library containing the dynamically loadable locales.
04
SYSOUT statement. Directs DFSORT messages and control statements to

sysout class A.
05
SORTIN DD statement. The input data set is named
Sort Examples
INPUT.FRENCH.CANADA and is cataloged. DFSORT determines the
06

OUTPUT.FRENCH.CANADA and is cataloged. DFSORT determines the
07
08

field starting at position 1, a one-byte descending binary control field
starting at position 25, and a 10-byte ascending character control field
starting at position 30. The character (CH) control fields will be sorted
according to the collating rules defined in locale FR_CA.
09
INCLUDE statement. COND specifies that only input records with equal
6-byte character compare fields starting in position 40 and position 50 are
to be included in the output data set. The character (CH) compare fields
will be compared according to the collating rules defined in locale FR_CA.
Example 13: Sort with z/OS UNIX files

INPUT
Concatenated z/OS UNIX Files
OUTPUT
z/OS UNIX File
WORK DATA SETS
USER EXITS
None
FUNCTIONS/OPTIONS
None
//EXAMP
JOB A400,PROGRAMMER
//S1
EXEC PGM=SORT
//SORTIN DD PATH=/user/hfs.inp1.txt,PATHOPTS=ORDONLY,
//
LRECL=80,BLKSIZE=240,RECFM=FB,FILEDATA=TEXT
//
DD PATH=/user/hfs.inp2.txt,PATHOPTS=ORDONLY,
//
LRECL=80,BLKSIZE=80,RECFM=F,FILEDATA=TEXT
//SORTOUT DD PATH=/user/hfs.ut.txt,PATHOPTS=OWRONLY,
//
LRECL=80,BLKSIZE=80,RECFM=F,FILEDATA=TEXT
//SYSIN DD *
01
02
03
04
05
06
07
08
09
10
11
Line
Explanation
01
02
03

04-05
SORTIN DD statement. The first input file is a z/OS UNIX file named
/user/hfs.inp1.txt. Only read access is allowed. The file is defined as a text
file. It has fixed-length records with a record size of 80 and a block size of
240.
06-07
The second input file is a z/OS UNIX file named /user/hfs.inp2.txt. Only
read access is allowed. The file is defined as a text file. It has fixed-length
records with a record size of 80 and a block size of 80.
837
Sort Examples
08-09
SORTOUT DD statement. The output file is a z/OS UNIX file named

/user/hfs.ut.txt. Only write access is allowed. The file is defined as a text
file. It has fixed-length records with a record size of 80 and a block size of
80.
10
11

Example 14. Sort with IFTHEN

INPUT
OUTPUT
WORK DATA SETS
USER EXITS
FUNCTIONS/OPTIONS
Three fixed-length record data sets

None
IFTHEN

//S1
EXEC PGM=SORT
//SORTIN DD DSN=INPUT.FILE1,DISP=SHR
//
DD DSN=INPUT.FILE2,DISP=SHR
//
DD DSN=INPUT.FILE3,DISP=SHR
//SORTOUT DD DSN=OUTPUT.FILE,DISP=(NEW,CATLG,DELETE),
//
SPACE=(CYL,(5,5)),UNIT=SYSDA
//SYSIN DD *
INREC IFTHEN=(WHEN=(1,3,CH,EQ,CHDR),
OVERLAY=(6:YDDD=(D4/),81:C0,82:SEQNUM,2,ZD)),
IFTHEN=(WHEN=(1,3,CH,EQ,CTRL),
OVERLAY=(11:YDDD=(D4/),81:C9,82:SEQNUM,2,ZD)),
IFTHEN=(WHEN=NONE,
OVERLAY=(81:C1))
SORT FIELDS=(81,1,CH,A,8,5,CH,A)
OUTFIL REMOVECC,
OMIT=(81,1,SS,EQ,C0,9,AND,82,2,ZD,GT,+1),
OUTREC=(1,80)
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
This example shows how you can use three input files, each with a header record
('HDR'), detail records ('DTL') and a trailer record ('TRL'), and create an output file
with one header record with the current date, the sorted detail records, and one
trailer record with the current date.
838
01
02
03

to sysout class A.
04-06
SORTIN DD statement. Consists of a concatenation of three input data sets:

INPUT.FILE1, INPUT.FILE2 and INPUT.FILE3. DFSORT determines from
the data set labels that each data set has RECFM=FB and LRECL=80. The
BLKSIZEs vary. Each input data set has a header record, detail records, and
a trailer record.
07-08
SORTOUT DD statement. Creates a new output data set: OUTPUT.FILE.

DFSORT sets RECFM=FB, LRECL=80 and selects an appropriate BLKSIZE.
The output data set will have one header record, the sorted detail records,
and one trailer record.
09
Sort Examples
10-15
INREC statement. The first IFTHEN WHEN=(logexp) clause identifies and

operates on header records ('HDR' in positions 1-3); OVERLAY puts today's
date in the form 'ddd/yyyy' in positions 6-13, adds a '0' in position 81,
adds a ZD sequence number in positions 82-83 and does not affect the rest
of the record.
The second IFTHEN WHEN=(logexp) clause identifies and operates on
trailer records ('TRL' in positions 1-3); OVERLAY puts today's date in the
form 'ddd/yyyy' in positions 11-18, adds a '9' in position 81, adds a ZD
sequence number in positions 82-83 and does not affect the rest of the
record.
The IFTHEN WHEN=NONE clause identifies and operates on detail
records (not 'HDR' or 'TRL' in positions 1-3); OVERLAY adds a '1' in
position 81 and does not affect the rest of the record.
DFSORT extends the reformatted input records from 80 bytes to 83 bytes to
accommodate the identifier byte added in position 81 and the sequence
number added in positions 82-83.
The '0', '1' or '9' identifier byte added in position 81 allows us to sort the
header records ('0') first, followed by the detail records ('1'), and then the
trailer records ('9'). The sequence number added in positions 82-83 will
allow us to keep only the first header record and the first trailer record.
The sequence number will be 1 for the first header record, 2 for the second
header record and 3 for the third header record. Likewise, the sequence
number will be 1 for the first trailer record, 2 for the second trailer record
and 3 for the third trailer record. Since the sequence number is not
specified for the detail records, it will be blank.
16

field at position 81 (the identifier byte added by INREC), and an ascending
5-byte character control field starting at position 8 (the key for the detail
records).
17-19
OUTFIL statement. REMOVECC removes the ANSI carriage control

characters and ensures that the RECFM is FB rather than FBA. OMIT
specifies that reformatted output records with '0' or '9' in position 81
(header or trailer records) and a sequence number in positions 82-83
greater than 1 (second and subsequent header or trailer records), are
omitted. OUTREC keeps only positions 1-80 for the OUTFIL output
records, thus removing the identifier byte and sequence number we added
in positions 81-83 with the INREC statement (we do not want these
temporary fields in the OUTFIL output records).
Example 15. Sort with 64-bit parameter lists, E15, E35 and
OUTFIL
INPUT
Variable-length records from E15
OUTPUT
Variable-length output data set
WORK DATA SETS
USER EXITS
E15 and E35
839
Sort Examples
FUNCTIONS/OPTIONS
64-bit parameter lists, OUTFIL
The JCL for running program PGM1, and the code used by PGM1 to invoke
DFSORT with the 64-bit invocation parameter list and use the 64-bit E15 and E35
parameter lists, are shown below.
//EXAMP
JOB A400,PROGRAMMER
01
//STEP1
EXEC PGM=PGM1
02
//SYSOUT
DD SYSOUT=A
03
//SORTOUT DD DSN=PL64.OUTPUT,DISP=(NEW,CATLG,DELETE),
04
05
--------------------------------------------------------------------------------------------------------------------------------------------PGM1
CSECT
06
PGM1
AMODE 64
PGM1
RMODE 31
STM
14,12,12(13)
SAVE LOW PART REGS
BASR 15,0
SET THE TEMPORARY
USING *,15
BASE REG
M0
DS
0H
STMH 0,15,HIGHREGS
SAVE HIGH PART REGS
LLGTR 12,15
SET THE PERMANENT
DROP 15
BASE
USING M0,12
REG
LMH
0,15,HIGHZERO
SET HIGH PART OF REGS TO ZERO
ST
13,SAVE+4
SAVE SAVE AREA ADDR
LR
11,13
RELOAD SAVE AREA ADDR
LA
13,SAVE
LOAD NEW SAVE AREA ADDR
ST
13,8(11)
SAVE NEW SA ADDR INTO OLD SA
* Obtain 64-bit storage (memory object):
* - for 64-bit Invocation Parameter List
* - for 64-bit Control Statements Area
* - for work area passed to E15, E35 exits via USER ADDRESS CONSTANT
STG
12,MOTOKEN
SAVE USERTOKEN
LA
2,2
SET NUMBER OF SEGMENTS
STG
2,MOSEGM
FOR MEMORY OBJECT (MO) - 2MB
IARV64 REQUEST=GETSTOR, GET MEMORY OBJECT
+
ORIGIN=MOADDR,
ADDRESS OF MEMORY OBJECT
+
USERTKN=MOTOKEN,
USERTOKEN
+
SEGMENTS=MOSEGM,
SIZE OF MEMORY OBJECT
+
COND=YES,
CONDITIONAL REQUEST
+
MF=(E,MOWORK)
LTR
15,15
IF MEMORY OBJECT NOT OBTAINED,
BNZ
NOMO
EXIT WITH ERROR RETURN CODE
LG
10,MOADDR
GET 64-BIT ADDR IN MO FOR
*
64-BIT INVOCATION PARMLIST
USING ICE64INV,10
MAKE ADDRESSABLE
XC
0(ICE64LNG,10),0(10) CLEAR 64-BIT INVOCATION PARM LIST
MVC
ICEPLID(8),PL64SORT MOVE 64-BIT PARM LIST IDENTIFIER
LA
2,1024(,10)
GET 64-BIT ADDR IN MEMORY OBJECT
*
FOR 64-BIT CONTROL STATEMENTS AREA
STG
2,ICECTL
AND STORE IN 64-BIT PARM LIST
MVC
0(CTLNG,2),CTLST
MOVE CONTROL STATEMENTS INTO MO
*
LA
2,1024(,2)
GET 64-BIT ADDR IN MO FOR WORK AREA
STG
2,MOWA
USED IN E15, E35 EXITS
MVC
ICEUC(8),UADCON
MOVE USER EXIT ADDRESS CONSTANT
*
LLGF 2,=A(E15)
STORE E15 ADDR
840
Sort Examples
STG
OI
OI
2,ICE15E32
ICEMDEX1,ICE15A64
ICEMDEX2,ICE15PLT
INTO PARM LIST

WITH AMODE 64 AND
WITH 64-BIT EXEC PARAMETER LIST
LLGF 2,=A(E35)
STG
2,ICE35
OI
ICEMDEX1,ICE35A64
OI
ICEMDEX2,ICE35PLT
STORE E35 ADDR

INTO PARM LIST
WITH AMODE 64 AND
WITH 64-BIT EXEC PARAMETER LIST
INTO 64-BIT INVOCATION PARM LIST
LOAD ADDR OF 64-BIT INVOCATION
PARAMETER LIST
INVOKE SORT WITH 64-BIT NAME
SAVE RETURN CODE
*
LGR
1,10
*
FREEMO
*
EXITALL
*
NOMO
*
SAVE
HIGHREGS
HIGHZERO
*
CTLST
LINK EP=SORT64
LR
11,15
DS
0H
IARV64 REQUEST=DETACH,
MATCH=USERTOKEN,
USERTKN=MOTOKEN,
COND=YES,
MF=(E,MOWORK)
LR
15,11
RESTORE RETURN CODE
DS
LMH
L
L
LM
BSM
0H
0,14,HIGHREGS
13,SAVE+4
14,12(13)
1,12,24(13)
0,14
RESTORE HIGH PART REGS

RESTORE OLD SA ADDR
RESTORE CALLERS
REGS
RETURN TO CALLER
LA
B
15,20
EXITALL
ERROR RETURN CODE (NOT MO)

RETURN TO CALLER
DC
DC
DC
18F0
16F0
16F0
FREE MEMORY OBJECT
+
+
+
+
SAVE AREA
DS
0H
ADDR OF CONTROL STMTS
DC
AL2(CTL2-CTL1)
LENGTH OF CONTROL STATEMENT STRING
CTL1
EQU
*
DC
C SORT FIELDS=(5,4,BI,A)
DC
C RECORD LENGTH=(80,80,80),TYPE=V
DC
C OPTION FILSZ=E1000
DC
C OUTFIL FNAMES=SORTOUT
CTL2
EQU
*
CTLNG
EQU
*-CTLST
LENGTH OF CONTROL STATEMENTS AREA
PL64SORT DC
CPL64SORT
IDENTIFIER OF 64-BIT PARM LIST
UADCON DC
F0,A(MOWA)
USER ADDRESS CONSTANT
MOWA
DC
D0
ADDR OF WORK AREA IN MO
*
MOTOKEN DC
D0
USERTOKEN
MOSEGM DC
D0
SIZE OF MO (IN MB)
MOADDR DC
D0
ADDR OF MO
*
MOWORK DS
0D
IARV64 MACRO WORK AREA
IARV64 MF=(L,MOV64L)
*
LTORG
*
ICEPL64
MAPPING OF 64-BIT PARM LISTS
*
DROP 12
841
Sort Examples
PGM1
CSECT
************** E15 **********************************************
*
E15
DS
0H
* THIS E15 EXIT ROUTINE FORMS ALL RECORDS IN MEMORY OBJECT
* AND INSERTS THEM TO DFSORT AS 64-BIT ADRESSED RECORDS.
*
STMG 14,12,8(13)
SAVE CALLERs REGISTERS
BASR 11,0
SET
USING *,11
ADDRESSABILITY
LLGTR 11,11
SET CLEAN BASE REG
LARL 14,SAVEE15
GET NEW SAVE AREA ADDRESS
STG
13,128(,14)
CHAIN TO PREVIOUS SAVE AREA
STG
14,136(,13)
CHAIN TO NEW SAVE AREA
LGR
13,14
SET R13 TO SAVE AREA ADDRESS
LMH
0,15,E15HZERO
SET HIGH PART OF REGS TO ZERO
LR
10,1
SAVE EXIT PARAMETER LIST ADDRESS
USING ICE64E15,10
SET ADDRESSABILITY
ICM
15,15,COUNT
EXIT IF ALL RECORDS WERE
BZ
EOF
INSERTED
LLGF 1,ICE15UC+4
GET USER CONSTANT ADDR
LG
1,0(1)
GET ADDR OF THE WORK AREA IN MO
MVC
0(80,1),RECE15
MOVE RECORD INTO MEMORY OBJECT
BCTR 15,0
SET NEW
ST
15,COUNT
SORT FIELD
*
LA
15,12
RC=12 (INSERT RECORD)
* RETURN TO DFSORT
BACKE15 DS
0H
LG
13,128(,13)
RESTORE CALLERs R13
LG
14,8(,13)
RESTORE CALLERs R14
LMG
2,12,40(13)
RESTORE CALLERs R2-R12
BSM
0,14
RETURN
*
EOF
DS
0H
* Create final record in MO for its insert to DFSORT as 64-bit
* addressed record in the E35 exit
LLGF 15,ICE15UC+4
LG
15,0(15)
GET ADDR IN MO FOR FINAL RECORD
XC
0(80,15),0(15)
CLEAR RECORDs DATA
MVI
1(15),8
SET NEW LENGTH OF FINAL RECORD
MVC
4(4,15),RECNUM
SET NUMBER OF INSERTED RECORDS
LA
15,8
SET RETURN CODE
B
BACKE15
RETURN CODE
*
E15HZERO DC
18F0
SAVEE15 DC
0D0,F0,CF4SA,17FD0
HIGHE15 DC
16F0
FLAGE15 DC
X00
E15MOADR DC
D0
ADDR OF MO
E15SEGM DC
D0
SIZE OF MO (IN MB)
E15TOKEN DC
D0
USERTOKEN
RECNUM
DC
F1000
NUMBER OF RECORDS
RECE15
DC
H80
LENGTH OF RECORD
DC
H0
COUNT
DC
F1000
SORT FIELD
DC
6CABCDEF
DATA
DC
6C123456
FIELDS
*
LTORG
842
Sort Examples
*
DROP
PGM1
CSECT
************** E35 ********************************************
E35
DS
0H
*
* THIS E35 EXIT ACCEPTS ALL RECORDS FROM DFSORT
* AND AT EOF INSERTS ONE RECORD TO DFSORT AS 64-BIT RECORD.
* THE ADDRESS OF THE FINAL RECORD IN MEMORY OBJECT IS PASSED
* VIA USER ADDRESS CONSTANT FIELD OF 64-BIT PARAMETER LIST.
*
STMG 14,12,8(13)
SAVE CALLERs REGISTERS
BASR 11,0
SET
USING *,11
ADDRESSABILITY
LLGTR 11,11
SET CLEAN BASE REG
LARL 14,SAVEE35
GET NEW SAVE AREA ADDRESS
STG
13,128(,14)
CHAIN TO PREVIOUS SAVE AREA
STG
14,136(,13)
CHAIN TO NEW SAVE AREA
LGR
13,14
SET R13 TO SAVE AREA ADDRESS
LMH
0,15,E35HZERO
SET HIGH PART REGS TO ZERO
LR
10,1
SAVE EXIT PARM LIST ADDRESS
USING ICE64E35,10
SET ADDRESSABILITY
ICM
1,15,ICE35RL+4
GET ADDR OF NEW RECORD
* IF ADDR OF NEW RECORD IT IS ZERO, EOF IS INDICATED
BZ
RC8E35
BR IF EOF
*
LA
15,0
ACCEPT RETURN CODE
* RETURN TO DFSORT
BACKE35 DS
0H
LG
13,128(,13)
RESTORE CALLERs R13
LG
14,8(,13)
RESTORE CALLERs R14
LMG
2,12,40(13)
RESTORE CALLERs R2-R12
BSM
0,14
RETURN
*
RC8E35 DS
0H
TM
FLAGE35,X01
IF FINAL FLAG IS ON
BO
RC8E35A
EXIT WITH 8 RETURN CODE
MVI
FLAGE35,X01
SET FINAL FLAG
LLGF 1,ICE35UC+4
LG
1,0(1)
GET MO ADDR WITH FINAL RECORD
LA
15,12
SET INSERT RETURN CODE
B
BACKE35
RETURN TO DFSORT
RC8E35A DS
0H
LA
15,8
SET RETURN CODE
B
BACKE35
RETURN TO DFSORT
*
E35HZERO DC
18F0
HIGHE35 DC
18F0
SAVEE35 DC
0D0,F0,CF4SA,17FD0
FLAGE35 DC
X00
*
LTORG
*
END
Line
Explanation
01
02
EXEC statement. Calls a program named PGM1 that in turn calls DFSORT.
843
Sort Examples
03

to SYSOUT class A.
04-05
SORTOUT DD statement. The OUTFIL data set.
06
Start of the complete source code for PGM1. PGM1 uses the ICEPL64
mapping macro in the DFSORT target library, SICEUSER, to provide
separate Assembler DSECTs for the 64-bit invocation and exit parameter
lists.
PGM1 LINKs to DFSORT using the name SORT64 to indicate that it is
using the 64-bit invocation parameter list. PGM1 passes SORT, RECORD,
OPTION and OUTFIL statements in the control statement area.
PGM1's E15 user exit passes 64-bit addressed records to DFSORT.
PGM1's E35 user exit accepts all of the sorted records and inserts one
additional record at EOF. DFSORT writes all of the records to the OUTFIL
data set.
MERGE examples
Although the MERGE operators in the examples in this section could all be
for clarity.
Example 1. Merge with EQUALS

INPUT
OUTPUT
WORK DATA SETS
Not applicable
USER EXITS
None
FUNCTIONS/OPTIONS
EQUALS
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC PGM=SORT
//SYSOUT
DD
SYSOUT=A
//SORTIN01 DD
DSN=M1234.INPUT1,DISP=SHR
//SORTIN02 DD
//SORTIN03 DD
//SORTOUT DD
DSN=M1234.MERGOUT,DISP=(,KEEP),
// SPACE=(CYL,(2,4)),UNIT=3390
//SYSIN
DD *
MERGE FIELDS=(1,8,CH,A,20,4,PD,A)
OPTION EQUALS
844
01
02
03
04
05
06
07
08
09
10
11
Line
Explanation
01
02
03

to sysout class A.
Merge Examples
04
SORTIN01 DD statement. The first input data set is named M1234.INPUT1

and is cataloged. DFSORT determines the RECFM, LRECL and BLKSIZE
from the data set label.
05
SORTIN02 DD statement. The second input data set is named

M1234.INPUT2 and is cataloged. DFSORT determines the RECFM, LRECL
06
SORTIN03 DD statement. The third input data set is named

M1234.INPUT3 and is cataloged. DFSORT determines the RECFM, LRECL
07-08
SORTOUT DD statement. The output data set is named M1234.MERGOUT

and is to be allocated on 3390 and kept. DFSORT sets the RECFM and
LRECL from the SORTINnn data sets and selects an appropriate BLKSIZE.
09
10
MERGE statement. FIELDS specifies an ascending 8-byte character control

field starting at position 1 and an ascending 4-byte packed-decimal field
starting at position 20. The records in each input data set must already be
in the order specified.
25
OPTION statement. EQUALS specifies that the order of output records

with equal control fields is to be based on the file number of the input data
sets and the original order of the records within each input data set.
Example 2. Merge with LOCALE and OUTFIL

INPUT
OUTPUT
Multiple fixed-length record data sets
WORK DATA SETS
Not applicable
USER EXITS
None
FUNCTIONS/OPTIONS
LOCALE, OUTFIL
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC PGM=SORT
//SYSOUT
DD SYSOUT=A
//SORTIN01 DD DSN=INPUT01.GERMAN.GERMANY,DISP=SHR
//GP1
DD DSN=OUTPUT.GERMAN.GP1,DISP=OLD
//GP2
DD DSN=OUTPUT.GERMAN.GP2,DISP=OLD
//GP3
DD DSN=OUTPUT.GERMAN.SAVE,DISP=OLD
//DFSPARM DD *
LOCALE=De_DE.IBM-1047
MERGE FIELDS=(25,5,CH,A,40,4,PD,D)
OUTFIL FNAMES=GP1,
INCLUDE=(23,1,CH,LE,C)
OUTFIL FNAMES=GP2,
INCLUDE=(23,1,CH,GT,C,AND,
23,1,CH,LT,C)
OUTFIL FNAMES=GP3,SAVE
Line
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
Explanation
845
Merge Examples
01
02
03
STEPLIB DD statement. Specifies the Language Environment run-time

library containing the dynamically loadable locales.
04
SYSOUT statement. Directs DFSORT messages and control statements to

sysout class A.
05
SORTIN01 DD statement. The first input data set is named

INPUT01.GERMAN.GERMANY and is cataloged. DFSORT determines the
06
SORTIN02 DD statement. The second input data set is named

07
SORTIN03 DD statement. The third input data set is named

08
GP1 DD statement. The first OUTFIL output data set is named

OUTPUT.GERMAN.GP1 and is cataloged. DFSORT determines the
09
GP2 DD statement. The second OUTFIL output data set is named

10
GP3 DD statement. The third OUTFIL output data set is named

11
DFSPARM DD statement. DFSORT control statements follow.
12
LOCALE parameter. Overrides the installation default for LOCALE. The

locale for the German language and the cultural conventions of Germany
based on the IBM-1047 encoded character set will be active.
13
MERGE statement. FIELDS specifies an ascending 5-byte character control

field starting at position 25, and a descending 4-byte packed decimal
control field starting at position 40. The character (CH) control field will be
merged according to the collating rules defined in locale De_DE.IBM-1047.
The records in each input data set must already be in the order specified.
14-15
OUTFIL statement. The subset of records with character values less than or
equal to '' in position 23 are written to the GP1 output data set. The
character (CH) compare field and character constant will be compared
according to the collating rules defined in locale De_DE.IBM-1047.
16-18
OUTFIL statement. The subset of records with character values greater

than '' but less than '' in position 23 are written to the GP2 output data
set. The character (CH) compare fields and character constants will be
compared according to the collating rules defined in locale
De_DE.IBM-1047.
19
OUTFIL statement. Any records not written to the GP1 or GP2 output data
sets are written to the GP3 output data set.
Copy examples
This section contains 2 copy examples.
846
Copy Examples
Example 1. Copy with EXEC PARMs, SKIPREC, MSGPRT and

ABEND
INPUT
Blocked fixed-length records on multivolume 3490
OUTPUT
WORK DATA SETS
Not applicable
USER EXITS
None
FUNCTIONS/OPTIONS
SKIPREC, MSGPRT, ABEND
//EXAMP
JOB A400,PROGRAMMER
//STEP1
EXEC
PGM=SORT,
// PARM=SKIPREC=500,MSGPRT=CRITICAL,ABEND
//SYSOUT
DD SYSOUT=A
//SORTIN
DD DSN=FLY.RECORDS,VOL=SER=(000333,000343),
// UNIT=(3490,2),DISP=OLD,LABEL=(,NL),
// DCB=(RECFM=FB,LRECL=12000,BLKSIZE=24000)
//SORTOUT DD DSN=FLY.RECORDS.COPY,DISP=OLD
//SYSIN DD *
SORT FIELDS=COPY
01
02
03
04
05
06
07
08
09
10
Line
Explanation
01
02-03
EXEC statement. Calls DFSORT directly by its alias SORT. SKIPREC=500

specifies that the first 500 input records are not to be included in the
output data set. MSGPRT=CRITICAL specifies that error messages, but not
informational messages, are to be printed. ABEND specifies that DFSORT
is to terminate with a user ABEND if it issues an error message.
04

to sysout class A.
05-07
SORTIN DD statement. The input data set is named FLY.RECORDS and

resides on 3490 volumes 000333 and 000343. The UNIT parameter requests
two tape drives, one for each volume of the data set. Because the tape is
unlabeled, DCB parameters must be supplied to indicate that the RECFM
is FB, the LRECL is 12000 and the BLKSIZE is 24000.
08

FLY.RECORDS.COPY and is cataloged. DFSORT determines the RECFM,
LRECL and BLKSIZE from the data set label.
09
10
SORT statement. FIELDS=COPY specifies a copy application.
Example 2. Copy with INCLUDE and VLSHRT

INPUT
Blocked spanned records on disk
OUTPUT
Blocked spanned records on SYSDA
847
Copy Examples
WORK DATA SETS
Not applicable
USER EXITS
None
FUNCTIONS/OPTIONS
INCLUDE, VLSHRT
//EXAMP
JOB A400,PROGRAMMER
//COPY
EXEC PGM=SORT
//SYSOUT
DD SYSOUT=A
//SORTIN
DD DSN=SMF.DATA,DISP=SHR
//SORTOUT DD DSN=SMF.VIOL,DISP=(,KEEP),SPACE=(CYL,(2,5)),
// UNIT=SYSDA
//SYSIN
DD *
INCLUDE COND=(6,1,FI,EQ,80,AND,19,1,BI,EQ,B1.......)
OPTION COPY,VLSHRT
01
02
03
04
05
06
07
08
09
Line
Explanation
01
02
03

to sysout class A.
04
SORTIN DD statement. The input data set is named SMF.DATA and is

VBS, the LRECL is 32760 and the BLKSIZE is 23476.
05-06
SORTOUT DD statement. The output data set is named SMF.VIOL and is

to be allocated on SYSDA and kept. DFSORT sets the RECFM and LRECL
from SORTIN and selects an appropriate BLKSIZE.
07
08
INCLUDE statement. COND specifies that only input records with decimal
80 in the 1-byte fixed-point field at position 6 and bit 0 on in the 1-byte
binary field at position 19 are to be included in the output data set.
09
OPTION statement. COPY specifies a copy application. VLSHRT specifies

that records that are too short to contain all of the INCLUDE compare
fields are not to be included in the output data set.
Example 3. Copy with OUTREC, PARSE and BUILD

INPUT
OUTPUT
WORK DATA SETS
USER EXITS
FUNCTIONS/OPTIONS

Not applicable
None
OUTREC, PARSE, BUILD

//S1 EXEC PGM=ICEMAN
//SORTIN DD DSN=FLY.VAR.FIELDS.IN,DISP=SHR
//SORTOUT DD DSN=FLY.FIX.FIELDS.OUT,DISP=OLD
//SYSIN DD *
OPTION COPY
OUTREC PARSE=(%00=(ABSPOS=21,STARTAFT=BLANKS,
ENDBEFR=C,,FIXLEN=5),
%=(ENDBEFR=C,),
%02=(FIXLEN=7)),
BUILD=(1,15,
848
01
02
03
04
05
06
07
08
09
10
11
12
13
Copy Examples
18:%00,JFY=(SHIFT=RIGHT,LEAD=C(,TRAIL=C),LENGTH=7),
30:%02,SFF,TO=ZD,LENGTH=7,
40:%01,60:X)
14
15
16
Line
Explanation
01
02
03

to sysout class A.
04
SORTIN DD statement. The input data set is named FLY.VAR.FIELDS.IN

and is cataloged. DFSORT determines from the data set label that the
RECFM is FB, the LRECL is 60 and the BLKSIZE is 27960.
05

FLY.FIX.FIELDS.OUT and is cataloged. DFSORT determines from the data
set label that the RECFM is FB, the LRECL is 60 and the BLKSIZE is 27960.
06
07
OPTION statement. COPY specifies a copy application.
08-16
OUTREC statement. PARSE and BUILD reformat the input records

containing one fixed position/length field and four variable
position/length fields to output records containing four fixed
position/length fields.
ICEGENER example
This section contains an ICEGENER example.
INPUT
Same as for IEBGENER job
OUTPUT
Same as for IEBGENER job
WORK DATA SETS
Not applicable
USER EXITS
None
FUNCTIONS/OPTIONS
None
//EXAMP
//GENR
//SYSPRINT
//SYSUT1
//SYSUT2
//SYSIN
JOB A400,PROGRAMMER
EXEC PGM=ICEGENER
DD SYSOUT=A
DD DSN=CTL.MASTER,DISP=SHR
DD DSN=CTL.BACKUP,DISP=OLD
DD DUMMY
01
02
03
04
05
06
This example shows how to use the ICEGENER facility for an IEBGENER job if
your site has not installed ICEGENER as an automatic replacement for IEBGENER.
The ICEGENER facility selects the more efficient DFSORT copy function for this
IEBGENER job.
Line
Explanation
01
849
ICEGENER Example
02
EXEC statement. Calls the ICEGENER facility. PGM=IEBGENER has been

replaced by PGM=ICEGENER.
03-06
No other changes to the IEBGENER job are required.
ICETOOL example
This section contains an example of ICETOOL with various operators.
INPUT
Multiple data sets
OUTPUT
Multiple data sets
WORK DATA SETS
Dynamically allocated (automatic)
USER EXITS
ICETOOL's E35 (automatic)
FUNCTIONS/OPTIONS
OCCUR, COPY, SORT, MODE, VERIFY, STATS, DISPLAY
//EXAMP
JOB A400,PROGRAMMER
//TOOLRUN EXEC PGM=ICETOOL,REGION=1024K
//DFSMSG
DD SYSOUT=A
//TOOLIN
DD *
* Print report showing departments with less than 5 employees
OCCUR FROM(IN1) LIST(LT5) LOWER(5) BLANK TITLE(Small Departments) PAGE HEADER(Department) HEADER(Employees) ON(45,3,CH)
ON(VALCNT)
* Sort by last name and first name
SORT FROM(IN1) TO(DEPTSD,DEPTSP) USING(ABCD)
* Do following operators even if a previous operator failed,
* but stop processing if a subsequent operator fails.
MODE STOP
* Verify decimal fields
VERIFY FROM(IN2) ON(22,6,PD) ON(30,3,ZD)
* Print statistics for record length and numeric fields
STATS FROM(IN2) ON(VLEN) ON(22,6,PD) ON(30,3,ZD)
* Sort and produce total for each unique key
SORT FROM(IN2) TO(OUT4) USING(CTL1)
* Print report containing:
*
- key and total for each unique key
*
- lowest and highest of the totals
DISPLAY FROM(OUT4) LIST(LIST1) TITLE(Unique key totals report) DATE TIME ON(5,10,CH) ON(22,6,PD) ON(30,3,ZD) MINIMUM(Lowest) MAXIMUM(Highest) PLUS
//LT5
DD SYSOUT=A
//IN1
DD DSN=FLY.INPUT1,DISP=SHR
//ABCDCNTL DD *
* Sort by last name, first name
SORT FIELDS=(12,15,CH,A,1,10,CH,A)
//DEPTSD
DD DSN=FLY.OUTPUT1,DISP=SHR
//DEPTSP
DD SYSOUT=A
//IN2
DD DSN=FLY.INPUT2,DISP=SHR
//OUT4
DD DSN=FLY.OUTPUT2,DISP=OLD
//CTL1CNTL DD *
* Sort and produce totals in one record for each unique key
SUM FIELDS=(22,6,PD,30,3,ZD)
//LIST1
DD SYSOUT=A
850
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
ICETOOL Example
This example shows how ICETOOL can be used to perform multiple operations in
a single step.
Line
Explanation
01
02
EXEC statement. Calls ICETOOL specifying the recommended REGION of

1024K.
03
TOOLMSG DD statement. Directs ICETOOL messages and statements to

system output class A.
04
DFSMSG DD statement. Directs DFSORT messages and control statements

to SYSOUT class A.
05
TOOLIN DD statement. ICETOOL statements follow. The MODE for the

ICETOOL run is initially set to STOP. If an error is detected for an
operator, SCAN mode will be entered.
06
07-10
OCCUR operator. Prints, in the LT5 data set, a report detailing each value
for the specified field in the IN1 data set and the number of times that
value occurs.
11
Comment statement.
12
SORT operator. Records from the IN1 data set are sorted to the DEPTSD
and DEPTSP data sets using the DFSORT control statements in the
ABCDCNTL data set. As a result, FLY.OUTPUT1 and DEPTSP (SYSOUT)
contain the sorted records from FLY.INPUT1.
13-14
Comment statements.
15
MODE operator. The MODE is reset to STOP (needed in case SCAN mode
was entered due to an error for a previous operator). If an error is detected
for a subsequent operator, SCAN mode will be entered. This divides the
previous operators and subsequent operators into two unrelated groups.
16
Comment statement.
17
VERIFY operator. Identifies invalid values, if any, in the specified decimal

fields of the IN2 data set. Used to stop subsequent operations if any
invalid value is found in FLY.INPUT2.
18
Comment statement.
19
STATS operator. Prints the minimum, maximum, average, and total for the
specified fields of the IN2 data set.
ON(VLEN) operates on the record length of the records in FLY.INPUT2.
Thus, the values printed for ON(VLEN) represent the shortest record, the
longest record, the average record length, and the total number of bytes for
FLY.INPUT2.
20
Comment statement.
21
SORT operator. Records from the IN2 data set are sorted and summarized
to the OUT4 data set using the DFSORT control statements in the
CTL1CNTL data set. As a result, FLY.OUTPUT2 contains one record from
FLY.INPUT2 for each unique sort field with totals for the sum fields.
22-24
Comment statements.
25-28
DISPLAY operator. Prints, in the LIST1 data set, a report detailing each sort
851
ICETOOL Example
and sum value for the OUT4 data set resulting from the previous
operation, and the lowest and highest value for each sum field.
29-42
852
DD statements. Defines the data sets and DFSORT control statements used
for the ICETOOL operations described previously in this section.
Appendix A. Using work space

Introduction
When a sort application cannot be performed entirely in virtual storage, DFSORT
must use work space. The amount of work space required depends on:
v The amount of data being sorted
v The amount of virtual storage available to DFSORT
v The amount of Hiperspace available to DFSORT
v The type of devices you use
v The DFSORT functions and features you use (for example, VLSHRT, locale
processing, EFS, and ALTSEQ can increase the amount of work space required).
There are three ways to supply work space for a DFSORT application:
v Hiperspace
v Dynamic allocation of work data sets
v JCL allocation of work data sets.
For best performance, an optimal amount of Hiperspace, in combination with
dynamically allocated disk work data sets, is strongly recommended. See Use
Hipersorting on page 811 for more information on using the HIPRMAX option.
The DYNAUTO installation option, or the DYNALLOC run-time option, can be
used to dynamically allocate work data sets.
Central storage
DFSORT leverages central storage as intermediate work space through the use of
either memory objects or Hiperspaces. Using the installation defaults of
HIPRMAX=OPTIMAL, MOSIZE=MAX and MOWRK=YES ensures that DFSORT
will use memory object or Hiperspace whenever possible. Sites can tune their
definition of HIPRMAX=OPTIMAL and MOSIZE=MAX through use of the
EXPMAX, EXPOLD, EXPRES and TUNE installation options. See z/OS DFSORT
Installation and Customization for more information.
DFSORTs use of memory object or Hiperspace storage depends upon the
availability of central storage, the needs of other concurrent memory object sorting,
Hipersorting, and dataspace sorting applications throughout the time the
application runs, and the settings of the EXPMAX, EXPOLD, EXPRES and TUNE
installation options. Consequently, it is possible for the same application to use
varying amounts of memory object or Hiperspace storage from run to run. If
enough memory object or Hiperspace storage is available, DFSORT uses it
exclusively for intermediate storage. If the amount of available storage is
insufficient, DFSORT uses a combination of work data sets with Hiperspace or
memory object storage, or even work data sets alone.
DFSORT's use of memory object and Hiperspace storage is very dynamic; multiple
concurrent sort applications always know each other's storage needs and never try
to back their memory objects or Hiperspaces with the same portion of storage. In
addition, DFSORT checks the available storage prior to creating each memory
object or Hiperspace. Depending on the TUNE installation default each SORT
application can be more or less dynamic:
853
Using Work Space

v by checking available storage once and then allocating all available storage
expected to be used at initialization, or
v by checking available storage multiple times and allocating it in increments
throughout the run.
DFSORT also has the capability of switching from using memory objects or
Hiperspaces to using disk work data sets when either of the following occur:
v allocation of an additional memory object or Hiperspace is likely to cause a
storage shortage (for more information on how DFSORT uses central storage, see
Recommendations under Hipersorting, Memory Object Sorting, and Data
Space Sorting in z/OS DFSORT Tuning Guide).
v the total Hipersorting, memory object sorting, and dataspace sorting activity on
the system reaches the limits set by the EXPMAX, EXPOLD, and EXPRES
installation options.
Memory object and Hipersorting requires that work data sets be available, as well
as central storage. DFSORT forces the use of dynamic allocation for memory object
sorting or Hipersorting if work data sets are not requested explicitly. For further
details, see the MOSIZE , MOWRK and HIPRMAX options of the OPTION
Work data set devices

The type of device selected for work data sets can have a significant effect on
performance. Consider the following when selecting devices for work data sets.
Disk and tape devices

For optimal performance, use disk devices to which little other activity is directed,
for work data sets. Specify high-speed devices such as IBM's TotalStorage DS8000
storage subsystems, and avoid specifying tape or virtual (VIO).
Using tape devices for work data sets rather than disk causes significant
performance degradation for the following reasons:
v Tape work data sets prevent DFSORT from using its more efficient sorting
techniques, Blockset and Peerage/Vale. Disk work data sets allow DFSORT to
use these techniques.
v Tape work data sets must be accessed sequentially. Disk data sets can be
accessed randomly.
v Disk control units can provide additional features, such as cache fast write, that
are not available with tape devices.
Number of devices
Although one work data set is sufficient, using two or more work data sets on
separate devices usually reduces the elapsed time of the application significantly.
In general, using more than three work data sets does not reduce elapsed time any
further, and is only necessary if the work data sets are small or the file size is
large.
For optimum allocation of resources such as virtual storage, avoid specifying a
large number of work data sets unnecessarily.
854
Using Work Space

No more than 255 work data sets can be specified. If you specify more than 32
work data sets, and the Blockset technique is not selected, a maximum of 32 work
data sets is used.
Non-synchronous storage subsystems

Allocation of work data sets on devices attached to non-synchronous storage
subsystems can affect the performance of certain DFSORT applications. Whether or
not a particular application is affected depends on many factors, the most critical
of which is the ratio of input file size to available storage.
In general, to maximize performance, DFSORT needs the following:
v Accurate knowledge of the size of the file being sorted
In most cases, DFSORT is able to calculate the file size accurately. However, for
applications that sort many input tapes that are not managed by DFSMSrmm or
a tape management system that uses ICETPEX (especially compacted input
tapes) or that use E15 exits that add or delete records, we recommend that you
specify the file size using the FILSZ or SIZE parameter.
v Adequate storage relative to the size of the file being sorted
Table 92 shows the minimum recommended storage to provide DFSORT based
on various input file sizes.
Table 92. Minimum Storage Required for Various File Sizes
Minimum storage
Less than 200MB
4MB
200MB to 500MB
8MB
500MB to 1GB
16MB
More than 1GB
16-32MB
Allocation of work data sets

Dynamic allocation has the following advantages over JCL allocation:
v Installation options can be used to automatically activate dynamic allocation for
all sort applications.
To use JCL allocation, appropriate DD statements must be specified for each
individual application.
v As the characteristics (file size, virtual storage, and so on) of an application
change over time, DFSORT can automatically optimize the amount of
dynamically allocated work space for the application. This eliminates unneeded
allocation of disk space.
JCL allocation is fixed; DFSORT cannot adjust it. Disk space might be wasted.
v Dynamically allocated work data sets are automatically allocated as large format
so they can use more than 65535 tracks on a single volume.
v When disk work space requirements are larger than expected, DFSORT can
automatically recover by increasing allocation sizes or using additional work
data sets.
v As the amount of Hiperspace available to the application varies from run to run,
DFSORT can automatically adjust the amount of space it dynamically allocates
to complement the amount of Hiperspace. This eliminates unneeded allocation
of disk space.
855
Allocation of Work Data Sets

JCL allocation is fixed; DFSORT cannot adjust it, even if all sorting can be done
in Hiperspace. Disk space might be wasted.
Dynamic allocation has one drawback: for certain applications, as described in
File size and dynamic allocation on page 857, you might need to give DFSORT a
reasonable estimate of the input file size. Later, if the input file size for the
application increases significantly, you must update the file size estimate
accordingly.
However, JCL allocation has a similar drawback, except that it applies to all
applications. Unless you overallocate the work data sets initially and waste space,
you have to update the JCL allocation when the input file size increases
significantly for any application to avoid out-of-space abends.
If you can allocate enough work data set space with JCL to guarantee your
applications will never exceed the space allocated, you do not need dynamic
allocation. However, since efficient use of disk space is usually desirable, dynamic
allocation is recommended over JCL allocation.
For both dynamic allocation and JCL allocation:
v The amount of work space actually used will often be less than the amount
allocated. DFSORT tries to minimize dynamic over-allocation while making
certain that the application will not fail due to lack of space. With JCL allocation,
you could minimize the amount of allocated space manually, but this might
require changes to JCL allocation as the characteristics of the application change
over time.
v Limiting the virtual storage available to DFSORT can increase the amount of
work space required. With a reasonable amount of storage, 4MB for example,
DFSORT can sort using a reasonable amount of work space. If storage is limited,
more work space might be required. If storage is drastically limited (for
example, to 200KB), significantly more work space might be required.
Dynamic allocation of work data sets

Installation options are available to request automatic dynamic allocation of work
data sets and to supply defaults for the device type and number of devices.
For certain applications, it is very important to specify a reasonable estimate of the
input file size when using dynamic allocation.
Automatic dynamic allocation

Your system programmer has set the DYNAUTO installation option to control
whether dynamic allocation is used automatically, or only when requested by the
DYNALLOC run-time option.
DYNAUTO also controls whether dynamic allocation or JCL allocation takes
precedence when JCL work data sets are specified.
If your system programmer selected DYNAUTO=IGNWKDD, dynamic allocation
takes precedence over JCL allocation (JCL work data sets are actually deallocated).
If you want the opposite precedence for selected applications, use the run-time
option USEWKDD.
If your system programmer selected DYNAUTO=YES, JCL allocation takes
precedence over dynamic allocation. If you want the opposite precedence, you
must remove the JCL allocation statements.
856

If your system programmer selected DYNAUTO=NO, dynamic allocation of work
data sets is not used unless you specify the DYNALLOC run-time option. JCL
allocation takes precedence over dynamic allocation.
Device defaults
When the device type, or the number of devices for dynamic allocation, is not
explicitly specified, DFSORT obtains the missing information from the DYNALOC
installation option information supplied by your system programmer.
Note: In rare cases where DFSORT is unable to dynamically allocate the required
space using the value supplied by d, allocation may be attempted with a more
generic unit name (such as 3390) in an effort to allocate the required space on
volumes from a different storage group. Refer to "System Managed Storage" in
z/OS DFSORT Tuning Guide for recommendations on controlling data set allocation.
File size and dynamic allocation

DFSORT bases the amount of work space it dynamically allocates on the number
of bytes to be sortedthe input file size. Generally, DFSORT can automatically
make an accurate determination of the file size by determining the number of
input records. However, DFSORT cannot always determine the input file size
accurately in the following cases:
v An E15 user exit routine supplies all input records (an input data set is not
present). DFSORT cannot automatically determine the number of records to be
inserted.
v An input data set is present, along with an E15 user exit routine. DFSORT can
automatically determine the number of records in the input data set, but cannot
automatically determine the number of records to be inserted or deleted.
v A spool (DD *) or pipe data set is used as input.
v The input consists of small data sets on tape that are not managed by
DFSMSrmm or a tape management system that uses ICETPEX. When the tape
data sets are not managed, DFSORT cannot know how much of the tapes are
used, so it determines the file size assuming full volumes at the maximum
regular density for the drives.
v The Improved Data Recording Capability (IDRC) feature is used for the input
device and the tape data sets are not managed by DFSMSrmm or a tape
management system that uses ICETPEX.
v Input data sets are members of partitioned data sets. DFSORT cannot determine
the size of a member in a partitioned data set. Therefore, when input data sets
are partitioned, DFSORT uses the size of the entire data set as the input file size.
This is usually an over-estimation, which leads to over-allocation of work space.
In these circumstances, if the number of records is not supplied by the FILSZ or
SIZE option, you will receive message ICE118I. If dynamic allocation of work data
sets is used, DFSORT allocates the primary space according to the DYNSPC value
in effect. This can result in underallocation or overallocation, possibly leading to
wasted space or an out-of-space condition, respectively. DFSORT will also allocate
additional work data sets with zero space that can be used to recover from an out
of space condition. The percentage of additional work data sets will be the greater
of 50% or the DYNAPCT value in effect. To avoid unknown file size situations,
you should specify FILSZ=En with a reasonably accurate estimate of the number of
records to be sorted. If you cannot specify FILSZ=En, you should use DYNSPC=n
to adjust the primary space for dynamically allocated work data sets and
DYNAPCT=x to increase the number of additional work data sets, as appropriate.
Note: FILSZ=E0 is ignored.
857

For variable-length records, DFSORT uses one-half of the maximum record length
(LRECL) in conjunction with the number of records to determine the input file
size, unless you specify AVGRLEN=n. If your actual average record length is
significantly different from one-half of the maximum record length, specifying
AVGRLEN=n can prevent DFSORT from overallocating or underallocating dynamic
work space.
See OPTION control statement on page 173 for more information about the
AVGRLEN, DYNSPC, FILSZ, and SIZE options.
Dynamic over-allocation of work space

DFSORT can dynamically over-allocate the work space even when you specify the
number of records under the following circumstances:
v When you delete a significant number of records using:
An INCLUDE or OMIT statement, or the SKIPREC option. Use of these
statements and options does not force DFSORT to use a SIZE=En or FILSZ=En
specification. DFSORT ignores the En value unless it cannot compute the
input file size.
One or more partitioned data set members as input. DFSORT uses the size of
the entire partitioned data set rather than the size of the member in its
calculations. DFSORT ignores any SIZE=En or FILSZ=En value unless it
cannot determine the input file size itself.
You can avoid over-allocation in these cases by specifying SIZE=Un or
FILSZ=Un.
v When the average record length of variable-length records is substantially
shorter than one-half of the maximum record length. If DFSORT uses your exact
or estimated number of records, it uses one-half of the maximum record length
to determine the file size. You can avoid over-allocation in this case by
specifying AVGRLEN=n.
Dynamic over-allocation of work space can occur when you do not specify the
number of records (for example, with small input data sets on tape), or even when
you do (for example, when a significant number of records is deleted). In these
cases, you might prefer to use JCL allocation of work data sets to control the
amount of space allocated. However, there are drawbacks to doing so, as
previously explained. If DYNAUTO=IGNWKDD is used, remember to specify
run-time option USEWKDD when you want to use JCL allocation of work data
sets.
JCL allocation of work data sets

The amount of required work space is dependent on many factors such as virtual
storage and type of devices used, but is especially sensitive to the file size of the
input data set.
Because of the number of variables involved, an exact formula cannot be given for
calculating the needed work space. However, the following guidelines usually hold
true:
v For fixed length record (FLR) sort applications, 1.5 to 2 times the input file size
is usually adequate.
v For variable-length record (VLR) sort applications, 1.5 to 2.5 times the input file
size is usually adequate.
858

These guidelines assume that a reasonable amount of storage (at least 1M) is
available to DFSORT. Limiting the available amount of storage can increase the
amount of needed work space.
DFSORT can often run with less than the amount of work space indicated by the
guidelines described previously in this section.
To get the best performance using JCL allocation of work data sets:
v Use devices without much activity on them.
v Use high-speed disk devices such as IBM's TotalStorage DS8000 subsystems for
work data sets, and avoid using tape or virtual (VIO) for work data sets.
v Allocate space in cylinders.
v Specify contiguous space for each work data set, and make sure there is enough
primary space so that secondary space is not needed.
v Allocate two or more work data sets.
v Use multiple channel paths to the devices.
v Use different volumes and separate channel paths for the work data sets and the
input/output data sets.
The following table shows the work data set space needed with 4M of storage for
applications with various characteristics when Hipersorting, dataspace sorting, and
memory object sorting are not used (HIPRMAX=0, DSPSIZE=0 and MOSIZE=0).
Table 93. Work Space Requirements for Various Input Characteristics
Input Data set Characteristics
Filesize (MB) FLR/VLR
Max LRECL
Cylinders (3390)
BLKSIZE
FLR
80
27920
Input Data
Set
Work Data
Set
6
FLR
160
27840
20
FLR
80
27920
26
36
20
FLR
160
27840
26
36
20
FLR
1000
27000
26
36
40
FLR
80
27920
51
56
40
FLR
160
27840
51
56
40
FLR
1000
27000
52
56
150
FLR
160
27840
189
198
VLR
300
27998
40
VLR
300
27998
51
63
40
VLR
6000
27998
55
59
150
VLR
300
27998
188
200
150
VLR
6000
27998
203
200
Disk capacity considerations

You can specify a mixture of disk devices for a given sort application. Any IBM
disk device supported by your operating system can be used for work data sets.
859
Disk Capacity Considerations

For best performance, use high-speed devices such as IBM's TotalStorage DS8000
storage subsystems for work data sets.
System performance is improved if work data sets are specified in cylinders, rather
than tracks or blocks. Storage on temporary work data sets will be readjusted to
cylinders if possible. The number of tracks per cylinder for disk devices is shown
in Table 94.
Table 94. Number of Tracks per Cylinder for Disk Devices
Tracks per Cylinder
Maximum Bytes used per

Track
3380
15
47476
3390
15
56664
9345
15
46456
If WRKSEC is in effect and the work data set is not allocated to VIO, DFSORT
allocates secondary extents as required, even if not requested in the JCL.
Exceeding disk work space capacity

If during sorting, the allocation of secondary space on one of the work data sets
fails, the system issues a B37 informational message. DFSORT can recover by
allocating space on one of the other work data sets, if one is available.
DFSORT normally allocates secondary extents for work data sets, even if not
requested in the JCL. This reduces the probability of exceeding work space
capacity.
If the disk work space is not sufficient to perform the sort, DFSORT issues a
message and terminates.
Tape capacity considerations

Any IBM tape device supported by your operating system can be used for work
space. However, using tape devices for work data sets rather than disk causes
significant performance degradation and should, therefore, be avoided.
Three different tape work data set techniques are available to DFSORT: Balanced,
Polyphase, and Oscillating. For information on how to calculate their requirements,
see Table 95.
Note: The value you obtain for min is literally a minimum value; if, for
example, your input uses a more efficient blocking factor than DFSORT or is
spanned, you need more work space. Space requirements are also summarized in
Table 95. DFSORT selects the most appropriate tape technique using these criteria.
Table 95. Work Space Requirements of the Various Tape Techniques
Maximum Input
Work Space Areas

Required
Balanced tape
(BALN)
15 volumes
Min=2(V+1)* tape units 32 volumes
Used if more than three work

storage tapes are provided
and file size is not given.
Polyphase tape
(POLY)
1 volume
Min=3 tape units
Used if three work storage

tapes are provided.
860
Max. No. of
Work Areas
17 volumes
Comments

Table 95. Work Space Requirements of the Various Tape Techniques (continued)
Maximum Input
Oscillating tape
(OSCL)
15 volumes
Work Space Areas

Required
Max. No. of
Work Areas
Min=V+2* or 4 tape
units, whichever is
greater
17 volumes
Comments
File size must be given. The
tape drive containing SORTIN
cannot be used as a work
unit.
Note:
V = Number of input volumes. Number of input volumes of blocking equals work space blocking.
Exceeding tape work space capacity

At the beginning of a sort using tape work data sets, DFSORT estimates the
maximum sort capacity (Nmax) and issues message ICE038I. See the explanation of
this message for details.
The value for Nmax printed in message ICE038I is an average value rounded
down to the nearest thousand. This value assumes random input. If you have a
reversed sequenced file and tape work space, sort capacity may be exceeded at a
lower value because of the higher number of partly empty, end-of-string blocks.
For magnetic tape, a tape length of 2400 feet is assumed in calculating Nmax. For
tapes of other lengths, the figure is not correct. When tapes with mixed density are
used, the smallest density is used in the calculation.
If you specify an actual data set size, and that size is larger than the maximum
capacity estimated by the program (Nmax), the program terminates before
beginning to sort. If you specify an estimated data set size, or none at all, and the
number of records reaches the maximum (Nmax), the program gives control to
your routine at user exit E16, if you have written and included one. This routine
can direct the program to take one of the following actions:
v Continue sorting the entire input data set with available work space. If the
estimate of the input data set size was high, enough work space may remain to
complete the application.
v Continue sorting with only part of the input data set; the remainder could be
sorted later and the two results merged to complete the application.
v Terminate the program without any further processing.
If you do not include an E16 routine, DFSORT continues to process records for as
long as possible. If the work space is sufficient to contain all the records in the
input data set, DFSORT completes normally; when work space is not sufficient,
DFSORT issues a message and terminates.
The program generates a separate message for each of the three possible error
conditions. They are:
1. ICE041AN GT NMAX: Generated before sorting begins when the exact file
size is greater than Nmax.
2. ICE046ASORT CAPACITY EXCEEDED: Generated when the sort has used
all available work space.
3. ICE048INMAX EXCEEDED: Generated when the sort has exceeded Nmax
and has transferred control to a user-written E16 routine for further action.
861

The test for message ICE041A is made with the maximum possible calculated
value, that is, DFSORT is sure it will fail. In case of doubt, the message is not
issued.
862
Appendix B. Specification/override of DFSORT options

Installation defaults on page 17 discusses DFSORT's installation options and
environments, and shows you how to use ICETOOL's DEFAULTS operator to list
the installation defaults selected at your site.
Listed below are the places in DFSORT where you can specify various options that
will override the IBM-supplied defaults. The sources for the options are listed in
override order; that is, any option specified in a higher place in the list overrides
one specified in a lower place.
Directly Invoked DFSORT
v DFSPARM data set
PARM options
DEBUG and OPTION control statements
Other control statements.
v EXEC statement PARM options
v SYSIN data set
v Installation options set with PARMLIB ICEPRMxx members (JCL, TSO or TDx).
v Installation options set with ICEMAC (JCL, TSO or TDx).
Program Invoked DFSORT
v DFSPARM data set
PARM options
v SORTCNTL data set
v Parameter list
v Installation options set with PARMLIB ICEPRMxx members (INV, TSOINV or
TDx).
v Installation options set with ICEMAC (INV, TSO or TDx).
Note:
1. DFSORT's EXEC PARMs can be specified in the EXEC statement when DFSORT
is directly invoked. For example:
//S1 EXEC PGM=ICEMAN,PARM=EQUALS,ABEND
DFSORT's EXEC PARMs are ignored if DFSORT is invoked from a program.

2. DFSPARM PARMs can be specified in the DFSPARM data set when DFSORT is
directly invoked or invoked from a program. For example:
863
Specification/Override Of Options
//DFSPARM DD *
EQUAL,ABEND
3. DFSORT's control statements can be specified in the DFSPARM or SYSIN data

set when DFSORT is directly invoked. At least one blank must precede the
operation field (SORT, MERGE, INCLUDE, and so on). For example:
//SYSIN DD *
OPTION EQUALS
DEBUG ABEND
4. DFSORT's control statements can be specified in the DFSPARM or SORTCNTL

data set, or in the parameter list, when DFSORT is invoked from a program. At
least one blank must precede the operation field (SORT, MERGE, INCLUDE,
and so on). For example:
//SORTCNTL DD *
OPTION EQUALS
DEBUG ABEND
5. For the DEBUG and OPTION statements, override is at the option level. For
example, with:
//DFSPARM DD *
OPTION EQUALS
//SYSIN DD *
OPTION NOEQUALS,SKIPREC=50
EQUALS from DFSPARM overrides NOEQUALS from SYSIN, but SKIPREC=50

from SYSIN in not affected by the OPTION statement in DFSPARM, so both
EQUALS and SKIPREC=50 will be used.
For control statements other than DEBUG and OPTION, override is at the
statement level. For example, with:
//DFSPARM DD *
MODS E15=(CHECK,4096,EXIT)
//SYSIN DD *
MODS E35=(MOVE,2048,EXITX)
the MODS statement in DFSPARM completely overrides the MODS statement

in SYSIN, so the E15 exit will be used, but the E35 exit will not.
6. An EFS program or an installation initialization exit (ICEIEXIT) routine can also
be used to override options. ICEIEXIT changes override any corresponding
changes made by an EFS program.
7. For OUTFIL statements, override is at the ddname level. See OUTFIL
statements notes on page 374 for details.
Main features of sources of DFSORT options

There are five sources for run-time options that allow you to override your sites
installation defaults. To help you decide which source is most efficient for you for
a given situation, compare their main features using the following lists:
DFSPARM data set

v Use with direct or program invocation.
v Overrides all other sources.
v Accepts all DFSORT program control statements, and all EXEC PARM options,
including those OPTION statement parameters ignored by SYSIN and
SORTCNTL.
v Permits comment statements, blank statements, and remarks.
864
EXEC statement PARM options

v Use with direct invocation only.
v Accepts all EXEC PARM options, including those equivalent to the OPTION
statement parameters ignored by SYSIN and SORTCNTL.
SORTCNTL data set

v Use with program invocation only.
v Accepts all DFSORT program control statements.
v Ignores these OPTION statement parameters: EFS, LIST, NOLIST, LISTX,
NOLISTX, LOCALE, MSGPRT, MSGDDN, SMF, SORTDD, SORTIN, SORTOUT,
and USEWKDD.
v Using multiple parameter lists to rename the SORTCNTL data set permits
different control statements to be used for a program that invokes DFSORT more
than once.
SYSIN data set

v Use with direct invocation only.
v Accepts all DFSORT program control statements.
v Ignores these OPTION statement parameters: EFS, LIST, NOLIST, LISTX,
NOLISTX, LOCALE, MSGPRT, MSGDDN, SMF, SORTDD, SORTIN, SORTOUT,
and USEWKDD.
v Can contain user exit routines in object deck format for binding or link-editing.
Parameter lists
v Use with program invocation only.
v Extended parameter list accepts all DFSORT program control statements,
including those OPTION statement parameters ignored by SYSIN and
SORTCNTL.
v 24-bit parameter list accepts a subset of DFSORT program control statements.
v Using multiple parameter lists to rename the SORTCNTL data set permits
different control statements to be used for a program that invokes DFSORT more
than once.
v Can be used to pass the addresses of any user exits that your program has
placed in main storage.
Note: The extended parameter list can perform a superset of the functions in the
24-bit parameter list.
Override tables
The following tables show the possible sources of specification and order of
override for individual options.
v The order of override between sources of specification is from left to right. A
specification overrides all specifications to its right.
v The order of override within a source is from top to bottom. A specification
overrides all specifications below it.
865
v For DFSPARM, 'keyword' (for example, BSAM) indicates a PARM option,
whereas 'operation keyword' (for example, DEBUG BSAM) indicates a control
statement option.
v The Function columns indicate which functions (S=sort, M=merge, or C=copy)
can use the option.
v Although alias names are available for many of the options, they are not shown
here.
Directly invoked DFSORT

Table 96 on page 867 shows where each sort, merge, or copy option may be
specified when DFSORT is directly invoked (that is, not invoked by programs).
DFSPARM: PARM options selectively override corresponding options in any other
source. DEBUG and OPTION control statement options selectively override
corresponding options in EXEC PARM and SYSIN. Control statements other than
DEBUG and OPTION completely override corresponding control statements in
SYSIN.
EXEC PARM options selectively override options in SYSIN.
SORT and MERGE are considered to be corresponding control statements.
INCLUDE and OMIT are considered to be corresponding control statements.
866
NO
DSA
DSPSIZE
NO
DSA
OPTION DSA
DSPSIZE
OPTION DSPSIZE
CINV|NOCINV
OPTION CINV|NOCINV
NO
CINV|NOCINV
OPTION CHECK|NOCHECK
NO
NO
OPTION CHALT|NOCHALT
NO
NO
DEBUG CFW|NOCFW
DEBUG CTRx
NO
BSAM
DEBUG BSAM
NO
BSAM
AVGRLEN
OPTION AVGRLEN
OPTION COPY
SORT|MERGE FIELDS
AVGRLEN
DEBUG NOASSIST
NO
NO
ARESALL
OPTION ARESALL
INCLUDE|OMIT COND|FORMAT
ARESALL
ALTSEQ CODE
COBEXIT
NO
DEBUG ABSTP
COBEXIT
OPTION COBEXIT
NO
NO
NO
Specified with EXEC

PARM
Specified with DFSPARM
OPTION DSPSIZE
OPTION DSA
NO
NO
DEBUG CTRx
OPTION COPY
SORT|MERGE FIELDS
OPTION COBEXIT
OPTION CINV|NOCINV
DEBUG CFW|NOCFW
DEBUG BSAM
OPTION AVGRLEN
DEBUG NOASSIST
OPTION ARESALL
ALTSEQ CODE
DEBUG ABSTP
NO
Specified with SYSIN
DSPSIZE
DSA
DIAGSIM
Time-of-day for
activation
NO
NO
NO
COBEXIT
CINV
CHECK
CHALT
CFW
NO
NO
NO
ARESALL
ALTSEQ
NO
ABCODE
Installation (JCL,
TSO or TDx)
S,M
S,M,C
S,M,C
Function
S,M
S,M,C
Dataspace sorting
Dynamic storage
adjustment limit
Simulate
SORTDIAG DD
Statement
Simulate
SORTDIAG DD
Statement
ABEND record
count
Copy records
Include|Omit
fields
COBOL library
Control interval
access
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
Record count check S,M,C
CH field sequence
Cache fast write
Force BSAM
Average record
length
Bypass Sorting
Instructions
System storage
S,M,C
above 16MB virtual
Alternate sequence
Abnormal stop
ABEND code
Description of
Option
Table 96. Directly Invoked DFSORT Option Specification/Override. Options are arranged alphabetically on the Installation column. If NO is specified in the
Installation column, move to the next column to the left and so on. The order of override is from left to right and from top to bottom within a row.
867
868
OPTION DYNALLOC
SORT DYNALLOC
NO
OPTION EQUALS|NOEQUALS
SORT|MERGE EQUALS|NOEQUALS
DYNALLOC
DYNAPCT
DYNSPC
EFS
NO
EQUALS|NOEQUALS
NO
ABEND|NOABEND
NO
NO
NO
NO
NO
DYNALLOC
OPTION DYNALLOC|USEWKDD
SORT DYNALLOC
DYNAPCT
OPTION DYNAPCT
DYNSPC
OPTION DYNSPC
EFS
OPTION EFS
NO
EQUALS|NOEQUALS
SORT|MERGE EQUALS|NOEQUALS
DEBUG EQUCOUNT

ABEND|NOABEND
DEBUG ABEND|NOABEND
DEBUG ESTAE|NOESTAE
OPTION EXITCK
NO
NO
NO
NO
NO
NO
OPTION EXITCK
DEBUG ESTAE|NOESTAE
DEBUG ABEND|NOABEND
DEBUG EQUCOUNT
EFS
NO2
EXPRES
EXPOLD
EXPMAX
EXITCK
ESTAE
ERET
NO
EQUALS
ENABLE
DYNSPC
DYNAPCT
DYNAUTO
DYNALOC
Installation (JCL,
TSO or TDx)
OPTION DYNSPC
OPTION DYNAPCT
OPTION DYNALLOC
SORT DYNALLOC
DYNALLOC
Specified with EXEC

PARM
DYNALLOC
OPTION DYNALLOC
SORT DYNALLOC

S
Function
S,M,C
S,M,C
S,M,C
S,M,C
S,M
Available expanded S
storage reserved
for
non-Hipersorting
use
Old expanded
storage limit for all
DFSORT
Hiperspaces
Available expanded S
DFSORT
Hiperspaces
E15/E35 return
code checking
ESTAE routine
Error action
Equal key count

message
Equal record order
Enable Time-of-Day S,M,C

modules
EFS program
specified
Dynamic allocation
default space
Additional work
data sets
Automatic dynamic S
allocation
Dynamic
SORTWKs
Description of
Option
Table 96. Directly Invoked DFSORT Option Specification/Override (continued). Options are arranged alphabetically on the Installation column. If NO is specified in
the Installation column, move to the next column to the left and so on. The order of override is from left to right and from top to bottom within a row.
MODS Exx|HILEVEL=YES
E15=COB E35=COB
NO
NO
E15=COB
PARM E35=COB
MODS Exx|HILEVEL=YES
INREC parameters
JOINKEYS parameters
NO
NO
LIST|NOLIST
RECORD LENGTH
LIST|NOLIST
OPTION LIST|NOLIST
NO
NO
OPTION CKPT
SORT CKPT4
NO
4
NO
FILSZ
FILSZ
OPTION FILSZ|SIZE
SORT|MERGE FILSZ|SIZE
NO
NO
MERGE FILES
NO
NO
SUM FIELDS/FORMAT
HIPRMAX
NO
SORT|MERGE FIELDS|FORMAT
HIPRMAX
OPTION HIPRMAX
MERGE FILES
OPTION FILSZ|SIZE
NO
REFORMAT parameters
JOIN parameters
OUTREC parameters
NO
RECORD LENGTH
NO
OPTION CKPT
SORT CKPT4
NO
NO
OPTION HIPRMAX
SUM FIELDS/FORMAT
REFORMAT parameters
OUTREC parameters
NO
NO
JOIN parameters
JOINKEYS parameters
INREC parameters
Specified with EXEC

PARM
LIST
NO
IOMAXBF
IGNCKPT
IEXIT
IDRCPCT
HIPRMAX
FSZEST
NO
NO
NO
NO
NO
NO
NO
NO
NO
Installation (JCL,
TSO or TDx)
Print DFSORT
control statements5
Record lengths
Maximum
SORTIN/
SORTOUT
data set buffer
space
Checkpoints
ICEIEXIT
IDRC compaction
Hipersorting
File size
Merge input files
Sum fields
Control fields
REFORMAT fields
OUTREC
reformatting
JOIN options
JOINKEYS
processing
INREC
reformatting
User Exit Exx

(xx=11,15-19,
31,35,37-39,
and 61)
Description of
Option
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M
S,M
S,C
S,M,C
S,C
S,C
S,M,C
S,M,C3
Function
869
870
LISTX|NOLISTX
LOCALE
NO
NO
NO
MOSIZE
MOWRK|NOMOWRK
MSGDDN
NO
MSGPRT
NO
NO
NULLOUT
ODMAXBF
OUTFIL9
OUTREL|NOOUTREL
LOCALE
OPTION LOCALE
NO
OPTION MERGEIN
NO
MOSIZE
OPTION MOSIZE
MOWRK|NOMOWRK
OPTION MOWRK|NOMOWRK
MSGDDN
OPTION MSGDDN
NO
MSGPRT
OPTION MSGPRT
OPTION NOBLKSET
NO
NULLOUT
OPTION NULLOUT
ODMAXBF
OPTION ODMAXBF
OUTFIL9
OUTREL|NOOUTREL
OPTION NOOUTREL
Specified with EXEC

PARM
LISTX|NOLISTX
OPTION LISTX|NOLISTX

OPTION NOOUTREL
OUTFIL9
OPTION ODMAXBF
OPTION NULLOUT
NO
OUTREL
NO
ODMAXBF
NULLOUT
NOMSGDD
NO
MSGPRT
NO2
OPTION NOBLKSET
MSGCON
MSGDDN
NO2
NO
MOWRK
MOSIZE
OPTION MOSIZE
MINLIM
NO
NO2
NO
MAXLIM
LOCALE
LISTX
Installation (JCL,
TSO or TDx)
NO
NO2
NO
Function
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
Release output data S,M,C

set space
OUTFIL processing
Maximum OUTFIL
data set buffer
space
Action when no
records for
SORTOUT
Action when
message data set
missing
Bypass Blockset
Print messages
Write messages on
master console
Alternate message
data set
Memory objects as
work storage
Memory object
sorting
Minimum storage
Alternate MERGE
ddnames
Maximum storage
below 16MB
virtual6
Locale processing
Print control
S,M,C
statements returned
by an EFS
program5
Description of
Option
NO
NO
OPTION SORTDD
7
OPTION SORTIN
SOLRF|NOSOLRF
NO
NO
NO
NO2
2
SOLRF
SMF
NO
SIZE
SDBMSG
SDB
RESET
RESALL
PARMDDN
PAD
OVFLO
OVERRGN
OUTSEC
Installation (JCL,
TSO or TDx)
OPTION SOLRF|NOSOLRF
NO
NO
SOLRF|NOSOLRF
NO
OPTION SDB
OPTION SMF
SDB
SDB
OPTION SDB
OPTION RESET|NORESET
OPTION SKIPREC
SORT SKIPREC
RESET|NORESET
RESET|NORESET
OPTION RESALL
SKIPREC
RESALL
RESALL
OPTION RESALL
NO
SKIPREC
OPTION SKIPREC
SORT SKIPREC
NO
NO
OPTION PAD
OPTION MAINSIZE
PAD
PAD
OPTION PAD
OPTION OVFLO
SIZE
OVFLO
OVFLO
OPTION OVFLO
NO
SIZE
OPTION MAINSIZE
NO
NO
OPTION NOOUTSEC
NO
NO
OPTION NOOUTSEC
NO
Specified with EXEC

PARM
Alternate SORTIN
ddname
ddname prefix
SORTOUT length
SMF records
Skip records
Storage
Systemdetermined
block size for
message and
list data sets
Systemdetermined
output data
set block size
NEW or MOD
VSAM output
System reserved
storage6
Alternate ddname
for DFSPARM
DFSORT LRECL
padding action
Summary fields
overflow action
Storage over
REGION
Output data set

secondary
allocation
Description of
Option
S,C
S,M,C
S,M,C
S,M,C
S,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
Function
871
872
OPTION SPANINC
OPTION STOPAFT
SORT STOPAFT
NO
SPANINC
STOPAFT
NO
SZERO|NOSZERO
NO
NO
NO
TRUNC
NO
VERIFY|NOVERIFY
NO
VLLONG|NOVLLONG
VLSCMP|NOVLSCMP
VLSHRT|NOVLSHRT
NO
OPTION SORTOUT8
SPANINC
OPTION SPANINC
STOPAFT
OPTION STOPAFT
SORT STOPAFT
NO
SZERO|NOSZERO
OPTION SZERO|NOSZERO
NO
NO

NO
TRUNC
OPTION TRUNC
RECORD TYPE
VERIFY|NOVERIFY
OPTION VERIFY|NOVERIFY
NO
VLLONG|NOVLLONG
OPTION VLLONG|NOVLLONG
VLSCMP|NOVLSCMP
OPTION VLSCMP|NOVLSCMP
VLSHRT|NOVLSHRT
OPTION VLSHRT|NOVLSHRT
NO
NO
NO
RECORD TYPE
OPTION TRUNC
NO
NO
NO
NO
NO2
NO
NO
NO
Specified with EXEC

PARM
VSAMBSP
VLSHRT
VLSCMP
VLLONG
VIO
VERIFY
NO
TRUNC
TUNE
TMAXLIM
TEXIT
SZERO
SVC
NO
SPANINC
NO
SORTLIB
Installation (JCL,
TSO or TDx)
S,M,C
S,C
S,M,C
S,M,C
S,M
Function
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M,C
VSAM buffer space S
Action for short

control or compare
fields
Pad short compare

fields
Truncate long
output records
SORTWK virtual
I/O
Sequence check
Record format
DFSORT LRECL
truncation action
Optimize central
storage or disk
work space
Maximum storage
above and below
16MB virtual6
ICETEXIT
Signed or unsigned S,M,C

zero
DFSORT SVC
Information
Input limit
Incomplete
spanned records
action
Alternate
SORTOUT ddname
Conventional
modules library
Description of
Option
OPTION WRKSEC|NOWRKSEC
OPTION Y2PAST
SORT|MERGE Y2PAST
WRKREL|NOWRKREL
WRKSEC|NOWRKSEC
Y2PAST
ZDPRINT|NZDPRINT
WRKREL|NOWRKREL
OPTION WRKREL|NOWRKREL
WRKSEC|NOWRKSEC
Y2PAST
OPTION Y2PAST
SORT|MERGE Y2PAST
ZDPRINT|NZDPRINT
OPTION ZDPRINT|NZDPRINT
OPTION VSAMIO|NOVSAMIO
VSAMIO|NOVSAMIO
VSAMIO|NOVSAMIO
OPTION VSAMEMT|NVSAMEMT
VSAMEMT|NVSAMEMT
Specified with EXEC

PARM
VSAMEMT|NVSAMEMT
ZDPRINT
Y2PAST
WRKSEC
WRKREL
VSAMIO
VSAMEMT
Installation (JCL,
TSO or TDx)
ZD SUM results
Set century
window
SORTWK
secondary
allocation
Release SORTWK
space
Same VSAM input

and output
Emty VSAM input
Description of
Option
S,M
S,M,C
S,M,C
Function
873
Notes to directly invoked DFSORT table

1
Does not request dynamic allocation; only supplies defaults.
Not used in SYSIN.
All functions do not apply to all exits. See Table 62 on page 491 and
Table 63 on page 491 for applicable exits.
Not used if Blockset is selected and IGNCKPT=YES was specified.
Not used if MSGPRT=NONE is in effect; in this case control statements are

not printed.
Not used unless MAINSIZE=MAX is in effect.
Overrides SORTDD for the SORT input ddname.
Overrides SORTDD for the SORT output ddname.
Override is at the ddname level.
Program invoked DFSORT with the extended parameter list

specified when DFSORT is program invoked and an extended parameter list is
passed to it.
corresponding options in SORTCNTL and the Parameter List. Control statements
other than DEBUG and OPTION completely override corresponding control
statements in SORTCNTL and the Parameter List.
SORTCNTL: DEBUG and OPTION control statement options selectively override
corresponding options in the Parameter List. Control statements other than DEBUG
and OPTION completely override corresponding control statements in the
Parameter List.
874
ALTSEQ CODE
OPTION ARESALL
OPTION ARESINV
DEBUG NOASSIST
OPTION AVGRLEN
DEBUG BSAM
DEBUG CFW|NOCFW
OPTION CINV|NOCINV
OPTION COBEXIT
DEBUG ABSTP
ALTSEQ CODE
ARESALL
OPTION ARESALL
OPTION ARESINV
DEBUG NOASSIST
AVGRLEN
OPTION AVGRLEN
BSAM
DEBUG BSAM
DEBUG CFW|NOCFW
CINV|NOCINV
OPTION CINV|NOCINV
COBEXIT
OPTION COBEXIT
OPTION COBEXIT
OPTION CINV|NOCINV
DEBUG CFW|NOCFW
DEBUG BSAM
OPTION AVGRLEN
DEBUG NOASSIST
OPTION ARESINV
OPTION ARESALL
Offset 16 entry
ALTSEQ CODE
DEBUG ABSTP
NO
COBEXIT
CINV
CHECK
CHALT
CFW
NO
NO
NO
ARESINV
ARESALL
ALTSEQ
NO
ABCODE
Specified with Extended Parameter Installation (INV,

List
TSOINV or TDx)
OPTION COPY
SORT|MERGE FIELDS2
DEBUG CTRx
NO
NO
OPTION DSA
OPTION DSPSIZE
OPTION COPY
SORT|MERGE FIELDS
DEBUG CTRx
NO
NO
DSA
OPTION DSA
DSPSIZE
OPTION DSPSIZE
OPTION DSPSIZE
OPTION DSA
NO
NO
DEBUG CTRx
OPTION COPY
SORT|MERGE FIELDS
DSPSIZE
DSA
DIAGSIM
day
NO
NO
INCLUDE|OMIT COND|FORMAT INCLUDE|OMIT COND|FORMAT INCLUDE|OMIT COND|FORMAT NO
NO
DEBUG ABSTP
NO
Specified with SORTCNTL
dataspace sorting
Dynamic storage
adjustment limit
Simulate SORTDIAG
DD statement
Time-of-day for
activation
ABEND record count
Copy records
Include|Omit fields
COBOL library
Control interval access
Record count check
CH field sequence
Cache fast write
Force BSAM
Average record length
Bypass Sorting
Instructions
Storage above 16MB

virtual for invoking
program
System storage above

16MB virtual
Alternate sequence
Abnormal stop
ABEND code
Description of Option
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
Function
Table 97. Extended Parameter List DFSORT Option Specification/Override. Options are arranged alphabetically on the Installation column. If NO is specified in the
Installation column, move to the next column to the left and so on. The order of override is from left to right and from top to bottom within a row.
875
876
OPTION DYNALLOC|
USEWKDD
SORT DYNALLOC
OPTION DYNSPC
OPTION EFS
NO
SORT|MERGE
EQUALS|NOEQUALS
NO
Offset 4 entry4
MODS E154|HILEVEL=YES
OPTION DYNALLOC
SORT DYNALLOC2
OPTION DYNALLOC
SORT DYNALLOC
OPTION DYNAPCT
OPTION DYNSPC
NO3
NO
SORT|MERGE
EQUALS|NOEQUALS2
DEBUG EQUCOUNT
DEBUG ABEND|NOABEND
DEBUG ESTAE|NOESTAE
OPTION EXITCK
NO
NO
NO
DYNALLOC
OPTION DYNALLOC
SORT DYNALLOC
DYNALLOC
OPTION DYNALLOC|
USEWKDD
SORT DYNALLOC
DYNAPCT
OPTION DYNAPCT
DYNSPC
OPTION DYNSPC
EFS
OPTION EFS
NO
EQUALS|NOEQUALS
SORT|MERGE
EQUALS|NOEQUALS
DEBUG EQUCOUNT

ABEND|NOABEND
DEBUG ABEND|NOABEND
DEBUG ESTAE|NOESTAE
OPTION EXITCK
NO
NO
NO
E15=COB
NO
NO
OPTION EXITCK
DEBUG ESTAE|NOESTAE
DEBUG ABEND|NOABEND
DEBUG EQUCOUNT
OPTION DYNAPCT
OPTION DYNALLOC
SORT DYNALLOC
NO
EXPRES
EXPOLD
EXPMAX
EXITCK
ESTAE
ERET
Exit E15
Available expanded
storage reserved for
non-Hipersorting use
Old expanded storage

limit for all DFSORT
Hiperspaces
Available expanded
DFSORT Hiperspaces
E15/E35 return code

checking
ESTAE routine
Error action
Equal key count

message
Equal record order
EQUALS
NO
Enable Time-of-Day
modules
EFS program specified
Dynamic allocation
default space
Additional work data

sets
Automatic
DYNALLOC
Dynamic SORTWKs
ENABLE
EFS
DYNSPC
DYNAPCT
DYNAUTO
DYNALOC

List
TSOINV or TDx)
S,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
Function
Table 97. Extended Parameter List DFSORT Option Specification/Override (continued). Options are arranged alphabetically on the Installation column. If NO is
specified in the Installation column, move to the next column to the left and so on. The order of override is from left to right and from top to bottom within a row.
REFORMAT parameters
NO
NO
NO
NO
OPTION HIPRMAX
NO
NO
OPTION CKPT6
SORT|MERGE CKPT2,6
NO
NO
NO
HIPRMAX
OPTION HIPRMAX
NO
NO
OPTION CKPT6
SORT|MERGE CKPT6
OPTION CKPT6
SORT|MERGE CKPT6
NO
NO
OPTION HIPRMAX
NO
NO
IGNCKPT
IEXIT
IDRCPCT
HIPRMAX
GNTRUNC
GNPAD
GENER
OPTION FILSZ|SIZE
SORT|MERGE FILSZ|SIZE2
NO
NO
FSZEST
MERGE FILES
OPTION FILSZ|SIZE
MERGE FILES
FILSZ
OPTION FILSZ|SIZE
NO
NO
MERGE FILES
SUM FIELDS|FORMAT
NO
NO
NO
NO
NO
SUM FIELDS|FORMAT
OUTREC parameters
JOIN parameters
JOINKEYS parameters
INREC parameters
MODS Exx
NO
NO
NO
NO
SUM FIELDS|FORMAT
REFORMAT parameters
REFORMAT parameters
JOINKEYS parameters
JOINKEYS parameters
JOIN parameters
INREC parameters
INREC parameters
OUTREC parameters
MODS Exx
MODS Exx
OUTREC parameters
MODS E394
MODS E394
JOIN parameters
Offset 8 entry4
Offset 28 entry4
MODS E394
Offset 4 entry
NO
Offset 24 entry
MODS E184

List
TSOINV or TDx)
E35=COB
MODS E18
NO
MODS E18
Checkpoints
ICEIEXIT
IDRC compaction
Hipersorting
ICEGENER LRECL
truncation action
ICEGENER LRECL
padding action
IEBGENER name
File size
Merge input files
Sum fields
Control fields
REFORMAT fields
OUTREC reformatting
JOIN options
JOINKEYS processing
INREC reformatting
User Exit Exx

(xx=11,16,17,19,
31,37,38, and 61)
Exit E39
Exit E35
Exit E32
Exit E18
S,M,C
S,M
S,M
S,M
S,C
S,M,C
S.C
S,C
S,M,C
S,M,C5
S,M,C
S,M,C
Function
877
878
RECORD LENGTH
RECORD LENGTH
OPTION LOCALE
OPTION MSGDDN
NO
OPTION MSGPRT
NO3
NO3
NO
NO
NO
OPTION MOSIZE
NO3
NO
NO3
OPTION NOBLKSET
NO
OPTION NULLOUT
OPTION ODMAXBF
OUTFIL11
LISTX|NOLISTX
LOCALE
OPTION LOCALE
NO
OPTION MERGEIN
NO

MOSIZE
OPTION MOSIZE
MOWRK|NOMOWRK
MSGDDN
OPTION MSGDDN
NO
MSGPRT
OPTION MSGPRT
OPTION NOBLKSET
NO
NULLOUT
OPTION NULLOUT
ODMAXBF
OPTION ODMAXBF
OUTFIL11
OUTFIL11
OPTION ODMAXBF
OPTION NULLOUT
NO
OPTION NOBLKSET
OPTION MOSIZE
NO
OPTION MERGEIN
NO
NO
OPTION LIST|NOLIST
RECORD LENGTH
NO
NO
ODMAXBF
NULLOUT
NOMSGDD
NO
MSGPRT
MSGCON
MSGDDN
MOWRK
MOSIZE
MINLIM
NO
MAXLIM
LOCALE
LISTX
LIST
NO
IOMAXBF

List
TSOINV or TDx)
LIST|NOLIST
OPTION LIST|NOLIST
NO
NO
S,M,C
S,M,C
S,M,C
Function
OUTFIL processing
Maximum OUTFIL
data set buffer space
Action when no
records for SORTOUT
Action when message

data set missing
Bypass Blockset
Print messages
Write messages on
master console
Alternate message
ddname
Memory objects as
work storage
Memory object sorting
Minimum storage
Alternate MERGE
ddnames
Maximum storage
below 16MB virtual8
Locale processing
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
Print control statements S,M,C

returned by an EFS
program7
Print DFSORT control

statements7
Record lengths
Maximum
SORTIN/SORTOUT
NO
OPTION SORTDD
NO3
OPTION SORTDD
OPTION SORTIN
OPTION SORTIN
OPTION SMF
OPTION SKIPREC
SORT|MERGE SKIPREC
OPTION MAINSIZE
NO
NO
OPTION SDB
OPTION RESINV
SOLRF|NOSOLRF
OPTION RESINV
OPTION RESINV
OPTION SMF
RESET|NORESET
OPTION RESALL
OPTION SKIPREC
SORT|MERGE SKIPREC2
OPTION RESALL
RESALL
OPTION RESALL
NO
SKIPREC
OPTION SKIPREC
SORT|MERGE SKIPREC
NO
NO
OPTION PAD
OPTION MAINSIZE
OPTION PAD
PAD
OPTION PAD
OPTION OVFLO
NO
SIZE
OPTION MAINSIZE
OPTION OVFLO
OVFLO
OPTION OVFLO
NO
NO
NO
OPTION NOOUTSEC
NO
OPTION NOOUTSEC
OPTION NOOUTSEC
OPTION NOOUTREL
OPTION SDB
OPTION NOOUTREL
OUTREL|NOOUTREL
OPTION NOOUTREL
NO
NO
SOLRF
SMF
NO
SIZE
SDBMSG
SDB
RESINV
RESET
RESALL
PARMDDN
PAD
OVFLO
OVERRGN
OUTSEC
OUTREL

List
TSOINV or TDx)
SDB
OPTION SDB
Alternate SORTIN
ddname
ddname prefix
SORTOUT length
SMF records
Skip records
Storage
System-determined
block size for message
and list data sets
System-determined
output data set block
size
Program reserved
storage8
NEW or MOD VSAM

output
System reserved
storage8
Alternate ddname for

DFSPARM
DFSORT LRECL
padding action
Summary fields
overflow action
Storage over REGION
Output data set

secondary allocation
Release output data set

space
S,C
S,M,C
S,M,C
S,M,C
S,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S,M,C
Function
879
880
OPTION SPANINC
OPTION STOPAFT
SORT|MERGE STOPAFT
NO
NO3
OPTION SPANINC
OPTION STOPAFT
SORT|MERGE STOPAFT2
NO
NO
NO
NO
OPTION TRUNC
RECORD TYPE
NO
NO
NO
OPTION SORTOUT10
SPANINC
OPTION SPANINC
STOPAFT
OPTION STOPAFT
SORT|MERGE STOPAFT
NO
SZERO|NOSZERO
NO
NO
NO

TRUNC
OPTION TRUNC
RECORD TYPE
VERIFY|NOVERIFY
NO
VLLONG|NOVLLONG
VLSCMP|NOVLSCMP
VLSHRT|NOVLSHRT
NO
NO
NO
RECORD TYPE
OPTION TRUNC
NO
NO
NO
NO
OPTION SORTOUT10
NO
VSAMBSP
VLSHRT
VLSCMP
VLLONG
VIO
VERIFY
NO
TRUNC
TUNE
TMAXLIM
TEXIT
SZERO
SVC
NO
SPANINC
NO
SORTLIB

List
TSOINV or TDx)
S,M,C
S,M,C
S,M,C
S,C
S,M,C
S,M,C
S,M
Function
S,M,C
S,M,C
S,M
S,M,C
S,M,C
VSAM buffer space
Action for short control S,M,C

field or compare field
Pad short compare

fields
Ttruncate long output

records
SORTWK virtual I/O
Sequence check
Record format
DFSORT LRECL
truncation action
Optimize central
storage or disk work
space
Maximum storage
S,M,C
above and below 16MB
virtual8
ICETEXIT
Signed or unsigned
zero
DFSORT SVC
information
Input limit
Incomplete spanned
records action
Alternate SORTOUT
ddname
Conventional modules
library
OPTION Y2PAST
SORT|MERGE Y2PAST
OPTION Y2PAST
SORT|MERGE Y2PAST2
VSAMIO|NOVSAMIO
WRKREL|NOWRKREL
OPTION
WRKREL|NOWRKREL
WRKSEC|NOWRKSEC
OPTION WRKSEC|
NOWRKSEC
Y2PAST
OPTION Y2PAST
SORT|MERGE Y2PAST
ZDPRINT|NZDPRINT
VSAMEMT|NVSAMEMT
OPTION VSAMEMT|
NVSAMEMT
ZDPRINT
Y2PAST
WRKSEC
WRKREL
VSAMIO
VSAMEMT

List
TSOINV or TDx)
S,M,C
Function
ZD SUM results
Set century window
SORTWK secondary
allocation
S,M
S,M,C
Release SORTWK space S
Same VSAM input and

output
Empty VSAM input
881
Notes to extended parameter list table

1
Does not override corresponding option in an OPTION statement specified

via the extended parameter list.
Not used in SORTCNTL.
DFSORT terminates if the exit is specified via the parameter list entry and
the exit is specified in a MODS statement.
All functions do not apply to all exits. See Table 62 on page 491 and
Table 63 on page 491 for applicable exits.
Not used if MSGPRT=NONE is in effect; in this case control statements are

not printed.
Overrides SORTDD for the sort input ddname.
10
Overrides SORTDD for the sort output ddname.
11
Program invoked DFSORT with the 24-bit parameter list

specified when DFSORT is program invoked and a 24-bit parameter list is passed
to it.
corresponding options in SORTCNTL and the Parameter List. Control statements
other than DEBUG and OPTION completely override corresponding control
statements in SORTCNTL and the Parameter List.
SORTCNTL: DEBUG control statement options selectively override corresponding
options in the Parameter List. Control statements other than DEBUG completely
override corresponding control statements in the Parameter List.
882
NO
DEBUG ABSTP
ALTSEQ CODE
OPTION ARESALL
OPTION ARESINV
DEBUG NOASSIST
OPTION AVGRLEN
DEBUG BSAM
DEBUG CFW|NOCFW
OPTION CINV|NOCINV
OPTION COBEXIT
OPTION COPY
SORT|MERGE FIELDS
DEBUG CTRx
NO
NO
OPTION DSA
NO
DEBUG ABSTP
ALTSEQ CODE
ARESALL
OPTION ARESALL
OPTION ARESINV
DEBUG NOASSIST
AVGRLEN
OPTION AVGRLEN
BSAM
DEBUG BSAM
DEBUG CFW|NOCFW
CINV|NOCINV
OPTION CINV|NOCINV
COBEXIT
OPTION COBEXIT
OPTION COPY
SORT|MERGE FIELDS
DEBUG CTRx
NO
NO
DSA
OPTION DSA
NO
NO
NO
DEBUG CTRx
SORT|MERGE FIELDS
INCLUDE|OMIT
COND|FORMAT
NO
NO
NO
NO
DEBUG CFW|NOCFW
DEBUG BSAM
NO
DEBUG NOASSIST
NO
NO
X'F6' entry
ALTSEQ CODE
DEBUG ABSTP
NO
Specified with 24-Bit List
DSA
DIAGSIM
day
NO
NO
NO
COBEXIT
CINV
CHECK
CHALT
CFW
NO
NO
NO
ARESINV
ARESALL
ALTSEQ
NO
ABCODE
S,M
S,M,C
S,M,C
Function
S,M,C
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
Dynamic storage
adjustment limit
Simulate SORTDIAG
DD statement
Time-of-day for
activation
S,M,C
S,M,C
ABEND record count S,M
Copy records
Include|Omit fields
COBOL library
Control interval
access
Record count check
CH field sequence
Cache fast write
Force BSAM
Average record
length
Bypass Sorting
Instructions
Storage above 16MB

virtual for invoking
program
System storage above S,M,C

16MB virtual
Alternate sequence
Abnormal stop
ABEND code
Installation (INV, TSOINV Description of

or TDx)
Option
Table 98. 24-Bit List DFSORT Option Specification/Override. Options are arranged alphabetically on the Installation column. If NO is specified in the Installation
column, move to the next column to the left and so on.
883
884
NO
SORT DYNALLOC
SORT DYNALLOC
NO
NO
OPTION DSPSIZE
OPTION DYNALLOC
SORT DYNALLOC
OPTION DYNALLOC
SORT DYNALLOC
OPTION DYNAPCT
OPTION DYNSPC
NO2
NO
SORT|MERGE EQUALS|
NOEQUALS
DEBUG EQUCOUNT
DEBUG ABEND|NOABEND
DEBUG ESTAE|NOESTAE
OPTION EXITCK
NO
NO
NO
DSPSIZE
OPTION DSPSIZE
DYNALLOC
OPTION DYNALLOC
SORT DYNALLOC
DYNALLOC
OPTION DYNALLOC|USEWKDD
SORT DYNALLOC
DYNAPCT
OPTION DYNAPCT
DYNSPC
OPTION DYNSPC
EFS
OPTION EFS
NO
EQUALS|NOEQUALS
SORT|MERGE EQUALS|
NOEQUALS

DEBUG EQUCOUNT
ABEND|NOABEND
DEBUG ABEND|NOABEND
DEBUG ESTAE|NOESTAE
OPTION EXITCK
NO
NO
NO
NO
EQUALS
ENABLE
EFS
DYNSPC
DYNAPCT
NO
NO
NO
NO
DEBUG ESTAE|NOESTAE
EXPRES
EXPOLD
EXPMAX
EXITCK
ESTAE
Function
S,M,C
S,M,C
S,M,C
S,M
S,M,C
S,M,C
S
Available expanded
storage reserved for
non-Hipersorting use
S
Old expanded
DFSORT Hiperspaces
S
Available expanded
DFSORT Hiperspaces
E15/E35 return code

checking
ESTAE routine
Error action
Equal key count

message
Equal record order
Enable Time-of-Day
modules
EFS program
specified
Dynamic allocation
default space
Additional work data S

sets
Automatic
DYNALLOC
Dynamic SORTWKs
DYNALOC1
DYNAUTO
dataspace sorting
DSPSIZE

or TDx)
Option
DEBUG ABEND|NOABEND ERET
DEBUG EQUCOUNT
SORT|MERGE
EQUALS|NOEQUALS
NO
NO
Table 98. 24-Bit List DFSORT Option Specification/Override (continued). Options are arranged alphabetically on the Installation column. If NO is specified in the
Installation column, move to the next column to the left and so on.
NO
MODS E35 |HILEVEL=YES
MODS Exx
NO
E35=COB
MODS Exx
X'04' entry
MERGE FILES
JOIN parameters
OUTREC parameters
REFORMAT parameters
SUM FIELDS|FORMAT
MERGE FILES
OPTION FILSZ|SIZE
NO
NO
NO
OPTION HIPRMAX
NO
NO
JOIN parameters
OUTREC parameters
REFORMAT parameters
SUM FIELDS|FORMAT
MERGE FILES
FILSZ
OPTION FILSZ|SIZE
NO
NO
NO
HIPRMAX
OPTION HIPRMAX
NO
NO
INREC parameters

NO
NO
NO
NO
NO
NO
SUM FIELDS|FORMAT
SORT|MERGE
FIELDS|FORMAT
REFORMAT parameters
OUTREC parameters
JOIN parameters
JOINKEYS parameters
INREC parameters
JOINKEYS parameters
JOINKEYS parameters
MODS Exx
Offset 22 entry
MODS E353|
HILEVEL=YES
Offset 18 entry
Offset 18 entry
MODS E153|
HILEVEL=YES
INREC parameters
MODS E15 |
HILEVEL=YES
E15=COB
MODS E153|
HILEVEL=YES
IEXIT
IDRCPCT
HIPRMAX
GNTRUNC
GNPAD
GENER
FSZEST
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
ICEIEXIT
IDRC compaction
Hipersorting
ICEGENER LRECL
truncation action
ICEGENER LRECL
padding action
IEBGENER name
File size
Merge input files
Sum fields
Control fields
REFORMAT fields
OUTREC
reformatting
JOIN options
JOINKEYS
processing
INREC reformatting
User Exit Exx

(xx=11,16-19,
31,37-39, and 61)
User exit E35
User exit E32
User exit E15

or TDx)
Option
S,M,C
S,M
S,M
S,M,C
S,C
S,M,C
S,C
S,C
S,M,C
S,M,C4
S,M,C
S,C
Function
885
886
NO
NO
NO
NO
NO
X'03' entry
NO
X'FF' entry
NO2
NO2
NO
NO2
NO
OPTION MOSIZE
NO2
NO
NO2
OPTION NOBLKSET
NO
LISTX|NOLISTX
LOCALE
OPTION LOCALE
NO
OPTION MERGEIN
NO
MOSIZE
OPTION MOSIZE
MOWRK|NOMOWRK
MSGDDN
OPTION MSGDDN
NO
MSGPRT
OPTION MSGPRT
OPTION NOBLKSET
NO
NO
NO
NO
NO
NO
RECORD LENGTH
NO
RECORD LENGTH
RECORD LENGTH
NO
LIST|NOLIST
OPTION LIST|NOLIST
NO
NO
SORT|MERGE CKPT
OPTION CKPT
SORT|MERGE CKPT5
OPTION CKPT
SORT|MERGE CKPT5
NOMSGDD
NO
MSGPRT
MSGCON
MSGDDN
MOWRK
MOSIZE
MINLIM
NO
MAXLIM
LOCALE
LISTX
LIST
NO
IOMAXBF
IGNCKPT
Action when
message data set
missing
Bypass Blockset
Print messages
Write messages on
master console
Alternate message
ddname
Memory objects as
work storage
Memory object
sorting
Minimum storage
Alternate MERGE
ddnames
Maximum storage
below 16MB virtual7
Locale processing
Print control
statements returned
by an EFS program6
Print DFSORT
control statements6
Record lengths
Maximum
SORTIN/
SORTOUT
data set buffer
space
Checkpoints

or TDx)
Option
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
Function

OPTION NULLOUT
OPTION ODMAXBF
OUTFIL10
OPTION NOOUTREL
OPTION NOOUTSEC
NO
OPTION OVFLO
OPTION PAD
NO
OPTION RESALL
OPTION RESINV
OPTION SDB
NO
OPTION MAINSIZE
OPTION SKIPREC
SORT|MERGE SKIPREC

NULLOUT
OPTION NULLOUT
ODMAXBF
OPTION ODMAXBF
OUTFIL10
OUTREL|NOOUTREL
OPTION NOOUTREL
OPTION NOOUTSEC
NO
OVFLO
OPTION OVFLO
PAD
OPTION PAD
NO
RESALL
OPTION RESALL
RESET|NORESET
OPTION RESINV
SDB
OPTION SDB
NO
SIZE
OPTION MAINSIZE
SKIPREC
OPTION SKIPREC
SORT|MERGE SKIPREC
SORT|MERGE SKIPREC
X'00' entry
NO
NO
X'01' entry
NO
NO
NO
NO
NO
NO
NO
NO
OUTFIL10
NO
NO
NO
SIZE
SDBMSG
SDB
RESINV
RESET
RESALL
PARMDDN
PAD
OVFLO
OVERRGN
OUTSEC
OUTREL
NO
ODMAXBF
NULLOUT
S,M,C
S,M
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
Function
S,M,C
S,M,C
S,M,C
S,M,C
Skip records
Storage
S,C
S,M,C
S,M,C
System-determined
block size for
message and list data
sets
System-determined
output data set block
size
Program reserved
storage7
NEW or MOD
VSAM output
System reserved
storage7
Alternate ddname for S,M,C

DFSPARM
DFSORT LRECL
padding action
Summary fields
overflow action
Storage over
REGION
Output data set

secondary allocation
Release output data

set space
OUTFIL processing
Maximum OUTFIL
Action when no
records for
SORTOUT

or TDx)
Option
887
888
OPTION SORTDD
8

OPTION TRUNC
TRUNC
OPTION TRUNC
VLLONG|NOVLLONG
NO
NO
NO
NO
NO
NO
NO
NO
RECORD TYPE
SZERO|NOSZERO
NO
NO
VERIFY|NOVERIFY
SORT|MERGE STOPAFT
OPTION STOPAFT
SORT|MERGE STOPAFT
STOPAFT
OPTION STOPAFT
SORT|MERGE STOPAFT
RECORD TYPE
NO
OPTION SPANINC
SPANINC
OPTION SPANINC
NO
NO
NO
RECORD TYPE
NO
NO
NO
NO
NO
NO
NO
NO2
OPTION SORTOUT9
NO
NO
NO
NO
NO
NO2
SOLRF|NOSOLRF
OPTION SORTIN
Prefix entry
NO
OPTION SMF
NO
NO
VLLONG
VIO
VERIFY
NO
TRUNC
TUNE
TMAXLIM
TEXIT
SZERO
SVC
NO
SPANINC
NO
SORTLIB
NO
NO
SOLRF
SMF
S,M
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,M,C
S,C
S,M,C
S,M,C
S,M
S,C
S,M,C
S,M,C
S,M,C
Function
Truncate long output

records
S,M,C
SORTWK virtual I/O S
Sequence check
Record format
DFSORT LRECL
truncation action
Optimize central
storage or disk work
space
Maximum storage
above and below
16MB virtual7
ICETEXIT
Signed or unsigned
zero
DFSORT SVC
information
Input limit
Incomplete spanned
records action
Alternate SORTOUT
ddname
Conventional
modules library
Alternate SORTIN
ddname
ddname prefix
SORTOUT length
SMF records

or TDx)
Option
NO
SORT|MERGE Y2PAST
OPTION WRKSEC|
NOWRKSEC
OPTION Y2PAST
SORT|MERGE Y2PAST
VSAMEMT|NVSAMEMT
OPTION VSAMEMT|
NVSAMEMT
VSAMIO|NOVSAMIO
WRKREL|NOWRKREL
WRKSEC|NOWRKSEC
Y2PAST
OPTION Y2PAST
SORT|MERGE Y2PAST
NO
NO
NO
ZDPRINT|NZDPRINT
NO
VLSHRT|NOVLSHRT
NO
NO
NO
NO
NO
VLSCMP|NOVLSCMP
ZDPRINT
Y2PAST
WRKSEC
WRKREL
VSAMIO
VSAMEMT
VSAMBSP
VLSHRT
VLSCMP
ZD SUM results
Set century window
SORTWK secondary
allocation
Release SORTWK
space
Same VSAM input

and output
Empty VSAM input
VSAM buffer space
Action for short

control or compare
field
Pad short compare

fields

or TDx)
Option
S,M
S,M,C
S,M,C
S,M,C
S,M,C
Function
889
Notes to 24-bit list table
890
Not used in SORTCNTL.
DFSORT terminates if the exit is specified via the parameter list entry and
the user exit is specified in a MODS statement.
All functions do not apply to all user exits. See Table 62 on page 491 and
Table 63 on page 491 for applicable user exits.
Not used if MSGPRT=NONE or MSGPRT=CRITICAL is in effect; in this

case control statements are not printed.
Overrides SORTDD for the sort input ddname.
Overrides SORTDD for the sort output ddname.
10

DFSORT data formats
DFSORT supports a large number of data formats as described below.
Description
CH
(character EBCDIC, unsigned). Each character is represented by its 8-bit EBCDIC code.
Example: AB7 becomes
C1
C2
F7
11000001 11000010 11110111
Note:
Hexadecimal
Binary
1. If CHALT is in effect, a format CH field collates according to the ALTSEQ (alternate collating sequence)
table in effect. AQ format can be used for the same purpose.
2. If locale processing is in effect, a format CH field collates according to the collating rules of the active
locale.
ZD
(zoned decimal, signed). Each digit of the decimal number is converted into its 8-bit EBCDIC representation.
The sign indicator replaces the first four bits of the low order byte of the number.
Example: -247 becomes
2
4
- 7
Decimal
F2
F4
D7
Hexadecimal
11110010 11110100 11010111 Binary
The number +247 becomes
F2
F4
C7
11110010 11110100 11000111
Note:
1. The following are treated as positive sign indicators: F, E, C, A, 8, 6, 4, 2, 0.
2. The following are treated as negative sign indicators: D, B, 9, 7, 5, 3, 1.
3. For SUM processing, 0 through 9 for the sign or A through F for a digit results in a data exception (0C7
ABEND). For example, a ZD value such as 3.5 (X'F34BF5') results in an 0C7 because B is treated as an
invalid digit. ICETOOL's DISPLAY or VERIFY operator can be used to identify ZD values with invalid
digits. ICETOOL's VERIFY operator can be used to identify ZD values with invalid signs.
4. The first four bits of the last digit is the sign indicator. The first four bits of each other digit is ignored.
Thus the EBCDIC strings '0025' and ' 25' are both treated as 25 because a leading blank (X'40') is
equivalent to a 0 digit (X'F0').
PD
(packed decimal, signed). Each digit of the decimal number is converted into its 4-bit binary equivalent. The
sign indicator is put into the rightmost four bits of the number.
Example: -247 becomes
2
4
7Decimal
2
4
7D
Hexadecimal
00100100 01111101
Binary
The number +247 becomes 247C in hexadecimal.
Note:
1. The following are treated as positive sign indicators: F, E, C, A, 8, 6, 4, 2, 0.
2. The following are treated as negative sign indicators: D, B, 9, 7, 5, 3, 1.
3. For SUM processing, 0 through 9 for the sign or A through F for a digit results in a data exception (0C7
ABEND). For example, a PD value such as X' 0123BF' results in an 0C7 because B is treated as an invalid
digit. ICETOOL's DISPLAY or VERIFY operator can be used to identify PD values with invalid digits.
ICETOOL's VERIFY operator can be used to identify PD values with invalid signs.
891
Data Format Examples

Description
PD0
(packed decimal, with sign and first digit ignored) The PD0 format can be represented as follows:
xddd...ds
x is hexadecimal 0-F and is ignored.
d is hexadecimal 0-9 and represents a decimal digit.
s is hexadecimal 0-F and is ignored.
PD0 can be used for parts of PD fields. For example, in the PD field P'mmddyy' (hexadecimal 0mmddyyC),
PD0 can be used separately for 0mmd (mm), mddy (dd) and dyyC (yy).
FI
(fixed-point, signed). The complete number is represented by its binary equivalent with the sign indicator
placed in the most significant bit position. 0 for + or 1 for -. Negative numbers are in 2's complement form.
Example: +247 becomes in halfword form
00F7
Hexadecimal
0000000011110111 Binary
-247 becomes in halfword form
FF09
Hexadecimal
1111111100001001 Binary
BI
(binary unsigned). Any bit pattern.
FL
( hexadecimal floating-point, signed). The specified number is in the two-part format of characteristic and
fraction with the sign indicator in bit position 0.
Example: +247 becomes
42F70000...
Hexadecimal
0 1000010 111101110000000....... Binary
+ chara. Fraction
-247 is identical, except that the sign bit is changed to 1.
C2F70000...
Hexadecimal
1 1000010 111101110000000....... Binary
- chara. Fraction
AQ
(character EBCDIC, with alternate collating sequence, unsigned). This is similar to format CH, but the
characters collate according to the ALTSEQ (alternate collating sequence) table in effect.
AC
(character EBCDIC, with ASCII collating sequence, unsigned). This is similar to format CH, but the
characters collate according to the ASCII collating sequence.
D1
(EFS type).
User-defined data type (requires an EFS program)
D2
(EFS type).
User-defined data type (requires an EFS program)
892

Description
CSF or FS
(signed numeric with optional leading floating sign).

The floating sign format can be represented as follows:
<s>d . . .d
s is an optional sign immediately to the left of the digits d . . .d. If s is a -, the number is treated as negative,
otherwise it is treated as positive. Thus, - must be used for a minus sign, but any other character (for
example, + or blank) can be used for a plus sign. The first non-decimal digit (that is, not 0-9) going from
right to left is treated as the sign and anything to the left of the sign is ignored.
Examples:
Value:
34
+34
00034
-003
--1234
1234
+01234
0
Treated as:
+34
+34
+34
-3
-1234
+1234
+1234
+0
The types of data handled by the CSF or FS format encompass those produced by several different
FORTRAN, PL/I and COBOL formats, such as those shown below (using a width of 4 for purposes of
illustration):
* FORTRAN: I4 ; G4.0 ; SP,I4 ; SP,I4.3 ; S,I4.3
* PL/I: F(4) ; P'S999' ; P'SSS9' ; P'---9'
* COBOL: PIC ++9 ; PIC +999 ; PIC ++++ ; PIC ---9 ; PIC ---- ; PIC ZZZZ
UFF
(unsigned free form numeric).

This format extracts decimal digits (0-9) from right to left anywhere in the field to form a positive number.
Any combination of characters is valid, but characters other than 0-9 are ignored.
Examples:
Value:
$58,272,300.10
$58,272,300.1
$58,272,300
12-31-2004
(402)-125-3721XXX
G1***
52 $ 21 R
000128637.240
+400.52
+400.1
173/821/9072/@3
ABC
SFF
Treated as:
+5827230010
+582723001
+58272300
+12312004
+4021253721
+15221
+128637240
+40052
+4001
+17382190723
+0
(signed free form numeric).

This format extracts decimal digits (0-9) from right to left anywhere in the field to form a positive or
negative number. If - or ) is found anywhere in the field, the number is treated as negative, otherwise it is
treated as positive. Any combination of characters is valid, but characters other than 0-9, - and ) are ignored.
Examples:
Value:
358,272,300.10
358,272,300.1
-358,272,300
(82,316.90)
12-31-2004
G1***
52 $ 21 R
G1*** ) 52 $ 21 R
000128637.240
400.52($400.5)
173/821/9072/@3
X,Y,Z
Treated as:
+35827230010
+3582723001
-358272300
-8231690
-12312004
+15221
-15221
+128637240
-40052
-4005
+17382190723
+0
893

Description
CSL or LS
(signed number, leading separate sign). This format refers to decimal data as punched into cards, and then
assembled into EBCDIC code.
Example: +247 punched in a card becomes
+
2
4
7
Punched numeric data
4E
F2
F4
F7
Hexadecimal
01001110 11110010 11110100 11110111 Binary EBCDIC code
-247 becomes
2
4
7
60
F2
F4
F7
Hexadecimal
01100000 11110010 11110100 11110111 Binary EBCDIC code
Note: A value with '-' as the leading sign character is treated as a negative value. A value with any leading
sign character other than '-' (for example, '+' (plus) or blank) is treated as a positive value.
CST or TS
(signed numeric, trailing separate sign). This has the same representation as the CSL format, except that the
sign indicator is punched after the number.
Example: 247+ punched on the card becomes
F2 F4 F7 4E Hexadecimal
Note: A value with '-' as the trailing sign character is treated as a negative value. A value with any trailing
CLO1 or OL1
(signed numeric, leading overpunch sign). This format again refers to decimal data punched into cards and
then assembled into EBCDIC code. The sign indicator is, however, overpunched with the first decimal digit
of the number.
Example:
+2
C2
11000010
Similarly
D2
CTO or OT
+247 with + overpunched on 2 becomes

4
7
F4
F7
Hexadecimal
11110100 11110111 Binary EBCDIC code
-247 becomes
F4
F7
(signed numeric, trailing overpunch sign). This format has the same representation as for the CLO format,
except that the sign indicator is overpunched on the last decimal digit of the number.
Example: +247 with + overpunched on 7 becomes
F2 F4 C7 hexadecimal
ASL
(signed numeric, ASCII, leading separate sign). Similar to the CSL format but with decimal data assembled
into ASCII code.
Example: +247 punched into card becomes
+
2
4
7
2B
32
34
37
Hexadecimal
0101011 00110010 00110100 00110111 Binary ASCII code
Similarly -247 becomes
2D 32 34 37 hexadecimal
Note: A value with '-' as the leading sign character is treated as a negative value. A value with any leading
AST
(signed numeric, ASCII, trailing separate sign). This gives the same bit representation as the ASL format,
except that the sign is punched after the number.
Example: 247+ becomes
32 34 37 2B hexadecimal
Note: A value with '-' as the trailing sign character is treated as a negative value. A value with any trailing
In the date formats below, unless specified otherwise, yy represents the two-digit
year (00-99), ddd represents the day of the year (001-366), ccyy represents the
four-digit year, mm represents the month (01-12), dd represents the day (01-31), x
represents a decimal digit (0-9), s is hexadecimal 0-F and is ignored, and c
represents the century indicator (c=0 is transformed to 19, c=1 is transformed to 20
and c>1 is transformed to 21).
894

Description
Y2T, Y2W, Y4T, Y4W
(character or zoned decimal date format with special indicators).

The date field can be represented as follows:
3,Y2T:
4,Y2T:
5,Y2T:
6,Y2T:
7,Y4T:
8,Y4T:
3,Y2W:
4,Y2W:
5,Y2W:
6,Y2W:
7,Y4W:
8,Y4W:
C'yyx' or Z'yyx'
C'yyxx' or Z'yyxx'
C'xyy' or Z'xyy'
C'xxyy' or Z'xxyy'
The special indicators are X'00...00' (BI zeros), X'40...40' (blanks), C'0...0' (CH zeros), Z'0...0' (ZD zeros), C'9...9'
(CH nines), Z'9...9' (ZD nines) and X'FF...FF' (BI ones).
Y2U, Y2V, Y2X, Y2Y,
Y4U, Y4V, Y4X, Y4Y
(packed decimal date format with special indicators).

The date field can be represented as follows:
2,Y2U:
3,Y2V:
3,Y2U:
4,Y2V:
4,Y4U:
5,Y4V:
P'yyx' (X'yyxs')
P'yyxx' (X'0yyxxs')
P'yyddd' (X'yyddds')
P'yymmdd' (X'0yymmdds')
P'ccyyddd' (X'ccyyddds')
P'ccyymmdd' (X'0ccyymmdds')
2,Y2X:
3,Y2Y:
3,Y2X:
4,Y2Y:
4,Y4X:
5,Y4Y:
P'xyy' (X'xyys')
P'xxyy' (X'0xxyys')
P'dddyy' (X'dddyys')
P'mmddyy' (X'0mmddyys')
P'dddccyy' (X'dddccyys')
P'mmddccyy' (X'0mmddccyys')
The special indicators are P'0...0' (PD zeros) and P'9...9' (PD nines).
Y2C or Y2Z
(two-digit, two-byte character or zoned-decimal year data). The two-digit year data can be represented as
follows:
xyxy
y is hexadecimal 0-9 and represents a year digit. x is hexadecimal 0-F and is ignored.
Thus, 96 might be represented as hexadecimal F9F6 (character 96) or as hexadecimal F9C6 or 0906 (zoned
decimal 96).
Y2P
(two-digit, two-byte packed-decimal year data). The two-digit year data can be represented as follows:
xyyx
Thus, 96 might be represented as hexadecimal 096F or 896C (packed decimal 96).
Y2D
(two-digit, one-byte decimal year data). The two-digit year data can represented as follows:
yy
y is hexadecimal 0-9 and represents a year digit.
Thus, 96 would be represented as hexadecimal 96 (decimal 96).
895

Description
Y2S
(two-digit, two-byte character or zoned-decimal year data with special indicators).

The two-digit year data can represented as follows:
xyxy
Thus, 96 might be represented as hexadecimal F9F6 (character 96) or as hexadecimal F9C6 or 0906 (zoned
decimal 96).
The special indicators can be represented as follows:
qxzx
qx is hexadecimal 00, 40 or FF. zx is hexadecimal 00-FF (although typically 00, 40 and FF).
Thus, special indicators might be hexadecimal 0000, 0005, 4040, FFFF, FF85 and so on.
Y2B
(two-digit, one-byte binary year data). The binary year data can be represented as follows:
hh
hh is the hexadecimal equivalent of a decimal yy value as follows:
Binary Values
X00-X63
X64-XC7
XC8-XFF
Decimal Values
00-99
100-199
200-255
yy
00-99
00-99
00-55
Thus, 96 might be represented as hexadecimal 60 (decimal 96) or C4 (decimal 196).

DT1
(SMF date interpreted as Z'yyyymmdd'). A 4-byte SMF date value in the form P'cyyddd' (X'0cyydddF') is
converted to a Z'yyyymmdd' value.
DT2
(SMF date interpreted as Z'yyyymm'). A 4-byte SMF date value in the form P'cyyddd' (X'0cyydddF') is
converted to a Z'yyyymm' value.
DT3
(SMF date interpreted as Z'yyyyddd'). A 4-byte SMF date value in the form P'cyyddd' (X'0cyydddF') is
converted to a Z'yyyyddd' value.
DC1
(TOD date interpreted as Z'yyyymmdd'). The 8 bytes of an input clock value, in the basic time-of-day (TOD)
format, is converted to a Z'yyyymmdd' value. The STCKCONV macro is used to do the conversion.
DC2
(TOD date interpreted as Z'yyyymm'). The 8 bytes of an input clock value, in the basic time-of-day (TOD)
format, is converted to a Z'yyyymm' value. The STCKCONV macro is used to do the conversion.
DC3
(TOD date interpreted as Z'yyyyddd'). The 8 bytes of an input clock value, in the basic time-of-day (TOD)
format, is converted to a Z'yyyyddd' value. The STCKCONV macro is used to do the conversion.
DE1
(ETOD date interpreted as Z'yyyymmdd'). The first 8 bytes of an input clock value, in the extended
time-of-day (ETOD) format, is converted to a Z'yyyymmdd' value. The STCKCONV macro is used to do the
conversion.
DE2
(ETOD date interpreted as Z'yyyymm'). The first 8 bytes of an input clock value, in the extended time-of-day
(ETOD) format is converted to a Z'yyyymm' value. The STCKCONV macro is used to do the conversion.
DE3
(ETOD date interpreted as Z'yyyyddd'). The first 8 bytes of an input clock value, in the extended
time-of-day (ETOD) format is converted to a Z'yyyyddd' value. The STCKCONV macro is used to do the
conversion.
In the time formats below, unless specified otherwise, hh represents the hour
(00-23), mm represents the minutes (00-59), ss represents the seconds (00-59), and
xx represents hundredths of a second (00-99).
Description
TM1
(SMF time interpreted as Z'hhmmss'). A 4-byte binary SMF time value in hundredths of a second is
converted to a Z'hhmmss' value.
TM2
(SMF time interpreted as Z'hhmm'). A 4-byte binary SMF time value in hundredths of a second is converted
to a Z'hhmm' value.
896

Description
TM3
(SMF time interpreted as Z'hh'). A 4-byte binary SMF time value in hundredths of a second is converted to a
Z'hh' value.
TM4
(SMF time interpreted as Z'hhmmssxx'). A 4-byte binary SMF time value in hundredths of a second is
converted to a Z'hhmmssxx' value.
TC1
(TOD time interpreted as Z'hhmmss'). The 8 bytes of an input clock value, in the basic time-of-day (TOD)
format, is converted to a Z'hhmmss' value. The STCKCONV macro is used to do the conversion.
TC2
(TOD time interpreted as Z'hhmm'). The 8 bytes of an input clock value, in the basic time-of-day (TOD)
format, is converted to a Z'hhmm' value. The STCKCONV macro is used to do the conversion.
TC3
(TOD time interpreted as Z'hh'). The 8 bytes of an input clock value, in the basic time-of-day (TOD) format,
is converted to a Z'hh' value. The STCKCONV macro is used to do the conversion.
TC4
(TOD time interpreted as Z'hhmmssxx'). The 8 bytes of an input clock value, in the basic time-of-day (TOD)
format, is converted to a Z'hhmmssxx' value. The STCKCONV macro is used to do the conversion.
TE1
(ETOD time interpreted as Z'hhmmss'). The first 8 bytes of an input clock value, in the extended time-of-day
(ETOD) format, is converted to a Z'hhmmss' value. The STCKCONV macro is used to do the conversion.
TE2
(ETOD time interpreted as Z'hhmm'). The first 8 bytes of an input clock value, in the extended time-of-day
(ETOD) format, is converted to a Z'hhmm' value. The STCKCONV macro is used to do the conversion.
TE3
(ETOD time interpreted as Z'hh'). The first 8 bytes of an input clock value, in the extended time-of-day
(ETOD) format, is converted to a Z'hh' value. The STCKCONV macro is used to do the conversion.
TE4
(ETOD time interpreted as Z'hhmmssxx'). The first 8 bytes of an input clock value, in the extended
time-of-day (ETOD) format, is converted to a Z'hhmmssxx' value. The STCKCONV macro is used to do the
conversion.
1
The overpunch sign bit is always 'C' for positive and 'D' for negative.
Where DFSORT formats can be used

The following tables show the statements, operands, and operators allowed with
each of the various data formats.
Table 99. Allowed with Frequently Used Data Types
CH
BI or FI
PD or
ZD
INCLUDE
MERGE
OMIT
SORT
Statement, Operand, or Operator
FS or
CSF
UFF or
SFF
DTn, DCn,
DEn, TMn,
TCn, or TEn
DFSORT statements
SUM
INREC statement operands
IFTHEN WHEN=(logexp)
IFTHEN BEGIN=(logexp)
IFTHEN END=(logexp)
FIELDS
BUILD
OVERLAY
IFTHEN BUILD
IFTHEN OVERLAY
OUTREC statement operands
897

Table 99. Allowed with Frequently Used Data Types (continued)
CH
BI or FI
IFTHEN END=(logexp)
FIELDS
BUILD
OVERLAY
IFTHEN BUILD
IFTHEN OVERLAY
Statement, Operand, or Operator
FS or
CSF
UFF or
SFF
DTn, DCn,
DEn, TMn,
TCn, or TEn
PD or
ZD
OUTFIL statement operands

INCLUDE
OMIT
IFTHEN END=(logexp)
OUTREC
BUILD
OVERLAY
IFTHEN BUILD
IFTHEN OVERLAY
TRAILERx
IFTRAIL TRLID=(logexp)
IFTRAIL TRLUPD
ICETOOL operators
DISPLAY (ON, BREAK)
OCCUR (ON)
RANGE (ON)
SELECT (ON)
SPLICE (ON)
STATS (ON)
UNIQUE (ON)
VERIFY (ON)
Table 100. Allowed with Other Data Types
DFSORT statements
INCLUDE
MERGE
898
AQ
AC
FL
X
X
X
X
LS
TS
OL
OT
or
or
or
or
CSL CST CLO CTO ASL AST
X
X
X
X
X
X
X
X
X
X
X
X
D1
D2
PD0
Y2x
X
X
X
X
Y4x

Table 100. Allowed with Other Data Types (continued)
OMIT
SORT
SUM
INREC statement
operands
IFTHEN
WHEN=(exp)
IFTHEN
BEGIN=(exp)
IFTHEN END=(exp)
FIELDS
BUILD
OVERLAY
IFTHEN BUILD
IFTHEN OVERLAY
OUTREC statement
operands
IFTHEN
WHEN=(exp)
IFTHEN
BEGIN=(exp)
IFTHEN END=(exp)
FIELDS
BUILD
OVERLAY
IFTHEN BUILD
IFTHEN OVERLAY
OUTFIL statement
operands
INCLUDE
OMIT
IFTHEN
WHEN=(exp)
IFTHEN
BEGIN=(exp)
IFTHEN END=(exp)
IFTRAIL
TRLID=(exp)
OUTREC
BUILD
OVERLAY
IFTHEN BUILD
IFTHEN OVERLAY
ICETOOL operators
DISPLAY (ON)
AQ
AC
X
X
X
X
FL
LS
TS
OL
OT
or
or
or
or
CSL CST CLO CTO ASL AST
D1
D2
PD0
Y2x
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Y4x
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
899
DFSORT formats for COBOL data types

Both DFSORT and COBOL support a large number of data types. COBOL
describes these data types in one way, and DFSORT describes them in another
way. If you SORT or MERGE with COBOL, the compiler automatically generates a
SORT or MERGE control statement for you with the correct DFSORT descriptions
for the COBOL fields you specify. But to take full advantage of DFSORT, you will
often want to describe your fields in your own DFSORT control statements (for
example, SORT, MERGE, INCLUDE, OMIT, INREC, OUTREC, OUTFIL, SUM)
either outside of COBOL or in a DFSPARM data set used with COBOL. The table
below will show you what DFSORT length and format to use for the various
commonly used COBOL data types.
For example, say you want to separate out records in a very large file into two
data sets based on the values in a PIC S9(4) COMP field starting in position 21. In
the first data set, you want records with values in the field that are greater than or
equal to +5000. In the second data set, you want records with values in the field
that are less than -1000. You could use the table below to determine that a PIC
S9(4) COMP field is equivalent to a DFSORT field with a length of 2 and a format
of FI, allowing you to code your DFSORT statements as follows:
OPTION COPY
OUTFIL FNAMES=OUT1,INCLUDE=(21,2,FI,GE,+5000)
OUTFIL FNAMES=OUT2,INCLUDE=(21,2,FI,LT,-1000)
Table 101. Equivalent DFSORT formats for various COBOL data types
DFSORT Length
DFSORT Format
PIC X(n) USAGE DISPLAY
CH
GROUP DATA ITEMS with n bytes
CH
PIC 9(n) DISPLAY
ZD
PIC S9(n) DISPLAY <TRAILING>
ZD
PIC S9(n) DISPLAY LEADING
CLO
PIC S9(n) DISPLAY SEPARATE <TRAILING>
n+1
CST
PIC S9(n) DISPLAY LEADING SEPARATE
n+1
CSL or FS
n = 1 to 4
BI
n = 5 to 9
BI
n >= 10
BI
FI
PIC 9(n) COMP|BINARY|COMP-4|COMP-5
PIC S9(n) COMP|BINARY|COMP-4|COMP-5

n = 1 to 4
n = 5 to 9
FI
n >= 10
FI
PIC 9(n) COMP-3|PACKED-DECIMAL
(n/2)+1
PD
PIC S9(n) COMP-3|PACKED-DECIMAL
(n/2)+1
PD
COMP-1
FL
COMP-2
FL
Note:
1. PIC 9(x)V9(y) can be treated like PIC 9(n) where n=x+y. (COBOL does NOT
store the decimal point internally.)
2. PIC S9(x)V9(y) can be treated like PIC S9(n) where n=x+y. (COBOL does NOT
store the decimal point internally.)
900
Appendix D. EBCDIC and ASCII collating sequences

EBCDIC
Table 102 shows the collating sequence for EBCDIC character and unsigned
decimal data. The collating sequence ranges from low (00000000) to high (11111111).
The bit configurations which do not correspond to symbols (that is, 0 through 73,
81 through 89, and so forth) are not shown. Some of these correspond to control
commands for the printer and other devices.
ALTSEQ, CHALT, and LOCALE can be used to select alternate collating sequences
for character data.
Packed decimal, zoned decimal, fixed-point, and normalized floating-point data are
collated algebraically, that is, each quantity is interpreted as having a sign.
Table 102. EBCDIC Collating Sequence
Bit Configuration
0
.
.
64
.
.
74
75
76
77
78
79
80
.
.
90
91
92
93
94
95
96
97
107
108
109
110
111
.
.
122
123
124
125
Symbol
Meaning
01000000
SP
Space
01001010
01001011
01001100
01001101
01001110
01001111
01010000
.
<
(
+
I
&
Cent sign
Period, decimal point
Less than sign
Left parenthesis
Plus sign
Vertical bar, Logical OR
Ampersand
01011010
01011011
01011100
01011101
01011110
01011111
01100000
01100001
01101011
01101100
01101101
01101110
01101111
!
$
*
)
;
/
,
%
_
>
?
Exclamation point
Dollar sign
Asterisk
Right parenthesis
Semicolon
Logical not
Minus, hyphen
Slash
Comma
Percent sign
Underscore
Greater than sign
Question mark
01111010
01111011
01111100
01111101
:
#
@
'
Colon
Number sign
Commercial At
Apostrophe, prime
00000000
901
Table 102. EBCDIC Collating Sequence (continued)
126
127
.
.
129
130
131
132
133
134
135
136
137
.
.
145
146
147
148
149
150
151
152
153
.
.
162
163
164
165
166
167
168
169
193
194
195
196
197
198
199
200
201
.
.
209
210
211
212
213
214
215
216
902
Bit Configuration
Symbol
Meaning
01111110
01111111
=
"
Equal sign
Quotation marks
10000001
10000010
10000011
10000100
10000101
10000110
10000111
10001000
10001001
a
b
c
d
e
f
g
h
i
10010001
10010010
10010011
10010100
10010101
10010110
10010111
10011000
10011001
j
k
l
m
n
0
p
q
r
10100010
10100011
10100100
10100101
10100110
10100111
10101000
10101001
11000001
11000010
11000011
11000100
11000101
11000110
11000111
11001000
11001001
s
t
u
v
w
x
y
z
A
B
C
D
E
F
G
H
I
11010001
11010010
11010011
11010100
11010101
11010110
11010111
11011000
J
K
L
M
N
O
P
Q
Table 102. EBCDIC Collating Sequence (continued)
217
.
.
226
227
228
229
230
231
232
233
.
.
240
241
242
243
244
245
246
247
248
249
.
.
255
Bit Configuration
Symbol
11011001
11100010
11100011
11100100
11100101
11100010
11100111
11101000
11101001
S
T
U
V
W
X
Y
Z
11110000
11110001
11110010
11110011
11110100
11110101
11110110
11110111
11111000
11111001
0
1
2
3
4
5
6
7
8
9
Meaning
11111111
ASCII
Table 103 shows the collating sequence for ASCII, character, and unsigned decimal
data. The collating sequence ranges from low (00000000) to high (01111111). Bit
configurations that do not correspond to symbols are not shown.
Packed decimal, zoned decimal, fixed-point normalized floating-point data, and the
signed numeric data formats are collated algebraically; that is, each quantity is
interpreted as having a sign.
Table 103. ASCII Collating Sequence
0
.
.
32
33
34
35
36
37
38
39
.
.
Bit Configuration
Symbol
00000000
Null
00100000
00100001
00100010
00100011
00100100
00100101
00100110
00100111
SP
!
"
#
$
%
&
'
Meaning
Space
Exclamation point
Quotation mark
Number sign
Dollar sign
Percent
Ampersand
Apostrophe, prime
903
Table 103. ASCII Collating Sequence (continued)
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
904
Bit Configuration
Symbol
Meaning
00101000
00101001
00101010
00101011
00101100
00101101
00101110
00101111
00110000
00110001
00110010
00110011
00110100
00110101
001101100
00110111
00111000
00111001
00111010
00111011
00111100
00111101
00111110
00111111
01000000
01000001
01000010
01000011
01000100
01000101
01000110
01000111
01001000
01001001
01001010
01001011
01001100
01001101
01001110
01001111
01010000
01010001
01010010
01010011
01010100
01010101
01010110
01010111
01011000
01011001
01011010
01011011
01011100
(
)
*
+
,
.
/
0
1
2
3
4
5
6
7
8
9
:
;
<
=
>
?
@
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z
[
\
Opening parenthesis
Closing parenthesis
Asterisk
Plus
Comma
Hyphen, minus
Period, decimal point
Slash
Colon
Semicolon
Less than
Equals
Greater than
Question mark
Commercial At
Opening bracket
Reverse slash
Table 103. ASCII Collating Sequence (continued)
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
Bit Configuration
Symbol
Meaning
01011101
01011110
01011111
01100000
01100001
01100010
01100011
01100100
01100101
01100110
01100111
01101000
01101001
01101010
01101011
01101100
01101101
01101110
01101111
01110000
01110001
01110010
01110011
01110100
01110101
01110110
01110111
01111000
01111001
01111010
01111011
01111100
01111101
01111110
]
^
_
`
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
{
I
}
~
Closing bracket
Circumflex, Logical NOT
Underscore
Grave Accent
Opening Brace
Vertical Line
Closing Brace
Tilde
905
906
Appendix E. DFSORT abend processing

This appendix explains how DFSORT processes an abend. It is intended to help
you get the dump you need to diagnose the error causing the abend.
All abend dumps produced by DFSORT are system abend dumps that can be
processed by standard dump analysis programs. A dump will be generated if you
have included a SYSUDUMP, SYSABEND, or SYSMDUMP DD statement in your
application. The actual output of the system dump depends on the system
parameters specified in the IEADMP00, IEAABD00 or IEADMR00 members of
SYS1.PARMLIB by your installation.
At the beginning of each run, DFSORT establishes an ESTAE recovery routine to
trap system or user abends for Blockset and Peer/Vale applications. You can delete
the routine by specifying installation option ESTAE=NO, or by specifying
NOESTAE on the DEBUG control statement. We recommend that you always run
with ESTAE in effect so that in the event of an abend the following benefits are
available:
v In general, you get dumps closer to the time of the abend.
v You get additional information useful in diagnosing the problem causing the
abend.
v If you have activated SMF, an ICETEXIT routine, or an EFS program, DFSORT
attempts to continue processing. That is, an SMF record is created, the
termination exit is called, or Major Calls 4 and 5 are made to the EFS program
before the application terminates processing. Of course, if one of these functions
caused the abend, that function will not complete successfully.
At the end of its recovery routine, DFSORT always returns control to the system to
allow termination to continue. The system will then invoke the next higher level
ESTAE recovery routine.
Checkpoint/restart
Checkpoint/Restart is a facility of the operating system that allows information
about an application to be recorded so that same application can be restarted after
abnormal termination or after some portion of the application has been completed.
Restart can take place immediately or be deferred until the application is
resubmitted.
DFSORT takes checkpoints when requested during a sort that uses the Peerage or
Vale techniques.
To have DFSORT record checkpoints you must code a SORTCKPT DD statement
and ensure the Peerage or Vale technique is selected. See SORTCKPT DD
statement on page 75 and OPTION control statement on page 173 for more
information on the SORTCKPT and CKPT options, respectively.
In general, no checkpoints are taken if the following conditions exist:
v No work data set is specified.
v The application is a copy or merge.
v Blockset is selected.
907
Abend Processing
Note:
1.
2. No ANSI Standard Label tape files can be open during Checkpoint/Restart
processing.
3. Do not specify CHKPT=EOV on any DFSORT DD statement.
For more information on the Checkpoint/Restart facility, see z/OS DFSMSdfp
Checkpoint/Restart.
DFSORT abend categories

There are two categories of abends for DFSORT: unexpected abends and user
abends issued by DFSORT.
v Unexpected abends
These are system abends or user abends not issued by DFSORT. The abend code
in these cases is the system abend code or the user abend code. See z/OS
DFSORT Messages, Codes and Diagnosis Guide for information about detecting
common user errors and reporting DFSORT program failures.
v User abends issued by DFSORT
DFSORT will issue user abends under the following circumstances:
The ABEND or ABSTP option is in effect and DFSORT encounters an error

that prevents completion of the run.
DFSORT detects an error in its internal logic.
See z/OS DFSORT Messages, Codes and Diagnosis Guide for complete information
about user abends issued by DFSORT.
Abend recovery processing for unexpected abends

DFSORT normally has an ESTAE recovery routine established to trap system or
user exit routine abends for Blockset and Peer/Vale applications. If an abend
occurs, the system will pass control to this routine. The DFSORT ESTAE recovery
routine functions are as follows:
v Abend dump
The recovery routine will first have the system issue an abend dump to capture
the environment at the time the error occurred.
v Termination functions
DFSORT tries to accomplish the following tasks when they are specified,
whether the program terminates normally or abnormally.
Calls 4 and 5 to an EFS program
Create the SMF record
Call the ICETEXIT routine
The DFSORT recovery routine runs any of the previous functions specified if
they have not already been run at the time of the abend.
v Abend information message
For unexpected system or user exit routine abends, the DFSORT recovery
routine issues message ICE185A giving information about when the abend
occurred. The description of this message is in z/OS DFSORT Messages, Codes and
Diagnosis Guide.
v Snap dumps
908
Abend Processing
The DFSORT recovery routine provides a snap dump of the system diagnostic
work area (SDWA). The snap dumps are written to a dynamically allocated data
set whether or not a SYSUDUMP (or SYSABEND or SYSMDUMP) DD statement
is included in the application.
v Copy system diagnostic work area
If an invoking program passes the address of an SDWA area in the 24-bit or
extended parameter list, DFSORT will copy the first 104 or 112 bytes of the
system diagnostic work area into the user SDWA area. See Chapter 6, Invoking
DFSORT from a program, on page 541 for more information.
v Continuation of an application after successful SORTOUT output
If an unexpected abend occurs after the sort, merge, or copy application writes
the SORTOUT data set successfully, DFSORT issues message ICE186A and
completes its normal cleanup and termination functions. The SORTOUT data set
written by DFSORT is closed. The run is successful except for the function
causing the abend. Message ICE186A says that the SORTOUT data set is usable
even though the run has abended. You can then decide to use the SORTOUT
data set or rerun the application.
v DFSORT returns control to the system at the end of its abend recovery
processing so that recovery routines can be invoked.
The DFSORT abend recovery routine functions described previously may not be
performed after an abend if NOESTAE is in effect. The DFSORT ESTAE recovery
routine is always established at the beginning of a run. It is deleted early in
DFSORT processing if NOESTAE is in effect.
Processing of error abends with A-type messages

When DFSORT encounters a critical error, it issues an A-type message and
terminates. You can specify that DFSORT is to terminate the application with an
abend through the ABEND or ABSTP options.
If abend termination is in effect and DFSORT encounters a critical error, DFSORT
first causes an abend dump to capture the environment at the time of the error.
Then, it issues the A-type message. It also runs the termination functions described
earlier before terminating with an abend. The abend code will be the error message
number, or a number between 1-99, as determined during installation with the
ABCODE installation option.
With NOESTAE and ABEND in effect, the abend dump is produced after the
A-type message is printed and other termination functions are run. As a result, the
dump produced might not reflect the conditions at the time of the error. It may not
include the module that encountered the error.
With NOESTAE and ABSTP in effect, the correct module will be dumped but the
A-type message will not be issued.
CTRx abend processing

The CTRx option can be used to diagnose a problem by requesting that DFSORT
abend during record input or output processing. See the DEBUG control statement
in Chapter 3, Using DFSORT program control statements, on page 81. DFSORT
will cause an 0C1 abend when the CTRx count is satisfied. The DFSORT ESTAE
recovery routine will process the abend in the same way as it does for an
unexpected abend described earlier.
Appendix E. DFSORT abend processing
909
Abend Processing
The DFSORT ESTAE recovery routine will return control to the system, which will
pass control to any ESTAE recovery routines established by invoking programs.
As described earlier, the DFSORT ESTAE recovery routine will save the first 104 or
112 bytes of the system diagnostic work area in the invoking program's SDWA area
if the address of the area is passed to DFSORT.
Since PL/I normally has an ESPIE in effect to intercept program checks (0Cx abend
codes), the DFSORT ESTAE recovery routine is not entered after these errors unless
you have specified NOSPIE. DFSORT abend recovery processing will occur for all
other types of abends.
Invocations from COBOL programs or use of COBOL exits can result in more than
one abend dump.
910
Appendix F. Accessibility
Accessible publications for this product are offered through IBM Knowledge
Center (http://www.ibm.com/support/knowledgecenter/SSLTBW/welcome).
If you experience difficulty with the accessibility of any z/OS information, send a
detailed message to the "Contact us" web page for z/OS (http://www.ibm.com/
systems/z/os/zos/webqs.html) or use the following mailing address.
IBM Corporation
Attention: MHVRCFS Reader Comments
Department H6MA, Building 707
2455 South Road
Poughkeepsie, NY 12601-5400
United States
Accessibility features
Accessibility features help users who have physical disabilities such as restricted
mobility or limited vision use software products successfully. The accessibility
features in z/OS can help users do the following tasks:
v Run assistive technology such as screen readers and screen magnifier software.
v Operate specific or equivalent features by using the keyboard.
v Customize display attributes such as color, contrast, and font size.
Consult assistive technologies

Assistive technology products such as screen readers function with the user
interfaces found in z/OS. Consult the product information for the specific assistive
technology product that is used to access z/OS interfaces.

You can access z/OS user interfaces with TSO/E or ISPF. The following
information describes how to use TSO/E and ISPF, including the use of keyboard
shortcuts and function keys (PF keys). Each guide includes the default settings for
the PF keys.
v z/OS TSO/E Primer
v z/OS TSO/E User's Guide
v z/OS V2R2 ISPF User's Guide Vol I
Dotted decimal syntax diagrams

Syntax diagrams are provided in dotted decimal format for users who access IBM
Knowledge Center with a screen reader. In dotted decimal format, each syntax
element is written on a separate line. If two or more syntax elements are always
present together (or always absent together), they can appear on the same line
because they are considered a single compound syntax element.
Each line starts with a dotted decimal number; for example, 3 or 3.1 or 3.1.1. To
hear these numbers correctly, make sure that the screen reader is set to read out
911
punctuation. All the syntax elements that have the same dotted decimal number
(for example, all the syntax elements that have the number 3.1) are mutually
exclusive alternatives. If you hear the lines 3.1 USERID and 3.1 SYSTEMID, your
syntax can include either USERID or SYSTEMID, but not both.
The dotted decimal numbering level denotes the level of nesting. For example, if a
syntax element with dotted decimal number 3 is followed by a series of syntax
elements with dotted decimal number 3.1, all the syntax elements numbered 3.1
are subordinate to the syntax element numbered 3.
Certain words and symbols are used next to the dotted decimal numbers to add
information about the syntax elements. Occasionally, these words and symbols
might occur at the beginning of the element itself. For ease of identification, if the
word or symbol is a part of the syntax element, it is preceded by the backslash (\)
character. The * symbol is placed next to a dotted decimal number to indicate that
the syntax element repeats. For example, syntax element *FILE with dotted decimal
number 3 is given the format 3 \* FILE. Format 3* FILE indicates that syntax
element FILE repeats. Format 3* \* FILE indicates that syntax element * FILE
repeats.
Characters such as commas, which are used to separate a string of syntax
elements, are shown in the syntax just before the items they separate. These
characters can appear on the same line as each item, or on a separate line with the
same dotted decimal number as the relevant items. The line can also show another
symbol to provide information about the syntax elements. For example, the lines
5.1*, 5.1 LASTRUN, and 5.1 DELETE mean that if you use more than one of the
LASTRUN and DELETE syntax elements, the elements must be separated by a comma.
If no separator is given, assume that you use a blank to separate each syntax
element.
If a syntax element is preceded by the % symbol, it indicates a reference that is
defined elsewhere. The string that follows the % symbol is the name of a syntax
fragment rather than a literal. For example, the line 2.1 %OP1 means that you must
refer to separate syntax fragment OP1.
The following symbols are used next to the dotted decimal numbers.
? indicates an optional syntax element
The question mark (?) symbol indicates an optional syntax element. A dotted
decimal number followed by the question mark symbol (?) indicates that all
the syntax elements with a corresponding dotted decimal number, and any
subordinate syntax elements, are optional. If there is only one syntax element
with a dotted decimal number, the ? symbol is displayed on the same line as
the syntax element, (for example 5? NOTIFY). If there is more than one syntax
element with a dotted decimal number, the ? symbol is displayed on a line by
itself, followed by the syntax elements that are optional. For example, if you
hear the lines 5 ?, 5 NOTIFY, and 5 UPDATE, you know that the syntax elements
NOTIFY and UPDATE are optional. That is, you can choose one or none of them.
The ? symbol is equivalent to a bypass line in a railroad diagram.
! indicates a default syntax element
The exclamation mark (!) symbol indicates a default syntax element. A dotted
decimal number followed by the ! symbol and a syntax element indicate that
the syntax element is the default option for all syntax elements that share the
same dotted decimal number. Only one of the syntax elements that share the
dotted decimal number can specify the ! symbol. For example, if you hear the
lines 2? FILE, 2.1! (KEEP), and 2.1 (DELETE), you know that (KEEP) is the
912
default option for the FILE keyword. In the example, if you include the FILE
keyword, but do not specify an option, the default option KEEP is applied. A
default option also applies to the next higher dotted decimal number. In this
example, if the FILE keyword is omitted, the default FILE(KEEP) is used.
However, if you hear the lines 2? FILE, 2.1, 2.1.1! (KEEP), and 2.1.1
(DELETE), the default option KEEP applies only to the next higher dotted
decimal number, 2.1 (which does not have an associated keyword), and does
not apply to 2? FILE. Nothing is used if the keyword FILE is omitted.
* indicates an optional syntax element that is repeatable
The asterisk or glyph (*) symbol indicates a syntax element that can be
repeated zero or more times. A dotted decimal number followed by the *
symbol indicates that this syntax element can be used zero or more times; that
is, it is optional and can be repeated. For example, if you hear the line 5.1*
data area, you know that you can include one data area, more than one data
area, or no data area. If you hear the lines 3* , 3 HOST, 3 STATE, you know
that you can include HOST, STATE, both together, or nothing.
Notes:
1. If a dotted decimal number has an asterisk (*) next to it and there is only
one item with that dotted decimal number, you can repeat that same item
more than once.
2. If a dotted decimal number has an asterisk next to it and several items
have that dotted decimal number, you can use more than one item from the
list, but you cannot use the items more than once each. In the previous
example, you can write HOST STATE, but you cannot write HOST HOST.
3. The * symbol is equivalent to a loopback line in a railroad syntax diagram.
+ indicates a syntax element that must be included
The plus (+) symbol indicates a syntax element that must be included at least
once. A dotted decimal number followed by the + symbol indicates that the
syntax element must be included one or more times. That is, it must be
included at least once and can be repeated. For example, if you hear the line
6.1+ data area, you must include at least one data area. If you hear the lines
2+, 2 HOST, and 2 STATE, you know that you must include HOST, STATE, or
both. Similar to the * symbol, the + symbol can repeat a particular item if it is
the only item with that dotted decimal number. The + symbol, like the *
symbol, is equivalent to a loopback line in a railroad syntax diagram.
Using assistive technologies

Assistive technology products, such as screen readers, function with the user
interfaces found in z/OS. Consult the assistive technology documentation for
specific information when using such products to access z/OS interfaces.

Users can access z/OS user interfaces using TSO/E or ISPF. Refer to z/OS TSO/E
Primer, z/OS TSO/E User's Guide, and z/OS V2R2 ISPF User's Guide Vol I for
information about accessing TSO/E and ISPF interfaces. These guides describe
how to use TSO/E and ISPF, including the use of keyboard shortcuts or function
keys (PF keys). Each guide includes the default settings for the PF keys and
explains how to modify their functions.
Appendix F. Accessibility
913
z/OS information
z/OS information is accessible using screen readers with the Library Server
versions of z/OS books in the Internet library at:
http://www.ibm.com/systems/z/os/zos/bkserv/
914
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in
other countries. Consult your local IBM representative for information on the
products and services currently available in your area. Any reference to an IBM
product, program, or service is not intended to state or imply that only that IBM
product, program, or service may be used. Any functionally equivalent product,
program, or service that does not infringe any IBM intellectual property right may
be used instead. However, it is the user's responsibility to evaluate and verify the
operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter
described in this document. The furnishing of this document does not give you
any license to these patents. You can send license inquiries, in writing, to:
IBM Director of Licensing
IBM Corporation
North Castle Drive
Armonk, NY 10504-1785
U.S.A.
For license inquiries regarding double-byte (DBCS) information, contact the IBM
Intellectual Property Department in your country or send inquiries, in writing, to:
IBM World Trade Asia Corporation Licensing
2-31 Roppongi 3-chome, Minato-ku
Tokyo 106, Japan
The following paragraph does not apply to the United Kingdom or any other
country where such provisions are inconsistent with local law:
INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS
PUBLICATIONAS IS WITHOUT WARRANTY OF ANY KIND, EITHER
EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY OR FITNESS
FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of express or
implied warranties in certain transactions, therefore, this statement may not apply
to you.
This information could include technical inaccuracies or typographical errors.
Changes are periodically made to the information herein; these changes will be
incorporated in new editions of the publication. IBM may make improvements
and/or changes in the product(s) and/or the program(s) described in this
publication at any time without notice.
IBM may use or distribute any of the information you supply in any way it
believes appropriate without incurring any obligation to you.
Licensees of this program who wish to have information about it for the purpose
of enabling: (i) the exchange of information between independently created
programs and other programs (including this one) and (ii) the mutual use of the
information which has been exchanged, should contact:
915
IBM Corporation
Mail Station P300
2455 South Road
Poughkeepsie, NY 12601-5400
USA
Such information may be available, subject to appropriate terms and conditions,
including in some cases, payment of a fee.
The licensed program described in this information and all licensed material
available for it are provided by IBM under terms of the IBM Customer Agreement,
IBM International Program License Agreement, or any equivalent agreement
between us.
Programming interface information

This publication primarily documents information that is NOT intended to be used
as Programming Interfaces of DFSORT.
This publication also documents intended Programming Interfaces that allow the
customer to write programs to obtain the services of DFSORT. This information is
identified where it occurs, either by an introductory statement to a chapter or
section or by the following marking:
Trademarks
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of
International Business Machines Corporation in the United States, other countries,
or both. If these and other IBM trademarked terms are marked on their first
occurrence in this information with a trademark symbol ( or ), these symbols
indicate U.S. registered or common law trademarks owned by IBM at the time this
information was published. Such trademarks may also be registered or common
law trademarks in other countries. A current list of IBM trademarks is available on
the Web at http://www.ibm.com/legal/copytrade.shtml.
Linux is a trademark of Linus Torvalds in the United States, other countries, or
both.
UNIX is a registered trademark of The Open Group in the United States and other
countries.
Microsoft, Windows, Windows NT, and the Windows logo are trademarks of
Microsoft Corporation in the United States, other countries, or both.
Other company, product, and service names may be trademarks or service marks
of others.
916
Index
Special characters
/OT (trailing overpunch sign) format
description 894
% parameter
of PARSE operand
on OUTFIL statement 241
%nn parameter
of PARSE operand
Numerics
24-bit parameter list
examples 560
format 544
64-bit parameter list
format 554
A
ABCODE
ABEND Code 35
installation option 18
abend
categories 908
checkpoint/restart 907
critical 909
CTRx processing 909
processing 907, 910
processing for unexpected
abends 908, 909
recovery 908, 909
ESTAE 907
ABEND
DEBUG control statement option
EXEC PARM option 35
ABSPOS=p parameter
of PARSE operand
ABSTP
processing 908
AC (ASCII character) format
description 892
where allowed 897
ACCEPT parameter
option 227, 235
accessibility 911
contact IBM 911
features 911
screen readers 915
action codes 779
adding fields and constants
INREC 136
OUTFIL 282
OUTREC 412
adding records 492
E15 user exit 499, 523
E35 user exit 529
91
92
ADDPOS=x parameter
of PARSE operand
addressing
EFS program 768
EFS program user exit routine 792
user exits 494
ALIAS statement 496
aliases
DFSORT 29
OPTION statement options 219
PARM options 60
alignment field 130, 407
allocating storage
intermediate storage 810
main storage 807
temporary work space 810
allocating temporary work space
efficiently 810, 811
altering records 492
See also reformatting records 125
ALTSEQ
ALTSEQ control statement 90
examples 89
function 83
TABLE Option 88
using 88, 90
ALTSEQ Statement Examples 89, 90
AMODE 494, 498
AQ (character EBCDIC, with alternate
collating sequence, unsigned) format
description 892
where allowed 897
ARESALL
EXEC PARM option 35
option 176
releasing main storage 809
using RESERVEX instead of
ARESALL 35
ARESINV
option 176
arithmetic
INREC 136
OUTFIL 282
OUTREC 412
ASL (ASCII leading sign) format
description 894
where allowed 897
Assembler user exit routines
input phase user exits 498, 509
output phase user exits 510, 517
assistive technologies 911, 913
AST (ASCII trailing sign) format
description 894
where allowed 897
ATTACH
description 541
writing macro instructions 559
Attention
"do not return" return code 495
Explanation field length 551
invalid syntax for incorrectly placed
blanks 87
performance degradation 36
RDW 264, 411
reformatted input record 135
unpredictable results with order of
data sets when using BSAM
processing with concatenated
SORTIN input 92
using the SIZE or FILSZ
parameter 190
when USING(xxxx) is not
specified 718
where USING(xxxx) is not
specified 646
AVGRLEN
EXEC PARM option 35
option 176
B
BI (binary) format
description 892
where allowed 897
binding
performance 806
bit comparison tests 112
bit operators 112
BLDINDEX 817
BLKCCH1 parameter
option 369
BLKCCH2 parameter
option 370
BLKCCT1 parameter
option 370
block
minimum length 14
blocking records 800
Blockset
DFSORT 26
BSAM
E18 user exit 504
E19 user exit 507
EXEC PARM option 36
BUILD parameter
INREC statement 129
OUTFIL statement 228, 252
OUTREC statement 406
92
917
C
cache fast write
specifying use of with OPTION
control statement 92
using to improve performance 803
cataloged procedures
SORT 30
SORT cataloged procedure 30
SORTD 31
SORTD cataloged procedure 31
cataloged procedures procedures,
catalogued
defined 30
specifying 30
century window 218, 385, 446, 873
CFW
using on OPTION control
statement 92
using to improve performance 803
CH (character) format
description 891
where allowed 897
CHALT
option 177
changing records 492
E35 512
E35 user exit 529
See also reformatting records 125
character constants 104, 131, 258, 340,
346, 408
character strings
for current date 105
for future date 106
for past date 106
CHECK
option 177
checkpoint/restart (CHKPT)
restrictions 908
using 907
CINV
EXEC PARM option 36
option 178
CKPT
efficiency 806
option 178
SORT control statement option 448
CLIST examples 821
CLO (leading overpunch sign) format
description 894
where allowed 897
closing data sets
E17 user exit 503
E37 user exit 517
housekeeping 776
with an EFS program 772, 776
with user exits 493
COBEXIT
EXEC PARM option 36
918
COBEXIT (continued)
option 179
COBOL
input phase user exits 523
output phase user exit 529
overview 521
requirements for copy processing 522
storage requirements 522
user exit routine requirements 521
user exit routines 521, 523, 529
COBOL E15 user exit
altering records 535
changing records for Sort 523
passing records for Sort 523
COBOL E35 user exit
changing records 529
inserting records 536
CODE
ALTSEQ control statement option 88
coding control statements 83
coding restrictions 88
collating sequence 88
altering with user exit 492
alternate 6
ASCII 6
defined 6
EBCDIC 6
modifying 6
combining data sets
See merging records 162
comment statement 87
comparison operator 99, 118
comparisons
OMIT control statement 170
COND
option 98
OMIT control statement option 171
considerations
data set 12
key-sequenced data set (KSDS) 16
QSAM data set 15
record descriptor word (RDW) 16
VSAM data set 16
constants
bit string 116, 125
character string 104, 131, 258, 340,
346, 408
for future date 106
for past date 106
date string 118
decimal number 102
for future date 103
for past date 104
for future date 259
for past date 260
hexadecimal string 107, 131, 258, 340,
346, 408
contact
z/OS 911
continuation column 85
continuation lines 85
continuing control statements 85

control field
defined 5, 6
deleting
with INREC control
statement 125
with OUTREC control
statement 402
describing on MERGE control
statement 163
describing on SORT control
statement 443
efficient design 800
equal 6
format 445
length 445
modifying with E61 user exit 508
modifying with user exit 492
overview 5, 7
reordering
with INREC control
statement 125
with OUTREC control
statement 402
Control Field Formats and Lengths
Table 445
control statement
coding 83
coding restrictions 88
comment statement 87
continuation column 85
EFS coding rules 779, 782
EFS interface request list 779
EFS string 779
examining, altering, or ignoring 774
format 83
functions 81, 83
label field 84
operand field 84
operation field 84
overview 81
preparing image 543
printing with an EFS program 776
remark field 85
request list 779
string returned by the EFS
program 781
string sent to the EFS program 779
summary 81, 83
control statement program control
statements
using with EXEC statement 32
control statements
using other IBM programs 88
using SET and PROC symbols 29
control word 443
CONVERT parameter
option 228, 309
COPY
option 179
copy examples 577, 846, 847
COPY operator (ICETOOL) 575
copy restrictions 561, 562
copying
data set requirements 12
copying (continued)
defined 1
overview 12
copying records
with MERGE control statement 163
COUNT operator (ICETOOL) 579
critical errors 909
CSF (floating sign) format
description 893
where allowed 897
CSL (leading sign) format
description 894
where allowed 897
CST (trailing sign) format
description 894
where allowed 897
CTO (trailing overpunch sign) format
description 894
where allowed 897
CTRx
abend processing 909
DEBUG control statement option 92
cultural environment
See LOCALE 7
current date
character string for 105
decimal number constant for 103
cylinders 801, 859
D
D1 (EFS type) format
description 892
where allowed 897
D2 (EFS type) format
description 892
where allowed 897
data formats
descriptions 891
where allowed 897
data set 12
closing 493
closing with user exit routines 503,
517
defining 12
handling input with user exit
routines 517
handling output with user exit
routines 517
input 11
shared tape unit 63
key-sequenced, considerations 16
message data set 25
notes and limitations 13, 15
opening with user exit routines 492,
499, 510
output 12
shared tape unit 63
page=end.considerations 15
QSAM considerations 15
requirements 12
valid types 12
VSAM considerations 16
data set data management rules

managing system data, rules rules, for
managing system data
system data management rules 13
data space
definition 180
specifying with EXEC PARM 37
specifying with OPTION control
statement 180
data types 13
DATASORT operator (ICETOOL) 585
dataspace sorting
advantages 812
considerations 812
definition 812
date comparisons 117
date constant 105, 118, 132, 408
date constants 258
date formats
descriptions 896
where allowed 897
DBCS ordering 767
DCn (TOD date) format
description 896
where allowed 897
DD statements
overview 61
program DD statements 65
summary 27
system DD statements 63
using 61, 78
ddnames
duplicate 63
DEBUG control statement
example 91, 95
function 83
special handling 781
using 91, 95
DEBUG Statement Examples 95
debugging jobs 91
decimal number constants 102
for future date 103
for past date 104
defaults
installation 17
listing with ICETOOL 17
DEFAULTS operator (ICETOOL) 590
definitions
cataloged procedures 30
collating sequence 6
control field 5
copying 1
DD statements 27
direct invocation 5
EXEC statement 29
installation options 18, 24
JOB statement 29
key 5
merging 1
program invocation 5
sorting 1
deleting control fields
with INREC 125
with OUTREC control statement 407
deleting records 492
deleting records (continued)

E35 user exit 529
with INCLUDE control statement 96,
169
with OMIT control statement 169
DEn (ETOD date) format
description 896
where allowed 897
designing applications to maximize
performance 799, 807
designing new applications 800
determining action when intermediate
storage is insufficient 493
devices, improving elapsed time
with 802
DFSORT 17
calls to your EFS program 769
dynamic invocation 541
exit routines 487
improving efficiency 799
invoking 4
job control statements 27, 78
logic examples for input/user
exit/output 490
messages 25
override of options 863
overview 1
processing order 7
processing OUTFIL operands 223
program control statements 81, 455
program phases 488, 768
terminating with user exit 493
DFSORT home page 4
DFSORT phases
definition 768
initialization 770, 796
input 772
termination 772, 798
DFSPARM data set 864
DFSPARM DD statement
defined 28
function 66
using 76, 78
DFSPARM statement
PARM options 32
alias PARM options 60
diagnosis
EFS program 795
diagnostic messages 25
DIAGSIM
direct invocation
definition 5
DFSORT processing 799
using JCL 866
disk
capacity considerations 859, 860
efficiency 801, 810
exceeding capacity 860
disk storage devices
See disk 801
disk work storage devices 810
DISPLAY operator (ICETOOL) 594
dividing fields and constants
INREC 136
OUTFIL 282
OUTREC 412
Index
919
Double Byte Character Set (DBCS)

ordering
See DBCS ordering 13
Double Byte Character Set Ordering
Support Program
See DBCS ordering 767
DSA
EXEC PARM option 37
option 179
DSA (Dynamic Storage Adjustment)
enhancing performance 803
limit 867, 875, 883
DSPSIZE
EXEC PARM option 37
option 180
DTn (SMF date) format
description 896
where allowed 897
duplicate ddnames 63
duplicate records
OCCUR operator (ICETOOL) 647
SELECT operator (ICETOOL) 668
SUM control statement 451
DYNALLOC
EXEC PARM option 38
option 181
DYNALLOC=OFF
EXEC PARM option 38
option 182
DYNALOC
dynamic link-editing
See link-editing 498
Dynamic Storage Adjustment (DSA)
limit 867, 875, 883
dynamically-invoked DFSORT
with the 24-bit parameter list 882,
890
with the extended parameter list 874,
882
DYNAPCT
EXEC PARM option 39
option 183
DYNAUTO
DYNSPC
EXEC PARM option 39
option 183
E
E11 user exit
initializing routines
920
499
E11 user exit (continued)

opening data sets 499
E15 user exit
changing records for sort and copy
applications 499
EXEC PARM option 40
passing records for sort and copy
applications 499
return codes 501
E15 User Exit
altering record length 518
interface with COBOL 523
LINKAGE SECTION code example for
fixed-length records 525
LINKAGE SECTION code example for
variable-length records 527
LINKAGE SECTION fields for
PROCEDURE DIVISION
requirements 529
return codes 527
E15/E35 return codes and EXITCK 537
E16 user exit
handling intermediate storage
miscalculation 503
return codes 503
sorting current records when NMAX
is exceeded 519
E17 user exit
closing data sets 503
E18 user exit
handling input data sets 504
using with QSAM/BSAM 504
using with VSAM 505
E19 user exit
handling output to work data
sets 507
E31 user exit
initializing routines 510
opening data sets 510
E32 user exit
handling input to a merge only 510
restriction with MERGE control
statement 166
return codes 512
E35 user exit
altering record length 519
Changing Records 512
EXEC PARM option 41
interface with COBOL 530
Procedure Division
Requirements 535
return codes 515
E37 user exit
E38 user exit
handling input data sets 517
using with VSAM 517
E39 user exit
handling output data sets 517
E39 user exit (continued)

using with VSAM 517
E61 user exit
altering control fields 520
information DFSORT passes to your
routine 509
modifying control fields 508
uses 508
edit masks
ICETOOL DISPLAY operator 601,
603
OUTFIL 271, 279
editing records
See reformatting records 125
efficiency
using main storage 799
EFS 783
efficiency 807
EXEC PARM option 40
exit routines 772
initialization phase 770
input phase 772
option 184
phases 768
processing 769
termination phase 772
using 767, 798
what you can do with EFS 773, 776
EFS interface
control statement length 785
control statement request list 779
control statement string 779, 781
D1 format 783
D2 format 784
defined 776
DFSORT action codes 779
extract buffer offsets list 785
function 768
information flags 786
message list 788
program context area 785
record lengths list 786
EFS program
addressing and residence mode 768
context area 785
examining, altering, ignoring control
statements 774
example 796
exit routine 776, 789, 790
function 789
functions 767, 773
interface parameter list 776, 788
opening and initializing data
sets 774
restrictions program in effect 88
restrictions when program in
effect 88
return codes you must supply 793
supplying messages 776
terminating DFSORT 776
user exit routine
addressing and residence
mode 792
EFS Program
example 798
EFS01
function description 789
parameter list 790
user exit routine 789
EFS02
address=0 798
function description 789
parameter list 792
user exit routine 790
EFSDPAFT 795
EFSDPBFR 795
elapsed time
improving with devices 802
END control statement
examples 95
function 83
using 95, 96
ENDAT=an parameter
of PARSE operand
ENDAT=BLANKS parameter
of PARSE operand
ENDAT=string parameter
of PARSE operand
ENDBEFR=an parameter
of PARSE operand
ENDBEFR=BLANKS parameter
of PARSE operand
ENDBEFR=string parameter
of PARSE operand
ENDREC parameter
option 227, 232
enhancing performance with installation
options 802
EODAD 505
EQUALS 6
efficiency 806
EXEC PARM option 40
MERGE control statement option 164
option 185
EQUCOUNT
efficiency 806
ERET
EROPT 505
error messages 25
error recovery routine
user exit 492
errors
critical 909
debugging jobs 91
diagnosing EFS 795
error recovery routines 492
ESTAE
recovery routine 907
ETOD date (DEn) and time (TEn) formats
descriptions 896, 897
where allowed 897
exceeding tape work space capacity 861,
862
EXEC statement
cataloged procedure
SORT 30, 65
SORTD 31, 63
cataloged procedures 30
defined 29
operands 32, 59
PARM options 32, 865
syntax 32
using 29, 60
using with control statements 32
execution phase run-time phase 768
exit
MODS control statement option 166
See also user exit 487
exit routine
EFS 788
EXITCK
ICEMAC installation option 499, 537
option 186
user exit return codes 537
EXLST 505, 508
EXPMAX installation option 20, 804, 853
EXPOLD installation option 20, 804, 853
EXPRES installation option 20, 804, 853
Extended Function Support
See EFS 767
extended parameter list
example 560
format 551, 553
extract buffer offsets list 785
F
FASTSRT
efficiency 803
FI (fixed-point) format
description 892
where allowed 897
Field and Constant Symbols
overview 731
field formats
control 445
ICETOOL operators
DISPLAY 599
RANGE 650, 662
SELECT 671
summary 452
fields
fixed position/length 1
variable position/length 1
FIELDS parameter
INREC statement 129
MERGE statement 163
FIELDS parameter (continued)

SORT statement 443
SUM statement 451
FIELDS=(COPY)
FIELDS=COPY
FILES parameter
option 226, 231
FILSZ
EXEC PARM option 41
variation summary 42
option 186
filtering records 96, 169
FINDREP parameter
INREC statement 143
OUTFIL statement 312
fixed century window 218
fixed parsed fields
extracting variable position/length
fields into
INREC control statement 128
OUTFIL control statement 236
OUTREC control statement 405
fixed position/length fields 1
FIXLEN=m parameter
of PARSE operand
FL ( hexadecimal floating-point) format
description 892
FL (hexadecimal floating-point) format
where allowed 897
floating-point data 124, 447, 492
floating-point fields 454
FNAMES parameter
option 226, 230
format
alternate character format 100
ASCII character format 100
ASCII leading sign format 100
ASCII trailing sign format 100
binary format 100
character format 100
date formats 100
fixed-point format 100
floating sign format 100
floating-point format 100
leading overpunch sign format 100
leading sign format 100
packed decimal format 100
signed free form format 100
time formats 100
trailing overpunch sign format 100
trailing sign format 100
unsigned free form format 100
user defined format (D1) 100
user defined format (D2) 100
zoned decimal format 100
format of 24-bit parameter list 544, 549
Index
921
format of 64-bit parameter list 554

format of extended parameter list 551,
553
FORMAT=f
option 96, 98
OMIT control statement option 169,
171
SUM control statement option 452
formatting
OUTFIL 269
four-digit year
transforming dates 218
FS (floating sign) format
description 893
where allowed 897
FSZEST installation option 20
FTOV parameter
option 228, 330
FTP site 4
functions of routines at user exits 490,
493
future date
G
GENER installation option 20
general coding rules 83, 88
general considerations 13, 14
GNPAD installation option 20, 816
GNTRUNC installation option 20, 816
group processing 322
INREC statement 147
OUTFIL statement 322
H
handling input data sets
E18 user exit 504
E38 user exit 517
handling input to a merge
E32 user exit 510
handling intermediate storage
miscalculation
E16 user exit 503
handling output data sets
E39 user exit 517
handling output to work data sets
E19 user exit 507
handling special I/O 492
HEADER parameter
DISPLAY operator 613
OCCUR operator 657
HEADER1 parameter
option 228, 338, 343
HEADER2 parameter
option 228, 351
922
HEADER3 parameter
option 361
hexadecimal constants 107, 131, 258,
340, 346, 408
hexadecimal display
OCCUR operator 653
HILEVEL=YES
MODS control statement option 167
Hipersorting
advantages to using 811
defined 811
Hiperspace
defined 811
limiting factors 190
HIPRMAX
efficiency 811
EXEC PARM option 43
option 190
home page (web) 4
how EFS works 768, 773
how user exit routines affect DFSORT
performance 495
I
I/O errors 492
ICEGENER
efficiency 813
example 849, 850
return codes 816
ICEGENER facility 813, 817
ICEMAC installation options 18, 24
ICETOOL 563
calling from a program 722
coding rules 574
complete sample job 850
COPY 565, 570, 575
COUNT 565, 579
DATASORT 585
DEFAULTS 590
description 563
DFSMSG DD statement 564
DISPLAY 565, 569, 594
example of simple job 566
examples 568, 569, 585, 588, 626, 645,
659, 664, 666, 673, 691, 710, 716, 719,
721
ICETOOL/DFSORT relationship 563
invoking 567
JCL 564, 571, 572, 573
JOBLIB DD statement 564
MERGE 642
MODE 565, 568, 570
OCCUR 565, 569, 646
operators 565, 566, 568, 569, 570, 575,
579, 585, 590, 594, 642, 646, 662, 664,
668, 677, 681, 708, 710, 718, 720
Parameter List Interface 567, 572, 722
RANGE 566, 569, 662
RESIZE 664
restrictions 573, 727
return codes 728
SELECT 566, 570, 668
ICETOOL (continued)
SORT 566, 570, 677
SPLICE 566, 681
statements 571, 574
STATS 566, 569, 708
STEPLIB DD statement 564
SUBSET 710
summary 564, 565
SYMNAMES DD statemen 564
SYMNOUT DD statemen 564
TOOLIN DD statement 564, 572
TOOLIN Interface 567, 572, 722
TOOLMSG DD statement 564, 571
UNIQUE 566, 570, 718
using SET and PROC symbols 567
using symbols 567
VERIFY 566, 568, 720
IDRCPCT installation option 20
IEBGENER 813
IEFUSI 809
IEXIT installation option 20
IFTHEN parameter
INREC statement 144
IFTRAIL parameter
option 370
IGNCKPT installation option 20
improving efficiency 799
efficiency 805
examples 108, 117
function 82
logical operator 124
relational condition 99
substring comparison operator 111
INCLUDE parameter
option 227, 233
INCLUDE/OMIT statement notes 124
INCLUDE/OMIT Statement Notes 124
including records 96, 169
user-defined data types 767
including records keeping records 1
information DFSORT passes to your
routine
E15 user exit 500
E32 user exit 510
E35 user exit 513
E61 user exit 509
information flags 786
Initialization Phase 770
initializing data sets 492, 770
initializing routines
E11 user exit 499
E31 user exit 510
initiating DFSORT
See invoking DFSORT 541
INPFIL control statement 88
input data set
requirements 12
valid types 12
input data sets
missing attributes of
set by DFSORT 68
input field 134

input file
size and efficiency 801
Input Phase 772
INREC control statement
column alignment 130
examples 150, 153, 154
function 82
input field 134
notes 148
separation field
binary zero separation 131
blank separation 130
character string separation 131
hexadecimal string separation 132
using 125, 153, 154
inserting comment statements 87
inserting records 492
installation defaults
displaying with DEFAULTS operator
(ICETOOL) 590
listing with ICETOOL 17
summary of options 18
installation defaults options, installation
defaults, installation 17
installation options 18, 24, 173
installation options, using to enhance
performance 802
insufficient intermediate storage 860
intermediate storage 862
Internet 4
introducing DFSORT 1, 26
invoking DFSORT
24-bit parameter list 543, 544
64-bit parameter list 553, 554
dynamically 541
extended parameter list 550, 553
from a program 541, 562
methods 4
using JCL 27
IOMAXBF installation option 20
J
Japanese characters 13, 767
JCL 27
cataloged procedure 63, 65
cataloged procedures, specifying 30
DD statement summary 27
EFS coding rules 779
EXEC statement 29
improving DFSORT efficiency 799
JOB statement 29
overview 27
procedures, cataloged 30
required 27
JCL DD statements 542
JCL DD Statements 552
JCL-invoked DFSORT 866, 874
job control language
see also JCL 27
JOB statement
defined 29
using 29
JOBLIB DD statement
defined 27
using 64
Join
SPLICE operator
JOINKEYS
overview 457
681
K
key-sequenced data set (KSDS)
key, defined 5
keyboard
navigation 911, 913
PF keys 911, 913
shortcut keys 911, 913
16
L
label field 84
length
altered control statement 785
LRECL for variable-length record 16
maximum record 14
original control statement 785
record descriptor word (RDW) 16
record lengths list 786
limitations
data set 13
length
maximum record 14
minimum block 14
minimum record 14
record
maximum length 14
storage constraints 14
LINES parameter
option 228
LINK 541
writing macro instructions 559
link-editing
performance 806
user exit routines 498
linkage conventions 496
linkage editor 65
linkage examples 497
LIST
EXEC PARM option 44
option 192
with an EFS program 776
LISTX
EXEC PARM option 44
option 192
loading user exit routines 496
locale
affecting INCLUDE and OMIT
processing 108
affecting MERGE processing 163
defined 7
restrictions
CHALT 177
EFS 40, 185
LOCALE
efficiency 805, 806
EXEC PARM option 45
option 193
using 7
logical operator 124
LookAt message retrieval tool xvi
lookup and change 228, 296, 383
LS (leading sign) format
description 894
where allowed 897
M
macro instructions
See system macro instructions 541
main features of sources of DFSORT
options 864, 865
main storage
allocating
consequences of increasing 808
allocating efficiently 807
minimum 807
releasing 809
tuning 807
using efficiently 807, 810
MAINSIZE 52
allocating storage 807
option 194
Major Call 1 796
Major Call 2 796
Major Call 3 797
Major Call 4 798
Major Call 5 798
major control field 6
master console messages 25
Match
SPLICE operator 681
MAX
EXEC PARM option 52
maximizing performance 799
maximum of fields and constants
INREC 136
OUTFIL 282
OUTREC 412
MAXLIM
memory object
definition 195, 197
specifying with EXEC PARM 45, 46
specifying with OPTION control
statement 195, 197
memory object sorting
advantages 813
considerations 813
definition 813
MERGE control statement
examples 165
function 81
using 162, 165
merge examples 845
Index
923
merge restriction 561

MERGEIN
option 195
merging
defined 1
overview 13
records 162
specifying the estimated number of
records to merge 42
specifying the exact number of records
to merge 41
specifying the number of records to
merge 42
user-defined data types 767, 772
message data set 25
message list 788
message retrieval tool, LookAt xvi
messages
master console messages 25
message data set 25
return codes 25
migration 24
minimum block length 14
minimum of fields and constants
INREC 136
OUTFIL 282
OUTREC 412
minimum record length 14
MINLIM
minor control field 6
missing attributes of input data sets
set by DFSORT 68
modifying control fields
E61 user exit 508
with user exit 492
modifying modifying the collating
sequence changing the collating
sequence 88
MODS control statement
examples 168, 169
function 83
using 166, 169
modulus of fields and constants
INREC 136
OUTFIL 282
OUTREC 412
MOSIZE
EXEC PARM option 45
option 195
MOWORK
EXEC PARM option 46
MOWRK
option 197
MSGCON installation option 21
MSGDDN
EXEC PARM option 46
924
MSGDDN (continued)
option 197
MSGPRT
alternate forms 47
EXEC PARM option 47
option 198
multiple output data sets
creating with OUTFIL 3, 225, 377
multiplying fields and constants
INREC 136
OUTFIL 282
OUTREC 412
N
navigation
keyboard 911, 913
NOABEND
EXEC PARM option 35
NOASSIST
NOBLKSET
efficiency 806
option 198
NOCFW
using on OPTION control
statement 92
NOCHALT
option 177
NOCHECK
option 177
NOCINV
efficiency 806
EXEC PARM option 36
option 178
NODETAIL parameter
option 228, 369
NOEQUALS
EXEC PARM option 40
option 185
NOESTAE
NOHEADER parameter
OCCUR operator 657
NOLIST
EXEC PARM option 44
option 192
NOLISTX
EXEC PARM option 44
option 192
NOMOWORK
EXEC PARM option 46
NOMOWRK
option 197
NOMSGDD installation option 21
NOOUTREL
EXEC PARM option 49
option 199
NOOUTSEC
option 199
NORESET
EXEC PARM option 50
NOSOLRF
EXEC PARM option 53
option 207
NOSZERO
EXEC PARM option 54
option 210
NOVERIFY
EXEC PARM option 55
option 212
NOVLLONG
EXEC PARM option 56
option 213
NOVLSCMP
EXEC PARM option 56
option 213
NOVLSHRT
EXEC PARM option 56
option 215
NOVSAMIO
EXEC PARM option 57
NOWRKREL
EXEC PARM option 58
option 217
NOWRKSEC
EXEC PARM option 58
option 218
NULLOFL installation option 21
NULLOFL parameter
option 337
NULLOUT
EXEC PARM option 47
option 199
NULLOUT installation option 21
numeric tests 120
numerice editing and formatting
NVSAMEMT
EXEC PARM option 57
NZDPRINT
EXEC PARM option 59
option 219
O
occurrences
ODMAXBF
EXEC PARM option 48
option 200
option 374
OL (leading overpunch sign) format
description 894
where allowed 897
OMIT control statement
efficiency 805
example 171, 172
function 82
using 172
OMIT parameter
option 227, 234
OMIT Statement Example 171, 172
omitting records 1, 169
opening and initializing data sets 492,
774
opening data sets
E11 user exit 499
E31 user exit 510
EFS 770
user exit routines 492
operand field 84
operation field 84
examples 220, 223
function 81
special handling 781
using 173, 223
OPTION Statement Examples 220, 223
OT (trailing overpunch sign) format
where allowed 897
OUTFIL
DD statement 73
digits needed for numeric fields 274
edit field formats and lengths 269
edit mask output field lengths 274
edit mask patterns 271
edit mask signs 273
efficiency 805
lookup and change 228, 296, 383
parsed input 296
producing reports 228, 256
storage limits 194, 374, 807
table lookup and change 296, 383
OUTFIL control statement
function 82
examples 377, 385, 386
function 82
using 223, 385, 386
outfil DD statement
defined 28
function 66
OUTFIL statements examples 377, 385,
386
OUTFIL statements notes 374
output data set

requirements 12
valid types 12
OUTREC control statement
column alignment 407
examples 153, 426, 428, 429
function 82
input field 410
separation field
binary zero separation 408
blank separation 407
character string separation 408
current date constant 408
future date constant 408
hexadecimal string separation 408
past date constant 408, 409
time constant 409
using 153, 428, 429
OUTREC control statements
using 402
OUTREC parameter
OUTREC statement examples 426, 428,
429
OUTREC statement notes 424
OUTREL
EXEC PARM option 49
OUTSEC installation option 22
overflow 150, 454
OVERLAY parameter
INREC statement 141
OVERRGN 809
override tables 865
overriding
defaults 863
installation defaults 173
Overriding control statements 542
overview, DFSORT 1
OVFLO
EXEC PARM option 49
option 201
P
PAD
EXEC PARM option 49
option 201
padding
GNPAD 816
INCLUDE/OMIT 107
records 14, 107
truncating 107
PAGEHEAD parameter
OUTFIL control statements 363
PAIR=APOST parameter
of PARSE operand
PAIR=QUOTE parameter
of PARSE operand
parameter list
control statements 543, 550, 554
description 865
format 544, 551
PARM options
PARMDDN installation option 22
PARSE parameter
INREC statement 127
parsing records 1
passing control to user exits 166
passing records
past date
PD (packed decimal) format
description 891
where allowed 897
PD0 (part of packed decimal) format
description 892
where allowed 897
performance
application design 800
dataspace sorting 812
efficient blocking 800
Hipersorting 811
HIPRMAX 811
ICEGENER 813
improving elapsed time with
devices 802
JCL 799
main storage 807
maximizing 799
merging techniques 801
ODMAXBF effects 374
options that degrade 806
options that enhance 802
sorting techniques 800
specifying data sets 801
temporary work space 810
using BLDINDEX support 817
using DFSORT's Performance Booster
for The SAS System 817
Pipe
Sort example 835
PROC symbol
using 29
processing and invoking programs 909
processing of error abends with A-type
messages 909
processing order, record 7
processing user-defined data types with
EFS program user exit routines 776
program control statements
64-bit parameter list 554
extended parameter list 550, 553
program DD statements 65
Program DD statements 78
program invocation, defined 5
program phase
defined 488
Index
925
program phase (continued)

initialization 770
input 772
termination 772
Q
QSAM
data set 12
data set considerations
E18 user exit 504
E19 user exit 507
15
R
RANGE operator (ICETOOL) 662
rearranging records
See sorting records 442
Recommendation
comparing padded bytes to excess
bytes in the binary field 116
record
blocking 800
changing with user exit routines 512
copying 163
data types 13
deleting 96, 169
with OMIT control statement 169
descriptor word (RDW) 16
EFS constraints 14
estimated number to be sorted 42
exact number to be sorted 41
formatting 125
inserting, deleting, and altering 492
maximum length 14
merging 162
minimum length 14
modifying with user exit 492
number to be sorted 42
padding 107, 816
passing with user exit routines 499
processing for OUTFIL 227
processing order 7, 124, 148, 149, 424
EFS 793
reformatting 402
sorting 442
storage constraints 14
summing 2, 451
E35 user exit 516
with user exits 492
truncating 107, 816
variable-length
efficiency 801
coding notes 441
examples 441
function 83
using 438
record processing order 793
record type
specifying 438
records
duplicate 451, 647, 668
unique
926
records (continued)
unique (continued)
UNIQUE operator
(ICETOOL) 718
recovering from unexpected abends
unexpected abends 908
REFORMAT control statement
using 442
reformatting records
after processing
examples 426
before processing
examples 150
with INREC statement
BUILD 125
FIELDS 125
IFTHEN 125
OVERLAY 125
with OUTFIL statement
BUILD 252
IFTHEN 252
OUTREC 252
OVERLAY 252
with OUTREC statement
BUILD 402
FIELDS 402
IFTHEN 402
OVERLAY 402
REGION
determining storage 807
size 807
Related reading
additional functions with ICETOOL
SELECT not available with
XSUM 454
relational condition
constants
character string format 104
date string format 118
decimal number format 102
hexadecimal string format 107
defined 99
description 99
format 99, 107, 117
relational condition format 121
remark field 85
REMOVECC parameter
option 370
RENT 496
reordering control fields
See reformatting records 125, 402
REPEAT parameter
option 334
REPEAT=v parameter
of PARSE operand
report
ANSI carriage control character 229,
256, 331, 345, 351, 354, 359, 361, 365,
370, 375
report (continued)
header, OUTFIL 339
ICETOOL DISPLAY 595, 636
ICETOOL OCCUR 647, 661
OUTFIL elements 3, 226
producing for OUTFIL 228, 256
trailer, OUTFIL 345
requesting a SNAP dump 795
requirements
input data set 12
JCL 27
output data set 12
RESALL
EXEC PARM option 50
option 202
RESERVEX 35
RESET
option 202
residence mode
EFS program 768
EFS program user exit routine 792
user exits 494
RESINV 809
option 203
RESIZE operator (ICETOOL) 664
restarting after an abend 907
Restriction
invoking DFSORT using ICEMAN 4
using OUTREC instead of INREC
could cause overflow 151
Restrictions
ICETOOL limitations 563
restrictions for dynamic invocation 561,
562
Return Code
DFSORT 25
return codes
EFS 793
Return Codes
ICEGENER 816
ICETOOL 728
REXX examples 820
RMODE 498
rules for parsing 781
running DFSORT with JCL 61, 78
S
sample job streams 819
sample jobs listing installation
defaults 17
SAMPLE parameter
option 232
sample routines written in
Assembler 518, 520
sample routines written in COBOL 535,
536
SAS
DFSORT's Performance Booster for
The SAS System 817
SAVE parameter
option 227, 234
screen readers
accessibility 915
SDB
EXEC PARM option 51
option 204
SDB (system-determined block size)
SDBMSG installation option 22
SECTIONS parameter
option 228, 359
sending comments to IBM xix
separation field 130, 407
sequence numbers
INREC 140
OUTFIL OUTREC 308
OUTREC 416
SET symbol
using 29
SFF (signed free form) format
description 893
where allowed 897
shared tape units 63
shortcut keys 911, 913
SIZE
EXEC PARM option 52
option 186, 206
SKIP parameter
OUTFIL control statements 360
SKIPREC
efficiency 805
EXEC PARM option 53
option 206
sliding century window 218
SmartBatch pipe
and ICETOOL 727
SmartBatch pipes
OUTFIL example 384
SMF
option 206
SMF date (DTn) and time (TMn) formats
descriptions 600, 621, 651, 896
where allowed 897
SNAP dump 795
SOLRF
EXEC PARM option 53
option 207
SORT cataloged procedure 30, 31, 65
SORT control statement

effects of EQUALS 443
examples 449, 451
field formats 445
function 81
using 442, 451
sort examples 679, 821, 836
SORT operator (ICETOOL) 677
SORT statement examples 449, 451
SORT statement image example 543, 544
SORT statement note 449
SORTCKPT DD statement
function 66
using 75
SORTCNTL data set 865
SORTCNTL DD statement
defined 28
function 66
using 75
SORTD cataloged procedure 63
SORTDD
option 208
SORTDIAG DD statement
defined 28
function 66
using 78
SORTDKdd DD statement
function 66
using 78
SORTIN
option 209
SORTIN DD statement
defined 28
function 66
using 67, 69
sorting
defined 1
identifying information to sort 5
overview 12
records 442
specifying the estimated number of
records to sort 42
specifying the exact number of records
to sort 41
specifying the number of records to
sort 42
user-defined data types 767, 772
using data space 803
sorting records 442
SORTINnn DD statement
defined 28
duplicate 63
function 66
using 69, 71
SORTLIB
ICEMAC installation option 67
SORTLIB DD statement
defined 27
function 65
using 67
SORTLIB installation option 22
SORTMODS DD statement
defined 29
function 67
SORTOUT
option 209
OUTFIL ddname 226
SORTOUT DD statement
defined 28
function 66
using 73, 75
SORTSNAP DD statement
defined 28
function 66
using 78
SORTWKdd DD statement
defined 28
duplicate 63
function 66
using 71
SORTWKdd DD Statement
dataspace sorting 71
SPANINC
EXEC PARM option 53
option control statement 209
special handling of OPTION and DEBUG
control statements 781
specification/override of DFSORT
options 863, 890
specifying efficient sort/merge
techniques 800
specifying input/output data set
characteristics accurately 801
SPLICE operator (ICETOOL) 681
SPLIT parameter
option 335
SPLIT1R parameter
option 337
SPLITBY parameter
option 336, 338
SS 111
STARTAFT=an parameter
of PARSE operand
STARTAFT=BLANKS parameter
of PARSE operand
STARTAFT=string parameter
of PARSE operand
STARTAT=an parameter
of PARSE operand
STARTAT=BLANKS parameter
of PARSE operand
STARTAT=NONBLANK parameter
of PARSE operand
STARTAT=string parameter
of PARSE operand
STARTREC parameter
option 227, 232
STATS operator (ICETOOL) 708
Index
927
STEPLIB DD statement
defined 27
using 64
STOPAFT
efficiency 805
EXEC PARM option 54
option 210
storage
efficient 801, 860
exceeding capacity 860, 861
intermediate 810
limits, OUTFIL 374
main
releasing 809
tuning 807
specifying for user exit routine 166
temporary 810
tracks versus cylinders 801, 859
user exit routine 493, 522
storage administrator examples 819
storage usage
records at E35 user exit 516
SUBPOS=y parameter
of PARSE operand
SUBSET operator (ICETOOL) 710
substring comparison operator 111
substring comparison tests 112
relational condition format 111
subtracting fields and constants
INREC 136
OUTFIL 282
OUTREC 412
SUM control statement 455
description 451
efficiency 805
examples 454, 455
function 83
summary field 451
using 455
SUM statement examples 454, 455
SUM statement notes 453
summarizing records 451
summary field
formats 451
table of formats and lengths 452
Summary Field Formats and Lengths
Table 452
summing
records 451, 492
records at E35 user exit 516
summing records adding record
values 2
supplying messages for printing to the
message data set 776
SVC installation option 23
symbols
using 4
Symbols
Comment and Blank Statement 734
example 732
for fields and constants 731
in DFSORT Statements 747
928
Symbols (continued)
in ICETOOL Operators
DISPLAY 757
ICETOOL Example 758
OCCUR 758
RANGE 757, 758
SELECT 758
SPLICE 758
STATS, UNIQUE and
VERIFY 758
in ICETOOL statements 757
INCLUDE and OMIT 749
INREC and OUTREC 750
Keyword Statements 744
Notes 765
OUTFIL 752
overview 731
SORT and MERGE 748
SUM 749
Symbol Statements 734
SYMNAMES DD Statement 733
SYMNAMES Statements 734
SYMNOUT DD Statement 734
using SET and PROC 760
SYMNAMES DD statement
defined 27
function 66
SYMNOUT DD statement
defined 28
function 66
SYNAD 504, 508
syntax diagrams
option control statement 173
SYSABEND DD statement
defined 28
using 65
SYSIN data set 865
SYSIN DD statement
defined 28
using 64
SYSLIN DD statement
defined 29
using 65
SYSLMOD DD statement
defined 29
using 65
SYSMDUMP DD statement
defined 28
using 65
SYSOUT DD statement
defined 27
using 64
SYSPRINT DD statement
defined 28
using 65
system DD statements 63, 65
system macro instructions
defined 541
using 541, 553
writing 559, 560
system-determined block size (SDB) 74
SYSUDUMP DD statement
defined 28
using 65
SYSUT1 DD statement
defined 28
using 65
SZERO
EXEC PARM option 54
option 210
T
tape
capacity considerations 860, 862
efficiency 806, 811, 860
insufficient intermediate storage 861
work space capacity 861
work storage devices 811
TCn (TOD time) format
description 897
where allowed 897
TEn (ETOD time) format
description 897
where allowed 897
terminating DFSORT
E35 user exit 529
with user exits 493
TEXIT installation option 23
time constant 133, 260, 409
time formats
where allowed 897
TMAXLIM
TMn (SMF time) format
description 896
where allowed 897
TOD date (DCn) and time (TCn) formats
where allowed 897
tracks 801, 859
TRAILER1 parameter
option 228, 343, 351
TRAILER2 parameter
option 228, 359
TRAILER3 parameter
option 363
Translate characters
ALTSEQ 88, 135
lowercase to uppercase 125, 265
uppercase to lowercase 125
TRUNC
EXEC PARM option 55
option 211
truncating
GNTRUNC 816
INCLUDE/OMIT 107
records 14
truncating records 107
TS (trailing sign) format
description 894
where allowed 897
TUNE
tuning main storage 807
two-digit year
conversion 218, 385
sorting 451
transforming dates 3, 226
TYPE
option 438
U
UFF (unsigned free form) format
description 893
where allowed 897
UNIQUE operator (ICETOOL) 718
unique records
UNIQUE operator (ICETOOL) 718
user exit
activating 487
addressing and residence mode 494
assembler routines
input phase 498
output phase 510
COBOL routines
input phase 523
output phase 529
overview 521
conventions for routines 495
DFSORT performance 495
E11 499
E15 499, 523
E16 503
E17 503
E18 504
E19 507
E31 510
E32 510
E35 512, 529
E37 517
E38 517
E39 517
E61 508
efficiency 806
functions 490
language requirements 487
link-editing 498
linkage conventions 496
loading routines 496
overview 487
passing control with MODS control
statement 166
summary of rules 495, 498
using RECORD control
statement 438
using routines 487, 517
using your own routines 518, 537
user exit linkage conventions 496
user interface
ISPF 911, 913
TSO/E 911, 913
USEWKDD
option 212
using control statements from other IBM

programs 88
using DD statements 61, 78
using DFSORT program control
statements 81, 455
using options that enhance
performance 802
V
variable position/length fields 1
extracting into parsed fields
INREC control statement 128
OUTFIL control statement 236
OUTREC control statement 405
variable-length record
longest record length 16
record descriptor word 16
VERIFY
efficiency 806
EXEC PARM option 55
option 212
VERIFY operator (ICETOOL) 720
VIO
ICEMAC installation option 78
VLFILL parameter
option 330
VLLONG
EXEC PARM option 56
option 213
VLSCMP
EXEC PARM option 56
option 213
VLSHRT
EXEC PARM option 56
option 215
VLTRAIL parameter
option 332
VLTRIM parameter
option 331
VSAM
data set 12
data set considerations 16
E18 user exit 505
E38 user exit 517
E39 user exit 517
key-sequenced data set (KSDS) 16
maximum record size
with INREC control
statement 149, 424
user exit functions 493
using RECORD control
statement 438
VSAMBSP installation option 23
VSAMEMT
EXEC PARM option 57
option 216
VSAMIO
EXEC PARM option 57
option 217
VTOF parameter
option 309
W
Web 4
web site 4
work space
requirements for DFSORT 853
using 853, 862
WRKREL
EXEC PARM option 58
option 217
WRKSEC
EXEC PARM option 58
option 218
X
XCTL
using 541
writing macro instructions
559
Y
Y2 formats
description 895
where allowed 897
Y2PAST
EXEC PARM option 59
option 218
Year 2000
century window 218
comparing dates 119
ordering dates 446
Z
z/OS file systems 16
ZD (zoned decimal) format
description 891
where allowed 897
ZDPRINT
EXEC PARM option 59
Index
929
ZDPRINT (continued)
option 219
930

Printed in USA
SC23-6878-01

DFSORT

Uploaded by

Copyright:

Available Formats

DFSORT

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DFSORT

Uploaded by

Copyright:

Available Formats

z/OS

DFSORT Application Programming Guide

How to send your comments to IBM

Summary of changes . . . . . . . . xxi

Chapter 1. Introducing DFSORT . . . . 1

Chapter 2. Invoking DFSORT with Job

Copyright IBM Corp. 1973, 2015

Using SET and PROC symbols in DFSORT control

Chapter 3. Using DFSORT program

Including records in the output data

Chapter 4. Using a JOINKEYS

z/OS V2R2 DFSORT Application Programming Guide

Example 6 - Paired and unpaired F1/F2 records

Chapter 5. Using your own user exit

COBOL E15 user exit: passing or changing

Chapter 6. Invoking DFSORT from a

Chapter 7. Using ICETOOL. . . . . . 563

Chapter 8. Using symbols for fields

Chapter 9. Using extended function

z/OS V2R2 DFSORT Application Programming Guide

EFS formats for SORT, MERGE, INCLUDE, and

Chapter 10. Improving efficiency . . . 799

Chapter 11. Examples of DFSORT job

Example 3. Sort with ASCII tapes . . . . .

Appendix A. Using work space . . . . 853

DFSPARM data set . . . . . . . .

Appendix C. Data format descriptions

DFSORT data formats . . . . . .

Appendix D. EBCDIC and ASCII

Appendix E. DFSORT abend

Appendix F. Accessibility . . . . . . 911

z/OS V2R2 DFSORT Application Programming Guide

Copyright IBM Corp. 1973, 2015

COBOL E15 Routine Example (FLR) . . . .

z/OS V2R2 DFSORT Application Programming Guide

Copyright IBM Corp. 1973, 2015

TRAN=ATOE ASCII-to-EBCDIC translation

87. Functions of an Extended Function Support

z/OS V2R2 DFSORT Application Programming Guide

95. Work Space Requirements of the Various Tape

About this document

How to use this document

Required product knowledge

z/OS DFSORT Messages, Codes and Diagnosis

z/OS MVS JCL Reference

z/OS MVS JCL User's Guide

z/OS DFSMS Using Data Sets

z/OS DFSMS Using Magnetic Tapes

z/OS V2R2 DFSORT Application Programming Guide

z/OS DFSMS Macro Instructions for Data Sets

z/OS DFSMS Using Data Sets

z/OS MVS JCL Reference

z/OS MVS JCL User's Guide

z/OS MVS Programming: Assembler Services

z/OS MVS Programming: Assembler Services

z/OS Install/Migration page

z/OS Install/Migration page

z/OS V2R2.0 UNIX System Services User's

The z/OS DFSORT Application Programming Guide is a part of a more extensive

Planning For and