InterBase 6

Embedded
SQL Guide

Borland/INPRISE
100 Enterprise Way, Scotts Valley, CA 95066 http://www.interbase.com
November 12, 1999 (C:\TechPubs60\60DocSet\Doc\EmbedSQL\EmbedSQLTitle.fm5)

Inprise/Borland may have patents and/or pending patent applications covering subject matter in this
document. The furnishing of this document does not convey any license to these patents.
Copyright 1999 Inprise/Borland. All rights reserved. All InterBase products are trademarks or registered
trademarks of Inprise/Borland. All Borland products are trademarks or registered trademarks of
Inprise/Borland. Other brand and product names are trademarks or registered trademarks of their
respective holders.

1INT0055WW21000 6E1R0699
Table of Contents

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

List of Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv

CHAPTER 1 Using the Embedded SQL Guide

Who should use this guide . . . . . . . . . . . . . . . . . . . . . 17
Topics covered in this guide . . . . . . . . . . . . . . . . . . . . 18
Sample database and applications . . . . . . . . . . . . . . . . . 19

CHAPTER 2 Application Requirements

Requirements for all applications . . . . . . . . . . . . . . . . . 21
Porting considerations for SQL . . . . . . . . . . . . . . . . 22
Porting considerations for DSQL . . . . . . . . . . . . . . . 22
Declaring host variables. . . . . . . . . . . . . . . . . . . . 22
Declaring and initializing databases . . . . . . . . . . . . . . . . 25
Using SET DATABASE . . . . . . . . . . . . . . . . . . . . . 26
Using CONNECT . . . . . . . . . . . . . . . . . . . . . . . . 27
Working with a single database . . . . . . . . . . . . . . . 27
SQL statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Error handling and recovery . . . . . . . . . . . . . . . . . . . . 28
Closing transactions . . . . . . . . . . . . . . . . . . . . . . . . . 28
Accepting changes . . . . . . . . . . . . . . . . . . . . . . . 29
Undoing changes . . . . . . . . . . . . . . . . . . . . . . . . 29
Closing databases . . . . . . . . . . . . . . . . . . . . . . . . . . 30
DSQL requirements . . . . . . . . . . . . . . . . . . . . . . . . . 30
Declaring an XSQLDA . . . . . . . . . . . . . . . . . . . . . 31
DSQL limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Using database handles . . . . . . . . . . . . . . . . . . . . 32
Using the active database . . . . . . . . . . . . . . . . . . . 33
Using transaction names . . . . . . . . . . . . . . . . . . . 33
Preprocessing programs . . . . . . . . . . . . . . . . . . . . . . 34

EMBEDDED SQL GUIDE iii

CHAPTER 3 Working with Databases
Declaring a database . . . . . . . . . . . . . . . . . . . . . . . . 35
Declaring multiple databases . . . . . . . . . . . . . . . . . 37
Preprocessing and run time databases . . . . . . . . . . . 38
Controlling SET DATABASE scope . . . . . . . . . . . . . . . 39
Specifying a connection character set . . . . . . . . . . . . . . . 40
Opening a database . . . . . . . . . . . . . . . . . . . . . . . . . 41
Using simple CONNECT statements. . . . . . . . . . . . . . 41
Additional CONNECT syntax. . . . . . . . . . . . . . . . . . 45
Attaching to multiple databases . . . . . . . . . . . . . . . 46
Handling CONNECT errors . . . . . . . . . . . . . . . . . . 46
Setting database cache buffers . . . . . . . . . . . . . . . . 47
Accessing an open database . . . . . . . . . . . . . . . . . . . . 48
Differentiating table names . . . . . . . . . . . . . . . . . . . . . 49
Closing a database . . . . . . . . . . . . . . . . . . . . . . . . . . 49
With DISCONNECT . . . . . . . . . . . . . . . . . . . . . . . 50
With COMMIT and ROLLBACK . . . . . . . . . . . . . . . . . 50

CHAPTER 4 Working with Transactions

Starting the default transaction . . . . . . . . . . . . . . . . . . 55
Starting without SET TRANSACTION . . . . . . . . . . . . . 55
Starting with SET TRANSACTION . . . . . . . . . . . . . . . 56
Starting a named transaction . . . . . . . . . . . . . . . . . . . . 57
Naming transactions . . . . . . . . . . . . . . . . . . . . . . 58
Specifying SET TRANSACTION behavior . . . . . . . . . . . 60
Using transaction names in data statements . . . . . . . . . . . 72
Ending a transaction . . . . . . . . . . . . . . . . . . . . . . . . 74
Using COMMIT . . . . . . . . . . . . . . . . . . . . . . . . . 75
Using ROLLBACK . . . . . . . . . . . . . . . . . . . . . . . . 78
Working with multiple transactions . . . . . . . . . . . . . . . . 79
The default transaction . . . . . . . . . . . . . . . . . . . . 80
Using cursors . . . . . . . . . . . . . . . . . . . . . . . . . . 80
A multi-transaction example . . . . . . . . . . . . . . . . . 81
Working with multiple transactions in DSQL . . . . . . . . . . . 83

iv INTERBASE 6
TABLE OF CONTENTS

Modifying transaction behavior with “?”. . . . . . . . . . . 83

CHAPTER 5 Working with Data Definition Statements

Creating metadata . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Creating a database . . . . . . . . . . . . . . . . . . . . . . 89
Creating a domain . . . . . . . . . . . . . . . . . . . . . . . 90
Creating a table. . . . . . . . . . . . . . . . . . . . . . . . . 91
Creating a view . . . . . . . . . . . . . . . . . . . . . . . . . 94
Creating an index . . . . . . . . . . . . . . . . . . . . . . . 96
Creating generators . . . . . . . . . . . . . . . . . . . . . . 98
Dropping metadata . . . . . . . . . . . . . . . . . . . . . . . . . 98
Dropping an index . . . . . . . . . . . . . . . . . . . . . . . 99
Dropping a view . . . . . . . . . . . . . . . . . . . . . . . . 99
Dropping a table . . . . . . . . . . . . . . . . . . . . . . . 100
Altering metadata . . . . . . . . . . . . . . . . . . . . . . . . . 100
Altering a table . . . . . . . . . . . . . . . . . . . . . . . . 101
Altering a view . . . . . . . . . . . . . . . . . . . . . . . . 105
Altering an index. . . . . . . . . . . . . . . . . . . . . . . 105

CHAPTER 6 Working with Data

Supported datatypes . . . . . . . . . . . . . . . . . . . . . . . . 108
Understanding SQL expressions . . . . . . . . . . . . . . . . . 110
Using the string operator in expressions . . . . . . . . . 112
Using arithmetic operators in expressions . . . . . . . . 113
Using logical operators in expressions . . . . . . . . . . 113
Using comparison operators in expressions . . . . . . . 114
Determining precedence of operators . . . . . . . . . . . 122
Using CAST( ) for datatype conversions . . . . . . . . . . 125
Using UPPER() on text data . . . . . . . . . . . . . . . . . 125
Understanding data retrieval with SELECT . . . . . . . . . . . 126
Listing columns to retrieve with SELECT. . . . . . . . . . 128
Specifying host variables with INTO . . . . . . . . . . . . 131
Listing tables to search with FROM. . . . . . . . . . . . . 132
Restricting row retrieval with WHERE . . . . . . . . . . . 134

EMBEDDED SQL GUIDE v

 Sorting rows with ORDER BY . . . . . . . . . . . . . . . . 138
Grouping rows with GROUP BY. . . . . . . . . . . . . . . 139
Restricting grouped rows with HAVING . . . . . . . . . . 140
Appending tables with UNION . . . . . . . . . . . . . . . 141
Specifying a query plan with PLAN. . . . . . . . . . . . . 142
Selecting a single row . . . . . . . . . . . . . . . . . . . . . . . 143
Selecting multiple rows . . . . . . . . . . . . . . . . . . . . . . 144
Declaring a cursor . . . . . . . . . . . . . . . . . . . . . . 145
Opening a cursor. . . . . . . . . . . . . . . . . . . . . . . 146
Fetching rows with a cursor . . . . . . . . . . . . . . . . 147
Closing the cursor . . . . . . . . . . . . . . . . . . . . . . 149
A complete cursor example . . . . . . . . . . . . . . . . . 149
Selecting rows with NULL values . . . . . . . . . . . . . . 151
Selecting rows through a view . . . . . . . . . . . . . . . 152
Selecting multiple rows in DSQL . . . . . . . . . . . . . . . . . 152
Declaring a DSQL cursor . . . . . . . . . . . . . . . . . . 153
Opening a DSQL cursor . . . . . . . . . . . . . . . . . . . 154
Fetching rows with a DSQL cursor . . . . . . . . . . . . . 154
Joining tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Choosing join columns . . . . . . . . . . . . . . . . . . . 156
Using inner joins . . . . . . . . . . . . . . . . . . . . . . . 156
Using outer joins . . . . . . . . . . . . . . . . . . . . . . . 159
Using nested joins . . . . . . . . . . . . . . . . . . . . . . 161
Using subqueries . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Simple subqueries . . . . . . . . . . . . . . . . . . . . . . 162
Correlated subqueries . . . . . . . . . . . . . . . . . . . . 163
Inserting data . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Using VALUES to insert columns . . . . . . . . . . . . . . 164
Using SELECT to insert columns . . . . . . . . . . . . . . 165
Inserting rows with NULL column values . . . . . . . . . 166
Inserting data through a view . . . . . . . . . . . . . . . 169
Specifying transaction names in an INSERT . . . . . . . . 169
Updating data . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

vi INTERBASE 6
TABLE OF CONTENTS

Updating multiple rows . . . . . . . . . . . . . . . . . . . 171

NULLing columns with UPDATE . . . . . . . . . . . . . . . 174
Updating through a view . . . . . . . . . . . . . . . . . . 174
Specifying transaction names in UPDATE . . . . . . . . . 175
Deleting data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Deleting multiple rows . . . . . . . . . . . . . . . . . . . 177
Deleting through a view. . . . . . . . . . . . . . . . . . . 180
Specifying transaction names in a DELETE . . . . . . . . 180

CHAPTER 7 Working with Dates

Selecting dates and times . . . . . . . . . . . . . . . . . . . . . 184
Inserting dates and times . . . . . . . . . . . . . . . . . . . . . 185
Updating dates and times . . . . . . . . . . . . . . . . . . . . . 186
Using CAST( ) to convert dates and times . . . . . . . . . . . . 186
Casting from SQL datatypes to datetime datatypes . . . . 187
Casting from datetime datatypes to other SQL datatypes 188
Using date literals . . . . . . . . . . . . . . . . . . . . . . . . . 189

CHAPTER 8 Working with Blob Data

What is a Blob? . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
How are Blob data stored? . . . . . . . . . . . . . . . . . . . . 192
Blob subtypes. . . . . . . . . . . . . . . . . . . . . . . . . 193
Blob database storage . . . . . . . . . . . . . . . . . . . . 194
Blob segment length . . . . . . . . . . . . . . . . . . . . . 195
Overriding segment length . . . . . . . . . . . . . . . . . 196
Accessing Blob data with SQL . . . . . . . . . . . . . . . . . . 196
Selecting Blob data . . . . . . . . . . . . . . . . . . . . . 196
Inserting Blob data . . . . . . . . . . . . . . . . . . . . . 199
Updating Blob data . . . . . . . . . . . . . . . . . . . . . 200
Deleting Blob data . . . . . . . . . . . . . . . . . . . . . . 201
Accessing Blob data with API calls . . . . . . . . . . . . . . . 202
Filtering Blob data . . . . . . . . . . . . . . . . . . . . . . . . . 203
Using the standard InterBase text filters . . . . . . . . . 203
Using an external Blob filter . . . . . . . . . . . . . . . . 203

EMBEDDED SQL GUIDE vii

 Writing an external Blob filter . . . . . . . . . . . . . . . . . . 205
Filter types . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Read-only and write-only filters . . . . . . . . . . . . . . 206
Defining the filter function . . . . . . . . . . . . . . . . . 206

CHAPTER 9 Using Arrays

Creating arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Multi-dimensional arrays . . . . . . . . . . . . . . . . . . 216
Specifying subscript ranges . . . . . . . . . . . . . . . . . 217
Accessing arrays . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Selecting data from an array . . . . . . . . . . . . . . . . 218
Inserting data into an array. . . . . . . . . . . . . . . . . 219
Selecting from an array slice . . . . . . . . . . . . . . . . 220
Updating data in an array slice . . . . . . . . . . . . . . . 222
Testing a value in a search condition . . . . . . . . . . . 223
Using host variables in array subscripts . . . . . . . . . . 224
Using arithmetic expressions with arrays . . . . . . . . . 224

CHAPTER 10 Working with Stored Procedures

Using stored procedures . . . . . . . . . . . . . . . . . . . . . 226
Procedures and transactions . . . . . . . . . . . . . . . . 226
Security for procedures . . . . . . . . . . . . . . . . . . . 227
Using select procedures . . . . . . . . . . . . . . . . . . . . . . 227
Calling a select procedure . . . . . . . . . . . . . . . . . . 228
Using a select procedure with cursors . . . . . . . . . . . 228
Using executable procedures . . . . . . . . . . . . . . . . . . . 229
Executing a procedure . . . . . . . . . . . . . . . . . . . . 229
Executing a procedure in a DSQL application . . . . . . 230

CHAPTER 11 Working with Events

Understanding the event mechanism . . . . . . . . . . . . . . 233
Signaling event occurrences . . . . . . . . . . . . . . . . . . . 234
Registering interest in events . . . . . . . . . . . . . . . . . . . 235
Registering interest in multiple events . . . . . . . . . . . . . 236
Waiting for events with EVENT WAIT . . . . . . . . . . . . . . . 236

viii INTERBASE 6
TABLE OF CONTENTS

Responding to events . . . . . . . . . . . . . . . . . . . . . . . 237

CHAPTER 12 Error Handling and Recovery

Standard error handling . . . . . . . . . . . . . . . . . . . . . 240
WHENEVER statements. . . . . . . . . . . . . . . . . . . . 240
Testing SQLCODE directly . . . . . . . . . . . . . . . . . . 243
Combining error-handling techniques . . . . . . . . . . . 244
Guidelines for error handling . . . . . . . . . . . . . . . 246
Additional InterBase error handling . . . . . . . . . . . . . . . 247
Displaying error messages . . . . . . . . . . . . . . . . . 247
Capturing SQL error messages . . . . . . . . . . . . . . . 248
Capturing InterBase error messages . . . . . . . . . . . . 249
Handling InterBase error codes . . . . . . . . . . . . . . 251

CHAPTER 13 Using Dynamic SQL

Overview of the DSQL programming process . . . . . . . . . 255
DSQL limitations . . . . . . . . . . . . . . . . . . . . . . . . . . 256
Accessing databases . . . . . . . . . . . . . . . . . . . . . 256
Handling transactions . . . . . . . . . . . . . . . . . . . . 257
Creating a database . . . . . . . . . . . . . . . . . . . . . 258
Processing Blob data . . . . . . . . . . . . . . . . . . . . 259
Processing array data . . . . . . . . . . . . . . . . . . . . 259
Writing a DSQL application . . . . . . . . . . . . . . . . . . . . 259
SQL statements that DSQL can process . . . . . . . . . . 260
SQL character strings . . . . . . . . . . . . . . . . . . . . 261
Value parameters in statement strings . . . . . . . . . . . 261
Understanding the XSQLDA . . . . . . . . . . . . . . . . . . . 262
XSQLDA field descriptions . . . . . . . . . . . . . . . . . 264
XSQLVAR field descriptions . . . . . . . . . . . . . . . . . 265
Input descriptors . . . . . . . . . . . . . . . . . . . . . . . 266
Output descriptors . . . . . . . . . . . . . . . . . . . . . . 266
Using the XSQLDA_LENGTH macro . . . . . . . . . . . . . 267
SQL datatype macro constants . . . . . . . . . . . . . . . 268
Handling varying string datatypes . . . . . . . . . . . . . 270

EMBEDDED SQL GUIDE ix

 NUMERIC and DECIMAL datatypes . . . . . . . . . . . . . 271
Coercing datatypes . . . . . . . . . . . . . . . . . . . . . . 271
Aligning numerical data . . . . . . . . . . . . . . . . . . . 272
DSQL programming methods . . . . . . . . . . . . . . . . . . . 273
Method 1: Non-query statements without parameters . . 273
Method 2: Non-query statements with parameters . . . . 275
Method 3: Query statements without parameters . . . . 278
Method 4: Query statements with parameters . . . . . . 283

CHAPTER 14 Preprocessing, Compiling, and Linking

Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Using gpre . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Using a file extension to specify language . . . . . . . . 295
Specifying the source file . . . . . . . . . . . . . . . . . . 295
Compiling and linking . . . . . . . . . . . . . . . . . . . . . . 297
Compiling an Ada program . . . . . . . . . . . . . . . . . 297
Linking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298

APPENDIX A InterBase Document Conventions

The InterBase documentation set . . . . . . . . . . . . . . . . 300
Printing conventions . . . . . . . . . . . . . . . . . . . . . . . 301
Syntax conventions . . . . . . . . . . . . . . . . . . . . . . . . 302

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

x INTERBASE 6
List of Tables

Table 1.1 Chapters in the InterBase 6 Embedded SQL Guide . . . . . . . . . . 18

Table 3.1 CONNECT syntax summary . . . . . . . . . . . . . . . . . . . . . . . 45
Table 4.1 SQL transaction management statements . . . . . . . . . . . . . . . 54
Table 4.2 Default transaction default behavior. . . . . . . . . . . . . . . . . . 56
Table 4.3 SET TRANSACTION parameters . . . . . . . . . . . . . . . . . . . . . 61
Table 4.4 ISOLATION LEVEL options. . . . . . . . . . . . . . . . . . . . . . . . 63
Table 4.5 InterBase management of classic transaction conflicts . . . . . . . . 65
Table 4.6 Isolation level Interaction with SELECT and UPDATE . . . . . . . . . . 68
Table 4.7 Table reservation options for the RESERVING clause . . . . . . . . . 71
Table 5.1 Data definition statements supported for embedded applications . . 88
Table 6.1 Datatypes supported by InterBase . . . . . . . . . . . . . . . . . . 108
Table 6.2 Elements of SQL expressions . . . . . . . . . . . . . . . . . . . . 110
Table 6.3 Arithmetic operators . . . . . . . . . . . . . . . . . . . . . . . . . 113
Table 6.4 InterBase comparison operators requiring subqueries . . . . . . . 115
Table 6.5 Operator precedence by operator type . . . . . . . . . . . . . . . 122
Table 6.6 Mathematical operator precedence . . . . . . . . . . . . . . . . . 123
Table 6.7 Comparison operator precedence . . . . . . . . . . . . . . . . . . 123
Table 6.8 Logical operator precedence . . . . . . . . . . . . . . . . . . . . 124
Table 6.9 Compatible datatypes for CAST() . . . . . . . . . . . . . . . . . . . . . 125
Table 6.10 SELECT statement clauses . . . . . . . . . . . . . . . . . . . . . . 127
Table 6.11 Aggregate functions in SQL . . . . . . . . . . . . . . . . . . . . . 129
Table 6.12 Elements of WHERE clause SEARCH conditions . . . . . . . . . . . 136
Table 7.1 Casting from SQL datatypes to datetime datatypes . . . . . . . . . 187
Table 7.2 Casting from datetime datatypes to other SQL datatypes . . . . . . 188
Table 8.1 Blob subtypes defined by InterBase . . . . . . . . . . . . . . . . 193
Table 8.2 API Blob calls . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Table 8.3 isc_blob_ctl structure field descriptions . . . . . . . . . . . . . . 209
Table 8.4 Blob access operations . . . . . . . . . . . . . . . . . . . . . . . . 211
Table 8.5 Blob filter status values . . . . . . . . . . . . . . . . . . . . . . . 213
Table 12.1 Possible SQLCODE values . . . . . . . . . . . . . . . . . . . . . . 240
Table 13.1 XSQLDA field descriptions . . . . . . . . . . . . . . . . . . . . . 264
Table 13.2 XSQLVAR field descriptions . . . . . . . . . . . . . . . . . . . . . 265
Table 13.3 SQL datatypes, macro expressions, and C datatypes . . . . . . . . 268

EMBEDDED SQL GUIDE xi

Table 13.4 SQL statement strings and recommended processing methods . . 273
Table 14.1 gpre language switches available on all platforms . . . . . . . . . 292
Table 14.2 Additional gpre language switches . . . . . . . . . . . . . . . . . 292
Table 14.3 gpre option switches . . . . . . . . . . . . . . . . . . . . . . . . . 293
Table 14.4 Language-specific gpre option switches . . . . . . . . . . . . . . . 294
Table 14.5 File extensions for language specification . . . . . . . . . . . . . 295
Table A.1 Books in the InterBase 6 documentation set . . . . . . . . . . . . 300
Table A.2 Text conventions . . . . . . . . . . . . . . . . . . . . . . . . . . 301
Table A.3 Syntax conventions . . . . . . . . . . . . . . . . . . . . . . . . . 302

xii INTERBASE 6
List of Figures

Figure 8.1 Relationship of a Blob ID to Blob segments in a database . . . . . 195

Figure 8.2 Filtering from lowercase to uppercase . . . . . . . . . . . . . . . 204
Figure 8.3 Filtering from uppercase to lowercase . . . . . . . . . . . . . . . 205
Figure 8.4 Filter interaction with an application and a database . . . . . . . 207
Figure 13.1 XSQLDA and XSQLVAR relationship . . . . . . . . . . . . . . . . . 263

EMBEDDED SQL GUIDE xiii

List of Examples

Example 2.1 Using host-language data structures to reference table columns . . .25
Example 2.2 Using SQL to read table data into a C struct . . . . . . . . . . . . . .25

EMBEDDED SQL GUIDE xv

 CHAPTER

Using the Chapter1

1
Embedded SQLGuide

The InterBase Embedded SQL Guide is a task-oriented explanation of how to write,

preprocess, compile, and link embedded SQL and DSQL database applications using
InterBase and a host programming language, either C or C++. This chapter describes
who should read this book, and provides a brief overview of its chapters.

Who should use this guide

The InterBase Embedded SQL Guide is intended for database applications programmers.
It assumes a general knowledge of:
g SQL
g Relational database programming
g C programming
The Embedded SQL Guide assumes little or no previous experience with InterBase. See
the Operations Guide for an introduction to InterBase and the Language Reference for
an introduction to SQL.

EMBEDDED SQL GUIDE 17

 CHAPTER 1 USING THE EMBEDDED SQL GUIDE

Note The Embedded SQL Guide focuses on embedded SQL and DSQL programming in
C or C++. It does not address Delphi-specific topics.

Topics covered in this guide

The following table provides a brief description of each chapter in this Embedded SQL
Guide:

Chapter Description
Chapter 1, “Using the Embedded SQL Guide” Introduces the structure of the book and describes its
intended audience.
Chapter 2, “Application Requirements” Describes elements common to programming all SQL and
DSQL applications.
Chapter 3, “Working with Databases” Describes using SQL statements that deal with databases.
Chapter 4, “Working with Transactions” Explains how to use and control transactions with SQL
statements.
Chapter 5, “Working with Data Definition Describes how to embed SQL data definition statements in
Statements” applications.
Chapter 6, “Working with Data” Explains how to select, insert, update, and delete standard
SQL data in applications.
Chapter 7, “Working with Dates” Describes how to select, insert, update, and delete DATE, TIME,
and TIMESTAMP data in applications.
Chapter 8, “Working with Blob Data” Describes how to select, insert, update, and delete Blob data
in applications.
Chapter 9, “Using Arrays” Describes how to select, insert, update, and delete array data
in applications.
Chapter 10, “Working with Stored Procedures” Explains how to call stored procedures in applications.
Chapter 11, “Working with Events” Explains how triggers interact with applications. Describes
how to register interest in events, wait on them, and respond
to them in applications.
Chapter 12, “Error Handling and Recovery” Describes how to trap and handle SQL statement errors in
applications.
TABLE 1.1 Chapters in the InterBase 6 Embedded SQL Guide

18 INTERBASE 6
SAMPLE DATABASE AND APPLICATIONS

Chapter Description
Chapter 13, “Using Dynamic SQL” Describes how to write DSQL applications.
Chapter 14, “Preprocessing, Compiling, and Linking” Describes how to convert source code into an executable
application.
Appendix A, “InterBase Document Conventions” Lists typefaces and special characters used in this book to
describe syntax and identify object types.
TABLE 1.1 Chapters in the InterBase 6 Embedded SQL Guide

Sample database and applications

@@The InterBase examples subdirectory contains several subidirectories containing useful
items worth noting, including:
g UDF subdirectory The ib_udf.sql file, which declares the UDFs in the library provided by
InterBase; the library of UDFs is in the InterBase lib subdirectory and is named ib_udf.dll
on Wintel platforms and ib_udf on UNIX platforms
g Database subdirectory A sample database, employee.gdb
g API subdirectory Sample application source code, which produces several sample
applications when compiled; see the makefile and the readme in the examples directory for
more information about the sample applications and how to compile them
The Embedded SQL Guide makes use of the sample database and source code for its
examples wherever possible.

EMBEDDED SQL GUIDE 19

 CHAPTER 1 USING THE EMBEDDED SQL GUIDE

20 INTERBASE 6
 CHAPTER

Application Requirements
Chapter2
2
This chapter describes programming requirements for InterBase SQL and dynamic SQL
(DSQL) applications. Many of these requirements may also affect developers moving
existing applications to InterBase.

Requirements for all applications

All embedded applications must include certain declarations and statements to ensure
proper handling by the InterBase preprocessor, gpre, and to enable communication
between SQL and the host language in which the application is written. Every application
must:
g Declare host variables to use for data transfer between SQL and the application.
g Declare and set the databases accessed by the program.
g Create transaction handles for each non-default transaction used in the program.
g Include SQL (and, optionally, DSQL) statements.
g Provide error handling and recovery.
g Close all transactions and databases before ending the program.

EMBEDDED SQL GUIDE 21

 CHAPTER 2 APPLICATION REQUIREMENTS

Dynamic SQL applications, those applications that build SQL statements at run time, or
enable users to build them, have additional requirements. For more information about
DSQL requirements, see “DSQL requirements” on page 30.
For more information about using gpre, see Chapter 14, “Preprocessing, Compiling,
and Linking.”

Porting considerations for SQL

When porting existing SQL applications to InterBase, other considerations may be
necessary. For example, many SQL variants require that host variables be declared
between BEGIN DECLARE SECTION and END DECLARE SECTION statements; InterBase has no
such requirements, but gpre can correctly handle section declarations from ported
applications. For additional portability, declare all host-language variables within
sections.

Porting considerations for DSQL

When porting existing DSQL applications to InterBase, statements that use another
vendor’s SQL descriptor area (SQLDA) must be modified to accommodate the extended
SQLDA (XSQLDA) used by InterBase.

Declaring host variables

A host variable is a standard host-language variable used to hold values read from a
database, to assemble values to write to a database, or to store values describing database
search conditions. SQL uses host variables in the following situations:
g During data retrieval, SQL moves the values in database fields into host variables where
they can be viewed and manipulated.
g When a user is prompted for information, host variables are used to hold the data until
it can be passed to InterBase in an SQL INSERT or UPDATE statement.
g When specifying search conditions in a SELECT statement, conditions can be entered
directly, or in a host variable. For example, both of the following SQL statement fragments
are valid WHERE clauses. The second uses a host-language variable, country, for
comparison with a column, COUNTRY:
… WHERE COUNTRY = 'Mexico';
… WHERE COUNTRY = :country;

22 INTERBASE 6
REQUIREMENTS FOR ALL APPLICATIONS

One host variable must be declared for every column of data accessed in a database. Host
variables may either be declared globally like any other standard host-language variable,
or may appear within an SQL section declaration with other global declarations. For more
information about reading from and writing to host variables in SQL programs, see
Chapter 6, “Working with Data.”
Host variables used in SQL programs are declared just like standard language variables.
They follow all standard host-language rules for declaration, initialization, and
manipulation. For example, in C, variables must be declared before they can be used as
host variables in SQL statements:
int empno; char fname[26], lname[26];

For compatibility with other SQL variants, host variables can also be declared between
BEGIN DECLARE SECTION and END DECLARE SECTION statements.

4 Section declarations
Many SQL implementations expect host variables to be declared between BEGIN DECLARE
SECTION and END DECLARE SECTION statements. For portability and compatibility,
InterBase supports section declarations using the following syntax:
EXEC SQL
BEGIN DECLARE SECTION;
<hostvar>;
. . .
EXEC SQL
END DECLARE SECTION;

For example, the following C code fragment declares three host variables, empno, fname,
and lname, within a section declaration:
EXEC SQL
BEGIN DECLARE SECTION;
int empno;
char fname[26];
char lname[26];
EXEC SQL
END DECLARE SECTION;

Additional host-language variables not used in SQL statements can be declared outside
DECLARE SECTION statements.

EMBEDDED SQL GUIDE 23

 CHAPTER 2 APPLICATION REQUIREMENTS

4 Using BASED ON to declare variables

InterBase supports a declarative clause, BASED ON, for creating C language character
variables based on column definitions in a database. Using BASED ON ensures that the
resulting host-language variable is large enough to hold the maximum number of
characters in a CHAR or VARCHAR database column, plus an extra byte for the
null-terminating character expected by most C string functions.
BASED ON uses the following syntax:
BASED ON <dbcolumn> hostvar;

For example, the following statements declare two host variables, fname, and lname,
based on two column definitions, FIRSTNAME, and LASTNAME, in an employee database:
BASED ON EMP.FIRSTNAME fname;
BASED ON EMP.LASTNAME lname;

Embedded in a C or C++ program, these statements generate the following host- variable
declarations during preprocessing:
char fname[26];
char lname[26];

To use BASED ON, follow these steps:

1. Use SET DATABASE to specify the database from which column definitions are
to be drawn.
2. Use CONNECT to attach to the database.
3. Declare a section with BEGIN DECLARE SECTION.
4. Use the BASED ON statement to declare a string variable of the appropriate
type.
The following statements show the previous BASED ON declarations in context:
EXEC SQL
SET DATABASE EMP = ’employee.gdb’;
EXEC SQL
CONNECT EMP;
EXEC SQL
BEGIN DECLARE SECTION;
int empno;
BASED ON EMP.FIRSTNAME fname;
BASED ON EMP.LASTNAME lname;
EXEC SQL
END DECLARE SECTION;

24 INTERBASE 6
DECLARING AND INITIALIZING DATABASES

4 Host-language data structures

If a host language supports data structures, data fields within a structure can correspond
to a collection of database columns. For example, the following C declaration creates a
structure, BILLING_ADDRESS, that contains six variables, or data members, each of which
corresponds to a similarly named column in a table:

EXAMPLE 2.1 Using host-language data structures to reference table columns

struct
{
char fname[25];
char lname[25];
char street[30];
char city[20];
char state[3];
char zip[11];
} billing_address;

SQL recognizes data members in structures, but information read from or written to a
structure must be read from or written to individual data members in SQL statements. For
example, the following SQL statement reads data from a table into variables in the C
structure, BILLING_ADDRESS:

EXAMPLE 2.2 Using SQL to read table data into a C struct

EXEC SQL
SELECT FNAME, LNAME, STREET, CITY, STATE, ZIP
INTO :billing_address.fname, :billing_address.lname,
:billing_address.street, :billing_address.city,
:billing_address.state, :billing_address.zip
FROM ADDRESSES WHERE CITY = ’Brighton’;

Declaring and initializing databases

An SQL program can access multiple InterBase databases at the same time. Each database
used in a multiple-database program must be declared and initialized before it can be
accessed in SQL transactions. Programs that access only a single database need not
declare the database or assign a database handle if, instead, they specify a database on
the gpre command line.

IMPORTANT DSQL programs cannot connect to multiple databases.

EMBEDDED SQL GUIDE 25

 CHAPTER 2 APPLICATION REQUIREMENTS

InterBase supports the following SQL statements for handling databases:

g SET DATABASE declares the name of a database to access, and assigns it to a database
handle.
g CONNECT opens a database specified by a handle, and allocates it system resources.
Database handles replace database names in CONNECT statements. They can also be used
to qualify table names within transactions. For a complete discussion of database
handling in SQL programs, see Chapter 3, “Working with Databases.”

Using SET DATABASE

The SET DATABASE statement is used to:
g Declare a database handle for each database used in an SQL program.
g Associate a database handle with an actual database name. Typically, a database handle
is a mnemonic abbreviation of the actual database name.
SET DATABASE instantiates a host variable for the database handle without requiring an
explicit host variable declaration. The database handle contains a pointer used to
reference the database in subsequent SQL statements. To include a SET DATABASE
statement in a program, use the following syntax:
EXEC SQL
SET DATABASE handle = ’dbname’;

A separate statement should be used for each database. For example, the following
statements declare a handle, DB1, for the employee.gdb database, and another handle,
DB2, for employee2.gdb:
EXEC SQL
SET DATABASE DB1 = ’employee.gdb’;
EXEC SQL
SET DATABASE DB2 = ’employee2.gdb’;

Once a database handle is created and associated with a database, the handle can be used
in subsequent SQL database and transaction statements that require it, such as CONNECT.
Note SET DATABASE also supports user name and password options. For a complete
discussion of SET DATABASE options, see Chapter 3, “Working with Databases.”

26 INTERBASE 6
DECLARING AND INITIALIZING DATABASES

Using CONNECT
The CONNECT statement attaches to a database, opens the database, and allocates system
resources for it. A database must be opened before its tables can be used. To include
CONNECT in a program, use the following syntax:
EXEC SQL
CONNECT handle;

A separate statement can be used for each database, or a single statement can connect to
multiple databases. For example, the following statements connect to two databases:
EXEC SQL
CONNECT DB1;
EXEC SQL
CONNECT DB2;

The next example uses a single CONNECT to establish both connections:

EXEC SQL
CONNECT DB1, DB2;

Once a database is connected, its tables can be accessed in subsequent transactions. Its
handle can qualify table names in SQL applications, but not in DSQL applications. For a
complete discussion of CONNECT options and using database handles, see Chapter 3,
“Working with Databases.”

Working with a single database

In single-database programs preprocessed without the gpre -m switch, SET DATABASE and
CONNECT are optional. The -m switch suppresses automatic generation of transactions.
Using SET DATABASE and CONNECT is strongly recommended, however, especially as a way
to make program code as self-documenting as possible. If you omit these statements, take
the following steps:
1. Insert a section declaration in the program code where global variables are
defined. Use an empty section declaration if no host-language variables are
used in the program. For example, the following declaration illustrates an
empty section declaration:
EXEC SQL
BEGIN DECLARE SECTION;
EXEC SQL
END DECLARE SECTION;

EMBEDDED SQL GUIDE 27

 CHAPTER 2 APPLICATION REQUIREMENTS

2. Specify a database name on the gpre command line at precompile time. A

database need not be specified if a program contains a CREATE DATABASE
statement.
For more information about working with a single database in an SQL program, see
Chapter 3, “Working with Databases.”

SQL statements
An SQL application consists of a program written in a host language, like C or C++, into
which SQL and dynamic SQL (DSQL) statements are embedded. Any SQL or DSQL
statement supported by InterBase can be embedded in a host language. Each SQL or
DSQL statement must be:
g Preceded by the keywords EXEC SQL.
g Ended with the statement terminator expected by the host language. For example, in C
and C++, the host terminator is the semicolon (;).
For a complete list of SQL and DSQL statements supported by InterBase, see the
Language Reference.

Error handling and recovery

Every time an SQL statement is executed, it returns an error code in the SQLCODE variable.
SQLCODE is declared automatically for SQL programs during preprocessing with gpre. To
catch run-time errors and recover from them when possible, SQLCODE should be
examined after each SQL operation.
SQL provides the WHENEVER statement to monitor SQLCODE and direct program flow to
recovery procedures. Alternatively, SQLCODE can be tested directly after each SQL
statement executes. For a complete discussion of SQL error handling and recovery, see
Chapter 12, “Error Handling and Recovery.”

Closing transactions
Every transaction should be closed when it completes its tasks, or when an error occurs
that prevents it from completing its tasks. Failure to close a transaction before a program
ends can cause limbo transactions, where records are entered into the database, but are
neither committed or rolled back. Limbo transactions can be cleaned up using the
database administration tools provided with InterBase.

28 INTERBASE 6
CLOSING TRANSACTIONS

Accepting changes
The COMMIT statement ends a transaction, makes the transaction’s changes available to
other users, and closes cursors. A COMMIT is used to preserve changes when all of a
transaction’s operations are successful. To end a transaction with COMMIT, use the
following syntax:
EXEC SQL
COMMIT TRANSACTION name;

For example, the following statement commits a transaction named MYTRANS:

EXEC SQL
COMMIT TRANSACTION MYTRANS;

For a complete discussion of SQL transaction control, see Chapter 4, “Working with
Transactions.”

Undoing changes
The ROLLBACK statement undoes a transaction’s changes, ends the current transaction,
and closes open cursors. Use ROLLBACK when an error occurs that prevents all of a
transaction’s operations from being successful. To end a transaction with ROLLBACK, use
the following syntax:
EXEC SQL
ROLLBACK TRANSACTION name;

For example, the following statement rolls back a transaction named MYTRANS:
EXEC SQL
ROLLBACK TRANSACTION MYTRANS;

To roll back an unnamed transaction (i.e., the default transaction), use the following
statement:
EXEC SQL
ROLLBACK;

For a complete discussion of SQL transaction control, see Chapter 4, “Working with
Transactions.”

EMBEDDED SQL GUIDE 29

 CHAPTER 2 APPLICATION REQUIREMENTS

Closing databases
Once a database is no longer needed, close it before the program ends, or subsequent
attempts to use the database may fail or result in database corruption. There are two ways
to close a database:
g Use the DISCONNECT statement to detach a database and close files.
g Use the RELEASE option with COMMIT or ROLLBACK in a program.
DISCONNECT, COMMIT RELEASE, and ROLLBACK RELEASE perform the following tasks:
g Close open database files.
g Close remote database connections.
g Release the memory that holds database descriptions and InterBase engine-compiled
requests.
Note Closing databases with DISCONNECT is preferred for compatibility with the SQL-92
standard.
For a complete discussion of closing databases, see Chapter 3, “Working with
Databases.”

DSQL requirements
DSQL applications must adhere to all the requirements for all SQL applications and meet
additional requirements as well. DSQL applications enable users to enter ad hoc SQL
statements for processing at run time. To handle the wide variety of statements a user
might enter, DSQL applications require the following additional programming steps:
g Declare as many extended SQL descriptor areas (XSQLDAs) as are needed in the
application; typically a program must use one or two of these structures. Complex
applications may require more.
g Declare all transaction names and database handles used in the program at compile time;
names and handles are not dynamic, so enough must be declared to accommodate the
anticipated needs of users at run time.
g Provide a mechanism to get SQL statements from a user.
g Prepare each SQL statement received from a user for processing.
PREPARE loads statement information into the XSQLDA.
g EXECUTE each prepared statement.

30 INTERBASE 6
DSQL REQUIREMENTS

EXECUTE IMMEDIATE combines PREPARE and EXECUTE in a single statement. For more
information, see the Language Reference.
In addition, the syntax for cursors involving Blob data differs from that of cursors for
other datatypes. For more information about Blob cursor statements, see the Language
Reference.

Declaring an XSQLDA
The extended SQL descriptor area (XSQLDA) is used as an intermediate staging area for
information passed between an application and the InterBase engine. The XSQLDA is used
for either of the following tasks:
g Pass input parameters from a host-language program to SQL.
g Pass output, from a SELECT statement or stored procedure, from SQL to the host-language
program.
A single XSQLDA can be used for only one of these tasks at a time. Many applications
declare two XSQLDAs, one for input, and another for output.
The XSQLDA structure is defined in the InterBase header file, ibase.h, that is automatically
included in programs when they are preprocessed with gpre.

IMPORTANT DSQL applications written using versions of InterBase prior to 3.3 use an older SQL
descriptor area, the SQLDA. The SQLDA and the gpre -sqlda switch are no longer supported.
Older applications should be modified to use the XSQLDA.
To create an XSQLDA for a program, a host-language datatype of the appropriate type must
be set up in a section declaration. For example, the following statement creates two
XSQLDA structures, inxsqlda, and outxsqlda:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
XSQLDA inxsqlda;
XSQLDA outxsqlda;
. . .
EXEC SQL
END DECLARE SECTION;
. . .

EMBEDDED SQL GUIDE 31

 CHAPTER 2 APPLICATION REQUIREMENTS

When an application containing XSQLDA declarations is preprocessed, gpre automatically

includes the header file, ibase.h, which defines the XSQLDA as a host-language datatype.
For a complete discussion of the structure of the XSQLDA, see Chapter 13, “Using
Dynamic SQL.”

DSQL limitations
DSQL enables programmers to create flexible applications that are capable of handling a
wide variety of user requests. Even so, not every SQL statement can be handled in a
completely dynamic fashion. For example, database handles and transaction names must
be specified when an application is written, and cannot be changed or specified by users
at run time. Similarly, while InterBase supports multiple databases and multiple
simultaneous transactions in an application, the following limitations apply:
g Only a single database can be accessed at a time.
g Transactions can only operate on the currently active database.
g Users cannot specify transaction names in DSQL statements; instead, transaction names
must be supplied and manipulated when an application is coded.

Using database handles

Database handles are always static, and can only be declared when an application is
coded. Enough handles must be declared to satisfy the expected needs of users. Once a
handle is declared, it can be assigned to a user-specified database at run time with SET
DATABASE, as in the following C code fragment:
. . .
EXEC SQL
SET DATABASE DB1 = ’dummydb.gdb’;
EXEC SQL
SET DATABASE DB2 = ’dummydb.gdb’;
. . .

printf("Specify first database to open: ");

gets(fname1);
printf("\nSpecify second database to open: ");
gets(fname2);

EXEC SQL
SET DATABASE DB1 = :fname1;
EXEC SQL

32 INTERBASE 6
DSQL LIMITATIONS

SET DATABASE DB2 = :fname2;

. . .

For a complete discussion of SET DATABASE, see Chapter 3, “Working with Databases.”

Using the active database

A DSQL application can only work with one database at a time, even if the application
attaches to multiple databases. All DSQL statements operate only on the currently active
database, the last database associated with a handle in a SET DATABASE statement.
Embedded SQL statements within a DSQL application can operate on any open database.
For example, all DSQL statements entered by a user at run time might operate against a
single database specified by the user, but the application might also contain non-DSQL
statements that record user entries in a log database.
For a complete discussion of SET DATABASE, see Chapter 3, “Working with Databases.”

Using transaction names

Many SQL statements support an optional transaction name parameter, used to specify
the controlling transaction for a specific statement. Transaction names can be used in
DSQL applications, too, but must be set up when an application is compiled. Once a
name is declared, it can be directly inserted into a user statement only by the application
itself.
After declaration, use a transaction name in an EXECUTE or EXECUTE IMMEDIATE statement
to specify the controlling transaction, as in the following C code fragment:
. . .
EXEC SQL
BEGIN DECLARE SECTION:
long first, second; /* declare transaction names */
EXEC SQL
END DECLARE SECTION;
. . .
first = second = 0L; /* initialize names to zero */
. . .
EXEC SQL
SET TRANSACTION first; /* start transaction 1 */
EXEC SQL
SET TRANSACTION second; /* start transaction 2 */

EMBEDDED SQL GUIDE 33

 CHAPTER 2 APPLICATION REQUIREMENTS

printf("\nSQL> ");
gets(userstatement);

EXEC SQL
EXECUTE IMMEDIATE TRANSACTION first userstatement;
. . .

For complete information about named transactions, see Chapter 4, “Working with
Transactions.”

Preprocessing programs
After an SQL or DSQL program is written, and before it is compiled and linked, it must
be preprocessed with gpre, the InterBase preprocessor. gpre translates SQL statements and
variables into statements and variables that the host-language compiler accepts. For
complete information about preprocessing with gpre, see Chapter 14, “Preprocessing,
Compiling, and Linking.”

34 INTERBASE 6
 CHAPTER

Working with Databases

Chapter3
3
This chapter describes how to use SQL statements in embedded applications to control
databases. There are three database statements that set up and open databases for access:
g SET DATABASE declares a database handle, associates the handle with an actual database
file, and optionally assigns operational parameters for the database.
g SET NAMES optionally specifies the character set a client application uses for CHAR,
VARCHAR, and text Blob data. The server uses this information to transliterate from a
database’s default character set to the client’s character set on SELECT operations, and to
transliterate from a client application’s character set to the database character set on
INSERT and UPDATE operations.
g CONNECT opens a database, allocates system resources for it, and optionally assigns
operational parameters for the database.
All databases must be closed before a program ends. A database can be closed by using
DISCONNECT, or by appending the RELEASE option to the final COMMIT or ROLLBACK in a
program.

Declaring a database
Before a database can be opened and used in a program, it must first be declared with
SET DATABASE to:

EMBEDDED SQL GUIDE 35

 CHAPTER 3 WORKING WITH DATABASES

g Establish a database handle.

g Associate the database handle with a database file stored on a local or remote node.
A database handle is a unique, abbreviated alias for an actual database name. Database
handles are used in subsequent CONNECT, COMMIT RELEASE, and ROLLBACK RELEASE
statements to specify which databases they should affect. Except in dynamic SQL (DSQL)
applications, database handles can also be used inside transaction blocks to qualify, or
differentiate, table names when two or more open databases contain identically named
tables.
Each database handle must be unique among all variables used in a program. Database
handles cannot duplicate host-language reserved words, and cannot be InterBase
reserved words.
The following statement illustrates a simple database declaration:
EXEC SQL
SET DATABASE DB1 = ’employee.gdb’;

This database declaration identifies the database file, employee.gdb, as a database the
program uses, and assigns the database a handle, or alias, DB1.
If a program runs in a directory different from the directory that contains the database
file, then the file name specification in SET DATABASE must include a full path name, too.
For example, the following SET DATABASE declaration specifies the full path to
employee.gdb:
EXEC SQL
SET DATABASE DB1 = ’/interbase/examples/employee.gdb’;

If a program and a database file it uses reside on different hosts, then the file name
specification must also include a host name. The following declaration illustrates how a
Unix host name is included as part of the database file specification on a TCP/IP network:
EXEC SQL
SET DATABASE DB1 = ’jupiter:/usr/interbase/examples/employee.gdb’;

On a Windows network that uses the Netbeui protocol, specify the path as follows:
EXEC SQL
SET DATABASE DB1 = ’//venus/C:/Interbase/examples/employee.gdb’;

36 INTERBASE 6
DECLARING A DATABASE

Declaring multiple databases

An SQL program, but not a DSQL program, can access multiple databases at the same
time. In multi-database programs, database handles are required. A handle is used to:
g Reference individual databases in a multi-database transaction.
g Qualify table names.
g Specify databases to open in CONNECT statements.
g Indicate databases to close with DISCONNECT, COMMIT RELEASE, and ROLLBACK RELEASE.
DSQL programs can access only a single database at a time, so database handle use is
restricted to connecting to and disconnecting from a database.
In multi-database programs, each database must be declared in a separate SET DATABASE
statement. For example, the following code contains two SET DATABASE statements:
. . .
EXEC SQL
SET DATABASE DB2 = ’employee2.gdb’;
EXEC SQL
SET DATABASE DB1 = ’employee.gdb’;
. . .

4 Using handles for table names

When the same table name occurs in more than one simultaneously accessed database,
a database handle must be used to differentiate one table name from another. The
database handle is used as a prefix to table names, and takes the form handle.table.
For example, in the following code, the database handles, TEST and EMP, are used to
distinguish between two tables, each named EMPLOYEE:
. . .
EXEC SQL
DECLARE IDMATCH CURSOR FOR
SELECT TESTNO INTO :matchid FROM TEST.EMPLOYEE
WHERE TESTNO > 100;
EXEC SQL
DECLARE EIDMATCH CURSOR FOR
SELECT EMPNO INTO :empid FROM EMP.EMPLOYEE
WHERE EMPNO = :matchid;
. . .

EMBEDDED SQL GUIDE 37

 CHAPTER 3 WORKING WITH DATABASES

IMPORTANT This use of database handles applies only to embedded SQL applications. DSQL
applications cannot access multiple databases simultaneously.

4 Using handles with operations

In multi-database programs, database handles must be specified in CONNECT statements
to identify which databases among several to open and prepare for use in subsequent
transactions.
Database handles can also be used with DISCONNECT, COMMIT RELEASE, and ROLLBACK
RELEASE to specify a subset of open databases to close.
To open and prepare a database with CONNECT, see “Opening a database” on page 41.
To close a database with DISCONNECT, COMMIT RELEASE, or ROLLBACK RELEASE, see
“Closing a database” on page 49. To learn more about using database handles in
transactions, see “Accessing an open database” on page 48.

Preprocessing and run time databases

Normally, each SET DATABASE statement specifies a single database file to associate with a
handle. When a program is preprocessed, gpre uses the specified file to validate the
program’s table and column references. Later, when a user runs the program, the same
database file is accessed. Different databases can be specified for preprocessing and run
time when necessary.

4 Using the COMPILETIME clause

A program can be designed to run against any one of several identically structured
databases. In other cases, the actual database that a program will use at runtime is not
available when a program is preprocessed and compiled. In such cases, SET DATABASE can
include a COMPILETIME clause to specify a database for gpre to test against during
preprocessing. For example, the following SET DATABASE statement declares that
employee.gdb is to be used by gpre during preprocessing:
EXEC SQL
SET DATABASE EMP = COMPILETIME ’employee.gdb’;

IMPORTANT The file specification that follows the COMPILETIME keyword must always be a
hard-coded, quoted string.

38 INTERBASE 6
DECLARING A DATABASE

When SET DATABASE uses the COMPILETIME clause, but no RUNTIME clause, and does not
specify a different database file specification in a subsequent CONNECT statement, the
same database file is used both for preprocessing and run time. To specify different
preprocessing and runtime databases with SET DATABASE, use both the COMPILETIME and
RUNTIME clauses.

4 Using the RUNTIME clause

When a database file is specified for use during preprocessing, SET DATABASE can specify
a different database to use at run time by including the RUNTIME keyword and a runtime
file specification:
EXEC SQL
SET DATABASE EMP = COMPILETIME ’employee.gdb’
RUNTIME ’employee2.gdb’;

The file specification that follows the RUNTIME keyword can be either a hard-coded,
quoted string, or a host-language variable. For example, the following C code fragment
prompts the user for a database name, and stores the name in a variable that is used later
in SET DATABASE:
. . .
char db_name[125];
. . .
printf("Enter the desired database name, including node
and path):\n");
gets(db_name);
EXEC SQL
SET DATABASE EMP = COMPILETIME ’employee.gdb’ RUNTIME :db_name;
. . .

Note Host-language variables in SET DATABASE must be preceded, as always, by a colon.

Controlling SET DATABASE scope

By default, SET DATABASE creates a handle that is global to all modules in an application.
A global handle is one that may be referenced in all host-language modules comprising
the program. SET DATABASE provides two optional keywords to change the scope of a
declaration:
g STATIC limits declaration scope to the module containing the SET DATABASE statement. No
other program modules can see or use a database handle declared STATIC.

EMBEDDED SQL GUIDE 39

 CHAPTER 3 WORKING WITH DATABASES

g EXTERN notifies gpre that a SET DATABASE statement in a module duplicates a

globally-declared database in another module. If the EXTERN keyword is used, then
another module must contain the actual SET DATABASE statement, or an error occurs
during compilation.
The STATIC keyword is used in a multi-module program to restrict database handle access
to the single module where it is declared. The following example illustrates the use of the
STATIC keyword:
EXEC SQL
SET DATABASE EMP = STATIC ’employee.gdb’;

The EXTERN keyword is used in a multi-module program to signal that SET DATABASE in
one module is not an actual declaration, but refers to a declaration made in a different
module. gpre uses this information during preprocessing. The following example
illustrates the use of the EXTERN keyword:
EXEC SQL
SET DATABASE EMP = EXTERN ’employee.gdb’;

If an application contains an EXTERN reference, then when it is used at run time, the
actual SET DATABASE declaration must be processed first, and the database connected
before other modules can access it.
A single SET DATABASE statement can contain either the STATIC or EXTERN keyword, but not
both. A scope declaration in SET DATABASE applies to both
COMPILETIME and RUNTIME databases.

Specifying a connection character set

When a client application connects to a database, it may have its own character set
requirements. The server providing database access to the client does not know about
these requirements unless the client specifies them. The client application specifies its
character set requirement using the SET NAMES statement before it connects to the
database.
SET NAMES specifies the character set the server should use when translating data from
the database to the client application. Similarly, when the client sends data to the
database, the server translates the data from the client’s character set to the database’s
default character set (or the character set for an individual column if it differs from the
database’s default character set).
For example, the following statements specify that the client is using the DOS437
character set, then connect to the database:
EXEC SQL

40 INTERBASE 6
OPENING A DATABASE

SET NAMES DOS437;

EXEC SQL
CONNECT ’europe.gdb’ USER ’JAMES’ PASSWORD ’U4EEAH’;

For more information about character sets, see the Data Definition Guide. For the
complete syntax of SET NAMES and CONNECT, see the Language Reference.

Opening a database
After a database is declared, it must be attached with a CONNECT statement before it can
be used. CONNECT:
g Allocates system resources for the database.
g Determines if the database file is local, residing on the same host where the application
itself is running, or remote, residing on a different host.
g Opens the database and examines it to make sure it is valid.
InterBase provides transparent access to all databases, whether local or remote. If the
database structure is invalid, the on-disk structure (ODS) number does not correspond to
the one required by InterBase, or if the database is corrupt, InterBase reports an error,
and permits no further access.
Optionally, CONNECT can be used to specify:
g A user name and password combination that is checked against the server’s security
database before allowing the connect to succeed. User names can be up to 31 characters.
Passwords are restricted to 8 characters.
g An SQL role name that the user adopts on connection to the database, provided that the
user has previously been granted membership in the role. Regardless of role
memberships granted, the user belongs to no role unless specified with this ROLE clause.
The client can specify at most one role per connection, and cannot switch roles except
by reconnecting.
g The size of the database buffer cache to allocate to the application when the default cache
size is inappropriate.

Using simple CONNECT statements

In its simplest form, CONNECT requires one or more database parameters, each specifying
the name of a database to open. The name of the database can be a:
g Database handle declared in a previous SET DATABASE statement.

EMBEDDED SQL GUIDE 41

 CHAPTER 3 WORKING WITH DATABASES

g Host-language variable.
g Hard-coded file name.

4 Using a database handle

If a program uses SET DATABASE to provide database handles, those handles should be
used in subsequent CONNECT statements instead of hard-coded names. For example,
. . .
EXEC SQL
SET DATABASE DB1 = ’employee.gdb’;
EXEC SQL
SET DATABASE DB2 = ’employee2.gdb’;
EXEC SQL
CONNECT DB1;
EXEC SQL
CONNECT DB2;
. . .

There are several advantages to using a database handle with CONNECT:

g Long file specifications can be replaced by shorter, mnemonic handles.
g Handles can be used to qualify table names in multi-database transactions. DSQL
applications do not support multi-database transactions.
g Handles can be reassigned to other databases as needed.
g The number of database cache buffers can be specified as an additional CONNECT
parameter.
For more information about setting the number of database cache buffers, see “Setting
database cache buffers” on page 47.

4 Using strings or host-language variables

Instead of using a database handle, CONNECT can use a database name supplied at run
time. The database name can be supplied as either a host-language variable or a
hard-coded, quoted string.
The following C code demonstrates how a program accessing only a single database
might implement CONNECT using a file name solicited from a user at run time:
. . .
char fname[125];
. . .
printf(’Enter the desired database name, including node
and path):\n’);

42 INTERBASE 6
OPENING A DATABASE

gets(fname);
. . .
EXEC SQL
CONNECT :fname;
. . .

Tip This technique is especially useful for programs that are designed to work with many
identically structured databases, one at a time, such as CAD/CAM or architectural
databases.

MULTIPLE DATABASE IMPLEMENTATION

To use a database specified by the user as a host-language variable in a CONNECT
statement in multi-database programs, follow these steps:
1. Declare a database handle using the following SET DATABASE syntax:
EXEC SQL
SET DATABASE handle = COMPILETIME ’dbname’;

Here, handle is a hard-coded database handle supplied by the programmer, dbname

is a quoted, hard-coded database name used by gpre during preprocessing.
2. Prompt the user for a database to open.
3. Store the database name entered by the user in a host-language variable.
4. Use the handle to open the database, associating the host-language variable
with the handle using the following CONNECT syntax:
EXEC SQL
CONNECT :variable AS handle;

The following C code illustrates these steps:

. . .
char fname[125];
. . .
EXEC SQL
SET DATABASE DB1 = ’employee.gdb’;
printf("Enter the desired database name, including node
and path):\n");
gets(fname);
EXEC SQL
CONNECT :fname AS DB1;
. . .

EMBEDDED SQL GUIDE 43

 CHAPTER 3 WORKING WITH DATABASES

In this example, SET DATABASE provides a hard-coded database file name for
preprocessing with gpre. When a user runs the program, the database specified in the
variable, fname, is used instead.

4 Using a hard-coded database names

IN SINGE-DATABASE PROGRAMS
In a single-database program that omits SET DATABASE, CONNECT must contain a
hard-coded, quoted file name in the following format:
EXEC SQL
CONNECT ’[host[path]]filename’;

host is required only if a program and a database file it uses reside on different nodes.
Similarly, path is required only if the database file does not reside in the current working
directory. For example, the following CONNECT statement contains a hard-coded file name
that includes both a Unix host name and a path name:
EXEC SQL
CONNECT ’valdez:usr/interbase/examples/employee.gdb’;

Note Host syntax is specific to each server platform.

IMPORTANT A program that accesses multiple databases cannot use this form of CONNECT.

IN MULTI-DATABASE PROGRAMS
A program that accesses multiple databases must declare handles for each of them in
separate SET DATABASE statements. These handles must be used in subsequent CONNECT
statements to identify specific databases to open:
. . .
EXEC SQL
SET DATABASE DB1 = ’employee.gdb’;
EXEC SQL
SET DATABASE DB2 = ’employee2.gdb’;
EXEC SQL
CONNECT DB1;
EXEC SQL
CONNECT DB2;
. . .

Later, when the program closes these databases, the database handles are no longer in
use. These handles can be reassigned to other databases by hard-coding a file name in a
subsequent CONNECT statement. For example,

44 INTERBASE 6
OPENING A DATABASE

. . .
EXEC SQL
DISCONNECT DB1, DB2;
EXEC SQL
CONNECT ’project.gdb’ AS DB1;
. . .

Additional CONNECT syntax

CONNECT supports several formats for opening databases to provide programming
flexibility. The following table outlines each possible syntax, provides descriptions and
examples, and indicates whether CONNECT can be used in programs that access single or
multiple databases:

Single Multiple
Syntax Description Example access access
CONNECT ‘dbfile’; Opens a single, hard-coded database file, EXEC SQL Yes No
dbfile. CONNECT ‘employee.gdb’;

CONNECT handle; Opens the database file associated with a EXEC SQL Yes Yes
previously declared database handle. This is CONNECT EMP;
the preferred CONNECT syntax.
CONNECT ‘dbfile’ AS Opens a hard-coded database file, dbfile, and EXEC SQL Yes Yes
handle; assigns a previously declared database handle CONNECT ‘employee.gdb’
to it. AS EMP;
CONNECT :varname AS Opens the database file stored in the EXEC SQL Yes Yes
handle; host-language variable, varname, and assigns a CONNECT :fname AS EMP;
previously declared database handle to it.
TABLE 3.1 CONNECT syntax summary

For a complete discussion of CONNECT syntax and its uses, see the Language Reference.

EMBEDDED SQL GUIDE 45

 CHAPTER 3 WORKING WITH DATABASES

Attaching to multiple databases

CONNECT can attach to multiple databases. To open all databases specified in previous SET
DATABASE statements, use either of the following CONNECT syntax options:
EXEC SQL
CONNECT ALL;

EXEC SQL
CONNECT DEFAULT;

CONNECT can also attach to a specified list of databases. Separate each database request
from others with commas. For example, the following statement opens two databases
specified by their handles:
EXEC SQL
CONNECT DB1, DB2;

The next statement opens two hard-coded database files and also assigns them to
previously declared handles:
EXEC SQL
CONNECT ’employee.gdb’ AS DB1, ’employee2.gdb’ AS DB2;

Tip Opening multiple databases with a single CONNECT is most effective when a program’s
database access is simple and clear. In complex programs that open and close several
databases, that substitute database names with host-language variables, or that assign
multiple handles to the same database, use separate CONNECT statements to make
program code easier to read, debug, and modify.

Handling CONNECT errors

The WHENEVER statement should be used to trap and handle runtime errors that occur
during database declaration. The following C code fragment illustrates an error-handling
routine that displays error messages and ends the program in an orderly fashion:
. . .
EXEC SQL
WHENEVER SQLERROR
GOTO error_exit;
. . .

46 INTERBASE 6
OPENING A DATABASE

:error_exit
isc_print_sqlerr(sqlcode, status_vector);
EXEC SQL
DISCONNECT ALL;
exit(1);
. . .

For a complete discussion of SQL error handling, see Chapter 12, “Error Handling and
Recovery.”

Setting database cache buffers

Besides opening a database, CONNECT can set the number of cache buffers assigned to a
database for that connection. When a program establishes a connection to a database,
InterBase allocates system memory to use as a private buffer. The buffers are used to store
accessed database pages to speed performance. The number of buffers assigned for a
program determine how many simultaneous database pages it can have access to in the
memory pool. Buffers remain assigned until a program finishes with a database.
The default number of database cache buffers assigned to a database is 256. This default
can be changed either for a specific database or for an entire server.
g Use the gfix utility to set a new default cache buffer size for a database. See the Operations
Guide for more information about setting database buffer size with gfix.
g Change the value of DATABASE_CACHE_PAGES in the InterBase configuration file to change
the default cache buffer size on a server-wide basis. Use this option with care, since it
makes it easy to overuse memory or create unusably small caches.

4 Setting individual database buffers

For programs that access or change many rows in many databases, performance can
sometimes be improved by increasing the number of buffers. The maximum number of
buffers allowed is system dependent.
g Use the CACHE n parameter with CONNECT to change the number of buffers assigned to a
database for the duration of the connection, where n is the number of buffers to reserve.
To set the number of buffers for an individual database, place CACHE n after the database
name. The following CONNECT specifies 500 buffers for the database pointed to by the
EMP handle:
EXEC SQL
CONNECT EMP CACHE 500;

EMBEDDED SQL GUIDE 47

 CHAPTER 3 WORKING WITH DATABASES

Note If you specify a buffer size that is less than the smallest one currently in use for the
database, the request is ignored.
The next statement opens two databases, TEST and EMP. Because CACHE is not specified
for TEST, its buffers default to 256. EMP is opened with the CACHE clause specifying 400
buffers:
EXEC SQL
CONNECT TEST, EMP CACHE 400;

4 Specifying buffers for all databases

To specify the same number of buffers for all databases, use CONNECT ALL with the
CACHE n parameter. For example, the following statements connect to two databases, EMP,
and EMP2, and allot 400 buffers to each of them:
. . .
EXEC SQL
SET DATABASE EMP = ’employee.gdb’;
EXEC SQL
SET DATABASE EMP2 = ’test.gdb’;
EXEC SQL
CONNECT ALL CACHE 400;
. . .

The same effect can be achieved by specifying the same amount of cache for individual
databases:
. . .
EXEC SQL
CONNECT EMP CACHE 400, TEST CACHE 400;
. . .

Accessing an open database

Once a database is connected, its tables can be accessed as follows:
g One database can be accessed in a single transaction.
g One database can be accessed in multiple transactions.
g Multiple databases can be accessed in a single transaction.
g Multiple databases can be accessed in multiple transactions.
For general information about using transactions, see Chapter 4, “Working with
Transactions.”

48 INTERBASE 6
DIFFERENTIATING TABLE NAMES

Differentiating table names

In SQL, using multiple databases in transactions sometimes requires extra precautions to
ensure intended behavior. When two or more databases have tables that share the same
name, a database handle must be prefixed to those table names to differentiate them from
one another in transactions.
A table name differentiated by a database handle takes the form:
handle.table

For example, the following cursor declaration accesses an EMPLOYEE table in TEST, and
another EMPLOYEE table in EMP. TEST and EMP are used as prefixes to indicate which
EMPLOYEE table should be referenced:
. . .
EXEC SQL
DECLARE IDMATCH CURSOR FOR
SELECT TESTNO INTO :matchid FROM TEST.EMPLOYEE
WHERE (SELECT EMPNO FROM EMP.EMPLOYEE WHERE EMPNO = TESTNO);
. . .

Note DSQL does not support access to multiple databases in a single statement.

Closing a database
When a program is finished with a database, the database should be closed. In SQL, a
database can be closed in either of the following ways:
g Issue a DISCONNECT to detach a database and close files.
g Append a RELEASE option to a COMMIT or ROLLBACK to disconnect from a database and
close files.
DISCONNECT, COMMIT RELEASE, and ROLLBACK RELEASE perform the following tasks:
g Close open database files.
g Disconnect from remote database connections.
g Release the memory that holds database metadata descriptions and InterBase
engine-compiled requests.
Note Closing databases with DISCONNECT is preferred for compatibility with the SQL-92
standard. Do not close a database until it is no longer needed. Once closed, a database
must be reopened, and its resources reallocated, before it can be used again.

EMBEDDED SQL GUIDE 49

 CHAPTER 3 WORKING WITH DATABASES

With DISCONNECT
To close all open databases by disconnecting from them, use the following DISCONNECT
syntax:
EXEC SQL
DISCONNECT {ALL | DEFAULT};

For example, each of the following statements closes all open databases in a
program:
EXEC SQL
DISCONNECT ALL;
EXEC SQL
DISCONNECT DEFAULT;

To close specific databases, specify their handles as comma-delimited parameters, using

the following syntax:
EXEC SQL
DISCONNECT handle [, handle ...];

For example, the following statement disconnects from two databases:

EXEC SQL
DISCONNECT DB1, DB2;

Note A database should not be closed until all transactions are finished with it, or it must
be reopened and its resources reallocated.

With COMMIT and ROLLBACK

To close all open databases when you COMMIT or ROLLBACK, use the following syntax:
EXEC SQL
{COMMIT | ROLLBACK} RELEASE;

For example, the following COMMIT closes all open databases:

EXEC SQL
COMMIT RELEASE;

To close specific databases, provide their handles as parameters following the RELEASE
option with COMMIT or ROLLBACK, using the following syntax:
EXEC SQL
COMMIT | ROLLBACK RELEASE handle [, handle ...];

50 INTERBASE 6
CLOSING A DATABASE

In the following example, the ROLLBACK statement closes two databases:

EXEC SQL
ROLLBACK RELEASE DB1, DB2;

EMBEDDED SQL GUIDE 51

 CHAPTER 3 WORKING WITH DATABASES

52 INTERBASE 6
 CHAPTER

Working with
Chapter4
4
Transactions

All SQL data definition and data manipulation statements take place within the context
of a transaction, a set of SQL statements that works to carry out a single task. This chapter
explains how to open, control, and close transactions using the following SQL transaction
management statements:

EMBEDDED SQL GUIDE 53

 CHAPTER 4 WORKING WITH TRANSACTIONS

Statement Purpose
SET TRANSACTION Starts a transaction, assigns it a name, and specifies its behavior. The following
behaviors can be specified:
Access mode describes the actions a transaction’s statements can perform.
Lock resolution describes how a transaction should react if a lock conflict occurs.
Isolation level describes the view of the database given a transaction as it relates to
actions performed by other simultaneously occurring transactions.
Table reservation, an optional list of tables to lock for access at the start of the
transaction rather than at the time of explicit reads or writes.
Database specification, an optional list limiting the open databases to which a
transaction may have access.
COMMIT Saves a transaction’s changes to the database and ends the transaction.
ROLLBACK Undoes a transaction’s changes before they have been committed to the database,
and ends the transaction.
TABLE 4.1 SQL transaction management statements

Transaction management statements define the beginning and end of a transaction.

They also control its behavior and interaction with other simultaneously running
transactions that share access to the same data within and across applications.
There are two types of transactions in InterBase:
g GDS__TRANS is a default transaction that InterBase uses when it encounters a statement
requiring a transaction without first finding a SET TRANSACTION statement. A default
behavior is defined for GDS__TRANS, but it can be changed by starting the default
transaction with SET TRANSACTION and specifying alternative behavior as parameters.
Treat GDS__TRANS as a global variable of type isc_tr_handle.
Note When using the default transaction without explicitly starting it with SET
TRANSACTION, applications must be preprocessed without the gpre -m switch.
g Named transactions are always started with SET TRANSACTION statements. These
statements provide unique names for each transaction, and usually include parameters
that specify a transaction’s behavior.
Except for naming conventions and use in multi-transaction programs, both the default
and named transactions offer the same control over transactions. SET TRANSACTION has
optional parameters for specifying access mode, lock resolution, and isolation level.

54 INTERBASE 6
STARTING THE DEFAULT TRANSACTION

For more information about gpre, see Chapter 14, “Preprocessing, Compiling,
and Linking.” For more information about transaction behavior, see “Specifying SET
TRANSACTION behavior” on page 60.

Starting the default transaction

If a transaction is started without a specified behavior, the following default behavior is
used:
READ WRITE WAIT ISOLATION LEVEL SNAPSHOT

The default transaction is especially useful for programs that use only a single
transaction. It is automatically started in programs that require a transaction context
where none is explicitly provided. It can also be explicitly started in a program with SET
TRANSACTION.
To learn more about transaction behavior, see “Starting the default transaction” on
page 55.

Starting without SET TRANSACTION

Simple, single transaction programs can omit SET TRANSACTION. The following program
fragment issues a SELECT statement without starting a transaction:
. . .
EXEC SQL
SELECT * FROM CITIES
WHERE POPULATION > 4000000
ORDER BY POPULATION, CITY;
. . .

A programmer need only start the default transaction explicitly in a single transaction
program to modify its operating characteristics or when writing a DSQL application that
is preprocessed with the gpre -m switch.
During preprocessing, when gpre encounters a statement, such as SELECT, that requires a
transaction context without first finding a SET TRANSACTION statement, it automatically
generates a default transaction as long as the -m switch is not specified. A default
transaction started by gpre uses a predefined, or default, behavior that dictates how the
transaction interacts with other simultaneous transactions attempting to access the same
data.

EMBEDDED SQL GUIDE 55

 CHAPTER 4 WORKING WITH TRANSACTIONS

IMPORTANT DSQL programs should be preprocessed with the gpre -m switch if they start a transaction
through DSQL. In this mode, gpre does not generate the default transaction as needed,
but instead reports an error if there is no transaction.
For more information about transaction behaviors that can be modified, see “Specifying
SET TRANSACTION behavior” on page 60. For more information about using the gpre
-m switch, see Chapter 14, “Preprocessing, Compiling, and Linking.”

Starting with SET TRANSACTION

SET TRANSACTION issued without parameters starts the default transaction, GDS__TRANS,
with the following default behavior:
READ WRITE WAIT ISOLATION LEVEL SNAPSHOT

The following table summarizes these settings:

Parameter Setting Purpose

Access mode READ WRITE Access mode. This transaction can select, insert, update, and
delete data.
Lock resolution WAIT Lock resolution. This transaction waits for locked tables and rows
to be released to see if it can then update them before reporting
a lock conflict.
Isolation level ISOLATION LEVEL SNAPSHOT This transaction receives a stable, unchanging view of the
database as it is at the moment the transaction starts; it never
sees changes made to the database by other active transactions.
TABLE 4.2 Default transaction default behavior

Note Explicitly starting the default transaction is good programming practice. It makes
a program’s source code easier to understand.
The following statements are equivalent. They both start the default transaction with the
default behavior.
EXEC SQL
SET TRANSACTION;
EXEC SQL
SET TRANSACTION NAME gds__trans READ WRITE WAIT ISOLATION LEVEL
SNAPSHOT;

56 INTERBASE 6
STARTING A NAMED TRANSACTION

To start the default transaction, but change its characteristics, SET TRANSACTION must be
used to specify those characteristics that differ from the default. Characteristics that do
not differ from the default can be omitted. For example, the following statement starts
the default transaction for READ ONLY access, WAIT lock resolution, and ISOLATION LEVEL
SNAPSHOT:
EXEC SQL
SET TRANSACTION READ ONLY;

As this example illustrates, the NAME clause can be omitted when starting the default
transaction.

IMPORTANT In DSQL, changing the characteristics of the default transaction is accomplished as with
PREPARE and EXECUTE in a manner similar to the one described, but the program must
be preprocessed using the gpre -m switch.
For more information about preprocessing programs with the -m switch, see Chapter 14,
“Preprocessing, Compiling, and Linking.” For more information about transaction
behavior and modification, see “Specifying SET TRANSACTION behavior” on page 60.

Starting a named transaction

A single application can start simultaneous transactions. InterBase extends transaction
management and data manipulation statements to support transaction names, unique
identifiers that specify which transaction controls a given statement among those
transactions that are active.
Transaction names must be used to distinguish one transaction from another in programs
that use two or more transactions at a time. Each transaction started while other
transactions are active requires a unique name and its own SET TRANSACTION statement.
SET TRANSACTION can include optional parameters that modify a transaction’s behavior.
There are four steps for using transaction names in a program:
1. Declare a unique host-language variable for each transaction name. In C and
C++, transaction names should be declared as long pointers.
2. Initialize each transaction name to zero.
3. Use SET TRANSACTION to start each transaction using an available transaction
name.
4. Include the transaction name in subsequent transaction management and
data manipulation statements that should be controlled by a specified
transaction.

EMBEDDED SQL GUIDE 57

 CHAPTER 4 WORKING WITH TRANSACTIONS

IMPORTANT Using named transactions in dynamic SQL statements is somewhat different. For
information about named transactions in DSQL, see “Working with multiple
transactions in DSQL” on page 83.
For additional information about creating multiple transaction programs, see “Working
with multiple transactions” on page 79.

Naming transactions
A transaction name is a programmer-supplied variable that distinguishes one transaction
from another in SQL statements. If transaction names are not used in SQL statements that
control transactions and manipulate data, then those statements operate only on the
default transaction, GDS__TRANS.
The following C code declares and initializes two transaction names using the
isc_tr_handle datatype. It then starts those transactions in SET TRANSACTION statements.
. . .
EXEC SQL
BEGIN DECLARE SECTION;
isc_tr_handle t1, t2; /* declare transaction names */
EXEC SQL
END DECLARE SECTION;
. . .
t1 = t2 = (isc_tr_handle) NULL; /* initialize names to zero */
. . .
EXEC SQL
SET TRANSACTION NAME t1; /* start trans. w. default behavior */
EXEC SQL
SET TRANSACTION NAME t2; /* start trans2. w. default behavior */
. . .

Each of these steps is fully described in the following sections.

A transaction name can be included as an optional parameter in any data manipulation
and transaction management statement. In multi-transaction programs, omitting a
transaction name causes a statement to be executed for the default transaction,
GDS__TRANS.
For more information about using transaction names with data manipulation statements,
see Chapter 6, “Working with Data.”

58 INTERBASE 6
STARTING A NAMED TRANSACTION

4 Declaring transaction names

Transaction names must be declared before they can be used. A name is declared as a
host-language pointer. In C and C++, transaction names should be declared as long
pointers.
The following code illustrates how to declare two transaction names:
EXEC SQL
BEGIN DECLARE SECTION;
isc_tr_handle t1;
isc_tr_handle t2;
EXEC SQL
END DECLARE SECTION;

Note In this example, the transaction declaration occurs within an SQL section
declaration. While InterBase does not require that host-language variables occur within
a section declaration, putting them there guarantees compatibility with other SQL
implementations that do require section declarations.
Transaction names are usually declared globally at the module level. If a transaction
name is declared locally, ensure that:
g The transaction using the name is completely contained within the function where the
name is declared. Include an error-handling routine to roll back transactions when errors
occur. ROLLBACK releases a transaction name, and sets its value to NULL.
g The transaction name is not used outside the function where it is declared.
To reference a transaction name declared in another module, provide an external
declaration for it. For example, in C, the external declaration for t1 and t2 might be as
follows:
EXEC SQL
BEGIN DECLARE SECTION;
extern isc_tr_handle t1, t2;
EXEC SQL
END DECLARE SECTION;

4 Initializing transaction names

Once transaction names are declared, they should be initialized to zero before being used
for the first time. The following C code illustrates how to set a starting value for two
declared transaction names:
/* initialize transaction names to zero */
t1 = t2 = (isc_tr_handle) NULL;

EMBEDDED SQL GUIDE 59

 CHAPTER 4 WORKING WITH TRANSACTIONS

Once a transaction name is declared and initialized, it can be used to:

g Start and name a transaction. Using a transaction name for all transactions except for the
default transaction is required if a program runs multiple, simultaneous transactions.
g Specify which transactions control data manipulation statements. Transaction names are
required in multi-transaction programs, unless a statement affects only the default
transaction.
g Commit or roll back specific transactions in a multi-transaction program.

Specifying SET TRANSACTION behavior

Use SET TRANSACTION to start a named transaction, and optionally specify its behavior.
The syntax for starting a named transaction using default behavior is:
SET TRANSACTION NAME name;

For a summary of the default behavior for a transaction started without specifying
behavior parameters, see table 4.2 on page 56. The following statements are equivalent:
they both start the transaction named t1, using default transaction behavior.
EXEC SQL
SET TRANSACTION NAME t1;
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE WAIT ISOLATION LEVEL SNAPSHOT;

60 INTERBASE 6
STARTING A NAMED TRANSACTION

The following table lists the optional SET TRANSACTION parameters for specifying the
behavior of the default transaction:

Parameter Setting Purpose

Access Mode READ ONLY or READ WRITE Describes the type of access this transaction is
permitted for a table. For more information about
access mode, see “Access mode” on page 62.
Lock WAIT or NO WAIT Specifies what happens when this transaction
Resolution encounters a locked row during an update or delete. It
either waits for the lock to be released so it can attempt
to complete its actions, or it returns an immediate lock
conflict error message. For more information about lock
resolution, see “Lock resolution” on page 69.
Isolation Level SNAPSHOT provides a view of the database at Determines this transaction’s interaction with other
the moment this transaction starts, but simultaneous transactions attempting to access the
prevents viewing changes made by other same tables.
active transactions. READ COMMITTED isolation level also enables a user to
SNAPSHOT TABLE STABILITY prevents other specify which version of a row it can read. There are two
transactions from making changes to tables options:
that this transaction is reading and • RECORD_VERSION: the transaction immediately reads
updating, but permits them to read rows in the latest committed version of a requested row, even
the table. if a more recent uncommitted version also resides on
READ COMMITTED reads the most recently disk.
committed version of a row during updates • NO RECORD_VERSION: if an uncommitted version of the
and deletions, and allows this transaction to requested row is present and WAIT lock resolution is
make changes if there is no update conflict specified, the transaction waits until the committed
with other transactions. version of the row is also the latest version; if NO WAIT is
specified, the transaction immediately returns an error
(“deadlock”) if the committed version is not the most
recent version.
Table RESERVING Specifies a subset of available tables to lock immediately
Reservation for this transaction to access.
Database USING Specifies a subset of available databases that this
Specification transaction can access; it cannot access any other
databases. The purpose of this option is to reduce the
amount of system resources used by this transaction.
Note: USING is not available in DSQL.
TABLE 4.3 SET TRANSACTION parameters

EMBEDDED SQL GUIDE 61

 CHAPTER 4 WORKING WITH TRANSACTIONS

The complete syntax of SET TRANSACTION is:

EXEC SQL
SET TRANSACTION [NAME name]
[READ WRITE| READ ONLY]
[WAIT | NO WAIT]
[[ISOLATION LEVEL] {SNAPSHOT [TABLE STABILITY]
| READ COMMITTED [[NO] RECORD_VERSION]}]
[RESERVING <reserving_clause>
| USING dbhandle [, dbhandle ...]];

<reserving_clause> = table [, table ...]

[FOR [SHARED | PROTECTED] {READ | WRITE}] [, <reserving_clause>]

Transaction options are fully described in the following sections.

4 Access mode
The access mode parameter specifies the type of access a transaction has for the tables it
uses. There are two possible settings:
g READ ONLY specifies that a transaction can select data from a table, but cannot insert,
update, or delete table data.
g READ WRITE specifies that a transaction can select, insert, update, and delete table data.
This is the default setting if none is specified.
InterBase assumes that most transactions both read and write data. When starting a
transaction for reading and writing, READ WRITE can be omitted from SET TRANSACTION
statement. For example, the following statements start a transaction, t1, for READ WRITE
access:
EXEC SQL
SET TRANSACTION NAME t1;
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE;

Tip It is good programming practice to specify a transaction’s access mode, even when it is
READ WRITE. It makes an application’s source code easier to read and debug because the
program’s intentions are clearly spelled out.
Start a transaction for READ ONLY access when you only need to read data. READ ONLY
must be specified. For example, the following statement starts a transaction, t1, for
read-only access:
EXEC SQL
SET TRANSACTION NAME t1 READ ONLY;

62 INTERBASE 6
STARTING A NAMED TRANSACTION

4 Isolation level
The isolation level parameter specifies the control a transaction exercises over table
access. It determines the:
g View of a database the transaction can see.
g Table access allowed to this and other simultaneous transactions.
The following table describes the three isolation levels supported by InterBase:

Isolation level Purpose

SNAPSHOT Provides a stable, committed view of the database at the time the transaction starts; this
is the default isolation level. Other simultaneous transactions can UPDATE and INSERT rows,
but this transaction cannot see those changes. For updated rows, this transaction sees
versions of those rows as they existed at the start of the transaction. If this transaction
attempts to update or delete rows changed by another transaction, an update conflict is
reported.
SNAPSHOT TABLE STABILITY Provides a transaction sole insert, update, and delete access to the tables it uses. Other
simultaneous transactions may still be able to select rows from those tables.
READ COMMITTED Enables the transaction to see all committed data in the database, and to update rows
updated and committed by other simultaneous transactions without causing lost update
problems.
TABLE 4.4 ISOLATION LEVEL options

The isolation level for most transactions should be either SNAPSHOT or READ COMMITTED.
These levels enable simultaneous transactions to select, insert, update, and delete data in
shared databases, and they minimize the chance for lock conflicts. Lock conflicts occur
in two situations:
g When a transaction attempts to update a row already updated or deleted by another
transaction. A row updated by a transaction is effectively locked for update to all other
transactions until the controlling transaction commits or rolls back. READ COMMITTED
transactions can read and update rows updated by simultaneous transactions after they
commit.
g When a transaction attempts to insert, update, or delete a row in a table locked by another
transaction with an isolation level of SNAPSHOT TABLE STABILITY. SNAPSHOT TABLE STABILITY
locks entire tables for write access, although concurrent reads by other SNAPSHOT and
READ COMMITTED transactions are permitted.

EMBEDDED SQL GUIDE 63

 CHAPTER 4 WORKING WITH TRANSACTIONS

Using SNAPSHOT TABLE STABILITY guarantees that only a single transaction can make
changes to tables, but increases the chance of lock conflicts where there are simultaneous
transactions attempting to access the same tables. For more information about the
likelihood of lock conflicts, see “Isolation level interactions” on page 68.

COMPARING SNAPSHOT, READ COMMITTED,

AND SNAPSHOT TABLE STABILITY
There are five classic problems all transaction management statements must address:
g Lost updates, which can occur if an update is overwritten by a simultaneous transaction
unaware of the last updates made by another transaction.
g Dirty reads, which can occur if the system allows one transaction to select uncommitted
changes made by another transaction.
g Non-reproducible reads, which can occur if one transaction is allowed to update or delete
rows that are repeatedly selected by another transaction. READ COMMITTED transactions
permit non-reproducible reads by design, since they can see committed deletes made by
other transactions.
g Phantom rows, which can occur if one transaction is allowed to select some, but not all,
new rows written by another transaction. READ COMMITTED transactions do not prevent
phantom rows.
g Update side effects, which can occur when row values are interdependent, and their
dependencies are not adequately protected or enforced by locking, triggers, or integrity
constraints. These conflicts occur when two or more simultaneous transactions randomly
and repeatedly access and update the same data; such transactions are called interleaved
transactions.

64 INTERBASE 6
STARTING A NAMED TRANSACTION

Except as noted, all three InterBase isolation levels control these problems. The following
table summarizes how a transaction with a particular isolation level controls access to its
data for other simultaneous transactions:

Problem SNAPSHOT, READ COMMITTED SNAPSHOT TABLE STABILITY

Lost updates Other transactions cannot update rows Other transactions cannot update tables
already updated by this transaction. controlled by this transaction.
Dirty reads Other SNAPSHOT transactions can only read a Other transactions cannot access tables
previous version of a row updated by this updated by this transaction.
transaction.
Other READ COMMITTED transactions can only
read a previous version, or committed
updates.
Non-reproducible SNAPSHOT and SNAPSHOT TABLE STABILITY SNAPSHOT and SNAPSHOT TABLE STABILITY
reads transactions can only read versions of rows transactions can only read versions of rows
committed when they started. committed when they started.
READ COMMITTED transactions must expect that Other transactions cannot access tables
reads cannot be reproduced. updated by this transaction.
Phantom rows READ COMMITTED transactions may encounter Other transactions cannot access tables
phantom rows. controlled by this transaction.
Update side effects Other SNAPSHOT transactions can only read a Other transactions cannot update tables
previous version of a row updated by this controlled by this transaction.
transaction. Use triggers and integrity constraints to avoid
Other READ COMMITTED transactions can only any problems with interleaved transactions.
read a previous version, or committed
updates.
Use triggers and integrity constraints to try to
avoid any problems with interleaved
transactions.
TABLE 4.5 InterBase management of classic transaction conflicts

CHOOSING BETWEEN SNAPSHOT AND READ COMMITTED

The choice between SNAPSHOT and READ COMMITTED isolation levels depends on an
application’s needs. SNAPSHOT is the default InterBase isolation level. READ COMMITTED
duplicates SNAPSHOT behavior, but can read subsequent changes committed by other
transactions. In many cases, using READ COMMITTED reduces data contention.

EMBEDDED SQL GUIDE 65

 CHAPTER 4 WORKING WITH TRANSACTIONS

SNAPSHOT transactions receive a stable view of a database as it exists the moment the
transactions start. READ COMMITTED transactions can see the latest committed versions of
rows. Both types of transactions can use SELECT statements unless they encounter the
following conditions:
g Table locked by SNAPSHOT TABLE STABILITY transaction for UPDATE.
g Uncommitted inserts made by other simultaneous transactions. In this case, a SELECT is
allowed, but changes cannot be seen.
READ COMMITTED transactions can read the latest committed version of rows. A SNAPSHOT
transaction can read only a prior version of the row as it existed before the update
occurred.
SNAPHOT and READ COMMITTED transactions with READ WRITE access can use INSERT,
UPDATE, and DELETE unless they encounter tables locked by SNAPSHOT TABLE STABILITY
transactions.
SNAPSHOT transactions cannot update or delete rows previously updated or deleted and
then committed by other simultaneous transactions. Attempting to update a row
previously updated or deleted by another transaction results in an update conflict error.
A READ COMMITTED READ WRITE transaction can read changes committed by other
transactions, and subsequently update those changed rows.
Occasional update conflicts may occur when simultaneous SNAPSHOT and READ
COMMITTED transactions attempt to update the same row at the same time. When update
conflicts occur, expect the following behavior:
g For mass or searched updates, updates where a single UPDATE modifies multiple rows in
a table, all updates are undone on conflict. The UPDATE can be retried. For READ
COMMITTED transactions, the NO RECORD_VERSION option can be used to narrow the
window between reads and updates or deletes. For more information, see “Starting a
transaction with READ COMMITTED isolation level” on page 67.
g For cursor or positioned updates, where rows are retrieved and updated from an active
set one row at a time, only a single update is undone. To retry the update, the cursor must
be closed, then reopened, and updates resumed at the point of previous conflict.
For more information about UPDATE through cursors, see Chapter 6, “Working with
Data.”

STARTING A TRANSACTION WITH SNAPSHOT ISOLATION LEVEL

InterBase assumes that the default isolation level for transactions is SNAPSHOT. Therefore,
SNAPSHOT need not be specified in SET TRANSACTION to set the isolation level. For
example, the following statements are equivalent. They both start a transaction, t1, for
READ WRITE access and set isolation level to SNAPSHOT.

66 INTERBASE 6
STARTING A NAMED TRANSACTION

EXEC SQL
SET TRANSACTION NAME t1;
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE SNAPSHOT;

When an isolation level is specified, it must follow the access and lock resolution modes.

Tip It is good programming practice to specify a transaction’s isolation level, even when it is
SNAPSHOT. It makes an application’s source code easier to read and debug because the
program’s intentions are clearly spelled out.

STARTING A TRANSACTION WITH READ COMMITTED ISOLATION LEVEL

To start a READ COMMITTED transaction, the isolation level must be specified. For example,
the following statement starts a named transaction, t1, for READ WRITE access and sets
isolation level to READ COMMITTED:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE READ COMMITTED;

Isolation level always follows access mode. If the access mode is omitted, isolation level
is the first parameter to follow the transaction name.
READ COMMITTED supports mutually exclusive optional parameters, RECORD_VERSION and
NO RECORD_VERSION, which determine the READ COMMITTED behavior when it encounters
a row where the latest version of that row is uncommitted:
g RECORD_VERSION specifies that the transaction immediately reads the latest committed
version of a row, even if a more recent uncommitted version also resides on disk.
g NO RECORD_VERSION, the default, specifies that the transaction can only read the latest
version of a requested row. If the WAIT lock resolution option is also specified, then the
transaction waits until the latest version of a row is committed or rolled back, and retries
its read. If the NO WAIT option is specified, the transaction returns an immediate
deadlock error.
Because NO RECORD_VERSION is the default behavior, it need not be specified with READ
COMITTED. For example, the following statements are equivalent. They start a named
transaction, t1, for READ WRITE access and set isolation level to READ COMMITTED NO
RECORD_VERSION.
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE READ COMMITTED;
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE READ COMMITTED
NO RECORD_VERSION;

EMBEDDED SQL GUIDE 67

 CHAPTER 4 WORKING WITH TRANSACTIONS

RECORD_VERSION must always be specified when it is used. For example, the following
statement starts a named transaction, t1, for READ WRITE access and sets isolation level to
READ COMMITTED RECORD_VERSION:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE READ COMMITTED
RECORD_VERSION;

STARTING A TRANSACTION WITH

SNAPSHOT TABLE STABILITY ISOLATION LEVEL
To start a SNAPSHOT TABLE STABILITY transaction, the isolation level must be specified. For
example, the following statement starts a named transaction, t1, for READ WRITE access
and sets isolation level to SNAPSHOT TABLE STABILITY:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE SNAPSHOT TABLE STABILITY;

Isolation level always follows the optional access mode and lock resolution parameters,
if they are present.

IMPORTANT Use SNAPSHOT TABLE STABILITY with care. In an environment where multiple transactions
share database access, SNAPSHOT TABLE STABILITY greatly increases the likelihood of lock
conflicts.

ISOLATION LEVEL INTERACTIONS

To determine the possibility for lock conflicts between two transactions accessing the
same database, each transaction’s isolation level and access mode must be considered.
The following table summarizes possible combinations.

SNAPSHOT or READ COMMITTED SNAPSHOT TABLE STABILITY

UPDATE SELECT UPDATE SELECT

SNAPSHOT or UPDATE Some simultaneous — Always conflicts Always conflicts

READ COMMITTED updates may conflict
SELECT — — — —
SNAPSHOT TABLE UPDATE Always conflicts — Always conflicts Always conflicts
STABILITY
SELECT Always conflicts — Always conflicts —
TABLE 4.6 Isolation level Interaction with SELECT and UPDATE

68 INTERBASE 6
STARTING A NAMED TRANSACTION

As this table illustrates, SNAPSHOT and READ COMMITTED transactions offer the least
chance for conflicts. For example, if t1 is a SNAPSHOT transaction with READ WRITE
access, and t2 is a READ COMMITTED transaction with READ WRITE access, t1 and t2 only
conflict when they attempt to update the same rows. If t1 and t2 have READ ONLY access,
they never conflict with any other transaction.
A SNAPSHOT TABLE STABILITY transaction with READ WRITE access is guaranteed that it
alone can update tables, but it conflicts with all other simultaneous transactions except
for SNAPSHOT and READ COMMITTED transactions running in READ ONLY mode. A SNAPSHOT
TABLE STABILITY transaction with READ ONLY access is compatible with any other read-only
transaction, but conflicts with any transaction that attempts to insert, update, or delete
data.

4 Lock resolution
The lock resolution parameter determines what happens when a transaction encounters
a lock conflict. There are two options:
g WAIT, the default, causes the transaction to wait until locked resources are released. Once
the locks are released, the transaction retries its operation.
g NO WAIT returns a lock conflict error without waiting for locks to be released.
Because WAIT is the default lock resolution, you don’t need to specify it in a SET
TRANSACTION statement. For example, the following statements are equivalent. They both
start a transaction, t1, for READ WRITE access, WAIT lock resolution, and READ COMMITTED
isolation level:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE READ COMMITTED;
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE WAIT READ COMMITTED;

To use NO WAIT, the lock resolution parameter must be specified. For example, the
following statement starts the named transaction, t1, for READ WRITE access, NO WAIT lock
resolution, and SNAPSHOT isolation level:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE NO WAIT READ SNAPSHOT;

When lock resolution is specified, it follows the optional access mode, and precedes the
optional isolation level parameter.

Tip It is good programming practice to specify a transaction’s lock resolution, even when it
is WAIT. It makes an application’s source code easier to read and debug because the
program’s intentions are clearly spelled out.

EMBEDDED SQL GUIDE 69

 CHAPTER 4 WORKING WITH TRANSACTIONS

4 RESERVING clause
The optional RESERVING clause enables transactions to guarantee themselves specific
levels of access to a subset of available tables at the expense of other simultaneous
transactions. Reservation takes place at the start of the transaction instead of only when
data manipulation statements require a particular level of access. RESERVING is only useful
in an environment where simultaneous transactions share database access. It has three
main purposes:
g To prevent possible deadlocks and update conflicts that can occur if locks are taken only
when actually needed (the default behavior).
g To provide for dependency locking, the locking of tables that may be affected by triggers
and integrity constraints. While explicit dependency locking is not required, it can assure
that update conflicts do not occur because of indirect table conflicts.
g To change the level of shared access for one or more individual tables in a transaction.
For example, a READ WRITE SNAPSHOT transaction may need exclusive update rights for a
single table, and could use the RESERVING clause to guarantee itself sole write access to
the table.

IMPORTANT A single SET TRANSACTION statement can contain either a RESERVING or a USING clause,
but not both. Use the SET TRANSACTION syntax to reserve tables for a transaction:
EXEC SQL
SET TRANSACTION [NAME name]
[READ WRITE| READ ONLY]
[WAIT | NO WAIT]
[[ISOLATION LEVEL] {SNAPSHOT [TABLE STABILITY]
| READ COMMITTED [[NO] RECORD_VERSION]}]
RESERVING <reserving_clause>;

<reserving_clause> = table [, table ...]

[FOR [SHARED | PROTECTED] {READ | WRITE}] [, <reserving_clause>]

70 INTERBASE 6
STARTING A NAMED TRANSACTION

Each table should only appear once in the RESERVING clause. Each table, or a list of tables
separated by commas, must be followed by a clause describing the type of reservation
requested. The following table lists these reservation options:

Reservation
option Purpose
PROTECTED READ Prevents other transactions from updating rows. All transactions can select
from the table.
PROTECTED WRITE Prevents other transactions from updating rows.
SNAPSHOT and READ COMMITTED transactions can select from the table, but only
this transaction can update rows.
SHARED READ Any transaction can select from this table. Any READ WRITE transaction can
update this table. This is the most liberal reservation mode.
SHARED WRITE Any SNAPSHOT or READ COMMITTED READ WRITE transaction can update this table.
Other SNAPSHOT and READ COMMITTED transactions can also select from this
table.
TABLE 4.7 Table reservation options for the RESERVING clause

The following statement starts a SNAPSHOT transaction, t1, for READ WRITE access, and
reserves a single table for PROTECTED WRITE access:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE WAIT SNAPSHOT
RESERVING EMPLOYEE FOR PROTECTED WRITE;

The next statement starts a READ COMMITTED transaction, t1, for READ WRITE access, and
reserves two tables, one for SHARED WRITE, and another for PROTECTED READ:
EXEC SQL
SET TRANSACTION NAME t1 READ WRITE WAIT READ COMMITTED
RESERVING EMPLOYEES FOR SHARED WRITE, EMP_PROJ
FOR PROTECTED READ;

SNAPSHOT and READ COMMITTED transactions use RESERVING to implement more restrictive
access to tables for other simultaneous transactions. SNAPSHOT TABLE STABILITY
transactions use RESERVING to reduce the likelihood of deadlock in critical situations.

EMBEDDED SQL GUIDE 71

 CHAPTER 4 WORKING WITH TRANSACTIONS

4 USING clause
Every time a transaction is started, InterBase reserves system resources for each database
currently attached for program access. In a multi-transaction, multi-database program,
the USING clause can be used to preserve system resources by restricting the number of
open databases to which a transaction has access. USING restricts a transaction’s access
to tables to a listed subset of all open databases using the following syntax:
EXEC SQL
SET TRANSACTION [NAME name]
[READ WRITE | READ ONLY]
[WAIT | NO WAIT]
[[ISOLATION LEVEL] {SNAPSHOT [TABLE STABILITY]
| READ COMMITTED [[NO] RECORD_VERSION]}]
USING dbhandle> [, dbhandle ...];

IMPORTANT A single SET TRANSACTION statement can contain either a USING or a RESERVING clause,
but not both.
The following C program fragment opens three databases, test.gdb, research.gdb, and
employee.gdb, assigning them to the database handles TEST, RESEARCH, and EMP,
respectively. Then it starts the default transaction and restricts its access to TEST and EMP:
. . .
EXEC SQL
SET DATABASE ATLAS = ’test.gdb’;
EXEC SQL
SET DATABASE RESEARCH = ’research.gdb’;
EXEC SQL
SET DATABASE EMP = ’employee.gdb’;
EXEC SQL
CONNECT TEST, RESEARCH, EMP; /* Open all databases */
EXEC SQL
SET TRANSACTION USING TEST, EMP;
. . .

Using transaction names in data statements

Once named transactions are started, use their names in INSERT, UPDATE, DELETE, and
OPEN statements to specify which transaction controls the statement. For example, the
following C code fragment declares two transaction handles, mytrans1, and mytrans2,
initializes them to zero, starts the transactions, and then uses the transaction names to
qualify the data manipulation statements that follow:

72 INTERBASE 6
USING TRANSACTION NAMES IN DATA STATEMENTS

. . .
EXEC SQL
BEGIN DECLARE SECTION;
long *mytrans1, *mytrans2;
char city[26];
EXEC SQL
END DECLARE SECTION;
mytrans1 = 0L;
mytrans2 = 0L;
. . .
EXEC SQL
SET DATABASE ATLAS = ’atlas.gdb’;
EXEC SQL
CONNECT;
EXEC SQL
DECLARE CITYLIST CURSOR FOR
SELECT CITY FROM CITIES
WHERE COUNTRY = ’Mexico’;
EXEC SQL
SET TRANSACTION NAME mytrans1;
EXEC SQL
SET TRANSACTION mytrans2 READ ONLY READ COMMITTED;
. . .
printf(’Mexican city to add to database: ’);
gets(city);
EXEC SQL
INSERT TRANSACTION mytrans1 INTO CITIES (CITY, COUNTRY)
VALUES :city, ’Mexico’;
EXEC SQL
COMMIT mytrans1;
EXEC SQL
OPEN TRANSACTION mytrans2 CITYLIST;
EXEC SQL
FETCH CITYLIST INTO :city;
while (!SQLCODE)
{
printf("%s\n", city);
EXEC SQL
FETCH CITYLIST INTO :city;
}
EXEC SQL
CLOSE CITYLIST;

EMBEDDED SQL GUIDE 73

 CHAPTER 4 WORKING WITH TRANSACTIONS

EXEC SQL
COMMIT;
EXEC SQL
DISCONNECT;
. . .

As this example illustrates, a transaction name cannot appear in a DECLARE CURSOR

statement. To use a name with a cursor declaration, include the transaction name in the
cursor’s OPEN statement. The transaction name is not required in subsequent FETCH and
CLOSE statements for that cursor.

Note The DSQL EXECUTE and EXECUTE IMMEDIATE statements also support transaction
names.
For more information about using transaction names with data manipulation statements,
see Chapter 6, “Working with Data.” For more information about transaction names
and the COMMIT statement, see “Using COMMIT” on page 75. For more information
about using transaction names with DSQL statements, see “Working with multiple
transactions in DSQL” on page 83.

Ending a transaction
When a transaction’s tasks are complete, or an error prevents a transaction from
completing, the transaction must be ended to set the database to a consistent state. There
are two statements that end transactions:
g COMMIT makes a transaction’s changes permanent in the database. It signals that a
transaction completed all its actions successfully.
g ROLLBACK undoes a transaction’s changes, returning the database to its previous state,
before the transaction started. ROLLBACK is typically used when one or more errors occur
that prevent a transaction from completing successfully.
Both COMMIT and ROLLBACK close the record streams associated with the transaction,
reinitialize the transaction name to zero, and release system resources allocated for the
transaction. Freed system resources are available for subsequent use by any application
or program.
COMMIT and ROLLBACK have additional benefits. They clearly indicate program logic and
intention, make a program easier to understand, and most importantly, assure that a
transaction’s changes are handled as intended by the programmer.

74 INTERBASE 6
ENDING A TRANSACTION

ROLLBACK is frequently used inside error-handling routines to clean up transactions when

errors occur. It can also be used to roll back a partially completed transaction prior to
retrying it, and it can be used to restore a database to its prior state if a program
encounters an unrecoverable error.

IMPORTANT If the program ends before a transaction ends, a transaction is automatically rolled back,
but databases are not closed. If a program ends without closing the database, data loss
or corruption is possible. Therefore, open databases should always be closed by issuing
explicit DISCONNECT, COMMIT RELEASE, or ROLLBACK RELEASE statements.
For more information about DISCONNECT, COMMIT RELEASE, and ROLLBACK RELEASE, see
Chapter 3, “Working with Databases.”

Using COMMIT
Use COMMIT to write transaction changes permanently to a database.
COMMIT closes the record streams associated with the transaction, resets the transaction
name to zero, and frees system resources assigned to the transaction for other uses. The
complete syntax for COMMIT is:
EXEC SQL
COMMIT [TRANSACTION name] [RETAIN [SNAPSHOT] | RELEASE dbhandle
[, dbhandle ...]]

For example, the following C code fragment contains a complete transaction. It gives all
employees who have worked since December 31, 1992, a 4.3% cost-of-living salary
increase. If all qualified employee records are successfully updated, the transaction is
committed, and the changes are actually applied to the database.
. . .
EXEC SQL
SET TRANSACTION SNAPSHOT TABLE STABILITY;
EXEC SQL
UPDATE EMPLOYEE
SET SALARY = SALARY * 1.043
WHERE HIRE_DATE < ’1-JAN-1993’;
EXEC SQL
COMMIT;
. . .

By default, COMMIT affects only the default transaction, GDS__TRANS. To commit another
transaction, use its transaction name as a parameter to COMMIT.

EMBEDDED SQL GUIDE 75

 CHAPTER 4 WORKING WITH TRANSACTIONS

Tip Even READ ONLY transactions that do not change a database should be ended with a
COMMIT rather than ROLLBACK. The database is not changed, but the overhead required
to start subsequent transactions is greatly reduced.

4 Specifying transaction names

To commit changes for transactions other than the default transaction, specify a
transaction name as a COMMIT parameter. For example, the following C code fragment
starts two transactions using names, and commits them:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
isc_tr_handle TR1, TR2;
EXEC SQL
END DECLARE SECTION;
TR1 = (isc_tr_handle) NULL;
TR2 = (isc_tr_handle) NULL;
. . .
EXEC SQL
SET TRANSACTION NAME TR1;
EXEC SQL
SET TRANSACTION NAME TR2;
. . .
/* do actual processsing here */
. . .
EXEC SQL
COMMIT TRANSACTION TR1;
EXEC SQL
COMMIT TRANSACTION TR2;
. . .

IMPORTANT In multi-transaction programs, transaction names must always be specified for COMMIT
except when committing the default transaction.

4 Committing without freeing a transaction

To write transaction changes to the database without releasing the current transaction
snapshot, use the RETAIN option with COMMIT. The COMMIT RETAIN statement commits
your work and opens a new transaction, preserving the old transaction’s snapshot. In a
busy multi-user environment, retaining the snapshot speeds up processing and uses
fewer system resources than closing and starting a new transaction for each action. The
disadvantage of using COMMIT RETAIN is that you do not see the pending transactions of
other users.

76 INTERBASE 6
ENDING A TRANSACTION

The syntax for the RETAIN option is as follows:

EXEC SQL
COMMIT [TRANSACTION name] RETAIN [SNAPSHOT];

Tip Developers who use Borland tools such as Delphi use this feature by specifying “soft
commits” in the BDE configuration.
For example, the following C code fragment updates the POPULATION column by
user-specified amounts for cities in the CITIES table that are in a country also specified by
the user. Each time a qualified row is updated, a COMMIT with the RETAIN option is issued,
preserving the current cursor status and system resources.
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char country[26], city[26], asciimult[10];
int multiplier;
long pop;
EXEC SQL
END DECLARE SECTION;
. . .
main ()
{
EXEC SQL
DECLARE CHANGEPOP CURSOR FOR
SELECT CITY, POPULATION
FROM CITIES
WHERE COUNTRY = :country;
printf("Enter country with city populations needing adjustment: ");
gets(country);
EXEC SQL
SET TRANSACTION;
EXEC SQL
OPEN CHANGEPOP;
EXEC SQL
FETCH CHANGEPOP INTO :city, :pop;
while(!SQLCODE)
{
printf("City: %s Population: %ld\n", city, pop);
printf("\nPercent change (100%% to -100%%:");
gets(asciimult);
multiplier = atoi(asciimult);
EXEC SQL

EMBEDDED SQL GUIDE 77

 CHAPTER 4 WORKING WITH TRANSACTIONS

UPDATE CITIES
SET POPULATION = POPULATION * (1 + :multiplier / 100)
WHERE CURRENT OF CHANGEPOP;
EXEC SQL
COMMIT RETAIN; /* commit changes, save current state */
EXEC SQL
FETCH CHANGEPOP INTO :city, :pop;
if (SQLCODE && (SQLCODE != 100))
{
isc_print_sqlerror(SQLCODE, isc_$status);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT;
exit(1);
}
}
EXEC SQL
COMMIT;
EXEC SQL
DISCONNECT;
}

Note If you execute a ROLLBACK after a COMMIT RETAIN, it rolls back only updates and
writes that occurred after the COMMIT RETAIN.

IMPORTANT In multi-transaction programs, a transaction name must be specified for COMMIT RETAIN,
except when retaining the state of the default transaction. For more information about
transaction names, see “Naming transactions” on page 58.

Using ROLLBACK
Use ROLLBACK to restore the database to its condition prior to the start of the transaction.
ROLLBACK also closes the record streams associated with the transaction, resets the
transaction name to zero, and frees system resources assigned to the transaction for other
uses. ROLLBACK typically appears in error-handling routines. The syntax for ROLLBACK is:
EXEC SQL
ROLLBACK [TRANSACTION name] [RELEASE [dbhandle [, dbhandle ...]]];

78 INTERBASE 6
WORKING WITH MULTIPLE TRANSACTIONS

For example, the following C code fragment contains a complete transaction that gives
all employees who have worked since December 31, 1992, a 4.3% cost-of-living salary
adjustment. If all qualified employee records are successfully updated, the transaction is
committed, and the changes are actually applied to the database. If an error occurs, all
changes made by the transaction are undone, and the database is restored to its condition
prior to the start of the transaction.
. . .
EXEC SQL
SET TRANSACTION SNAPSHOT TABLE STABILITY;
EXEC SQL
UPDATE EMPLOYEES
SET SALARY = SALARY * 1.043
WHERE HIRE_DATE < ’1-JAN-1993’;
if (SQLCODE && (SQLCODE != 100))
{
isc_print_sqlerror(SQLCODE, isc_$status);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT;
exit(1);
}
EXEC SQL
COMMIT;
EXEC SQL
DISCONNECT;
. . .

By default, ROLLBACK affects only the default transaction, GDS__TRANS. To roll back other
transactions, use their transaction names as parameters to
ROLLBACK.

Working with multiple transactions

Because InterBase provides support for transaction names, a program can use as many
transactions at once as necessary to carry out its work. Each simultaneous transaction in
a program requires its own name. A transaction’s name distinguishes it from other active
transactions. The name can also be used in data manipulation and transaction
management statements to specify which transaction controls the statement. For more
information about declaring and using transaction names, see “Starting a named
transaction” on page 57.

EMBEDDED SQL GUIDE 79

 CHAPTER 4 WORKING WITH TRANSACTIONS

There are four steps for using named transactions in a program:

1. Declare a unique host-language variable for each transaction name.
2. Initialize each transaction variable to zero.
3. Use SET TRANSACTION to start each transaction using an available transaction
name.
4. Use the transaction names as parameters in subsequent transaction
management and data manipulation statements that should be controlled by
a specified transaction.

The default transaction

In multi-transaction programs, it is good programming practice to supply a transaction
name for every transaction a program defines. One transaction in a multi-transaction
program can be the default transaction, GDS__TRANS. When the default transaction is used
in multi-transaction programs, it, too, should be started explicitly and referenced by
name in data manipulation statements.
If the transaction name is omitted from a transaction management or data manipulation
statement, InterBase assumes the statement affects the default transaction. If the default
transaction has not been explicitly started with a SET TRANSACTION statement, gpre inserts
a statement during preprocessing to start it.

IMPORTANT DSQL programs must be preprocessed with the gpre -m switch. In this mode, gpre does
not generate the default transaction automatically, but instead reports an error. DSQL
programs require that all transactions be explicitly started.

Using cursors
DECLARE CURSOR does not support transaction names. Instead, to associate a named
transaction with a cursor, include the transaction name as an optional parameter in the
cursor’s OPEN statement. A cursor can only be associated with a single transaction. For
example, the following statements declare a cursor, and open it, associating it with the
transaction, T1:
. . .
EXEC SQL
DECLARE S CURSOR FOR
SELECT COUNTRY, CUST_NO, SUM(QTY_ORDERED)
FROM SALES

80 INTERBASE 6
WORKING WITH MULTIPLE TRANSACTIONS

GROUP BY CUST_NO
WHERE COUNTRY = ’Mexico’;
EXEC SQL
SET TRANSACTION T1 READ ONLY READ COMMITTED;
. . .
EXEC SQL
OPEN TRANSACTION T1 S;
. . .

An OPEN statement without the optional transaction name parameter operates under
control of the default transaction, GDS__TRANS.
Once a named transaction is associated with a cursor, subsequent cursor statements
automatically operate under control of that transaction. Therefore, it does not support a
transaction name parameter. For example, the following statements illustrate a FETCH and
CLOSE for the S cursor after it is associated with the named transaction, t2:
. . .
EXEC SQL
OPEN TRANSACTION t2 S;
EXEC SQL
FETCH S INTO :country, :cust_no, :qty;
while (!SQLCODE)
{
printf("%s %d %d\n", country, cust_no, qty);
EXEC SQL
FETCH S INTO :country, :cust_no, :qty;
}
EXEC SQL
CLOSE S;
. . .
Multiple cursors can be controlled by a single transaction, or each transaction can control
a single cursor according to a program’s needs.

A multi-transaction example
The following C code illustrates the steps required to create a simple multi-transaction
program. It declares two transaction handles, mytrans1, and mytrans2, initializes them
to zero, starts the transactions, and then uses the transaction names to qualify the data
manipulation statements that follow. It also illustrates the use of a cursor with a named
transaction.
. . .

EMBEDDED SQL GUIDE 81

 CHAPTER 4 WORKING WITH TRANSACTIONS

EXEC SQL
BEGIN DECLARE SECTION;
long *mytrans1 = 0L, *mytrans2 = 0L;
char city[26];
EXEC SQL
END DECLARE SECTION;
. . .
EXEC SQL
DECLARE CITYLIST CURSOR FOR
SELECT CITY FROM CITIES
WHERE COUNTRY = ’Mexico’;
EXEC SQL
SET TRANSACTION NAME mytrans1;
EXEC SQL
SET TRANSACTION mytrans2 READ ONLY READ COMMITTED;
. . .
printf("Mexican city to add to database: ");
gets(city);
EXEC SQL
INSERT TRANSACTION mytrans1 INTO CITIES
VALUES :city, ’Mexico’, NULL, NULL, NULL, NULL;
EXEC SQL
COMMIT mytrans1;
EXEC SQL
OPEN TRANSACTION mytrans2 CITYLIST;
EXEC SQL
FETCH CITYLIST INTO :city;
while (!SQLCODE)
{
printf("%s\n", city);
EXEC SQL
FETCH CITYLIST INTO :city;
}
EXEC SQL
CLOSE CITYLIST;
EXEC SQL
COMMIT mytrans2;
EXEC SQL
DISCONNECT
. . .

82 INTERBASE 6
WORKING WITH MULTIPLE TRANSACTIONS IN DSQL

Working with multiple transactions in DSQL

In InterBase, DSQL applications can also use multiple transactions, but with the following
limitations:
g Programs must be preprocessed with the gpre -m switch.
g Transaction names must be declared statically. They cannot be defined through
user-modified host variables at run time.
g Transaction names must be initialized to zero before appearing in DSQL statements.
g All transactions must be started with explicit SET TRANSACTION
statements.
g No data definition language (DDL) can be used in the context of a named transaction in
an embedded program; DDL must always occur in the context of the default transaction,
GDS__TRANS.
g As long as a transaction name parameter is not specified with a SET TRANSACTION
statement, it can follow a PREPARE statement to modify the behavior of a subsequently
named transaction in an EXECUTE or EXECUTE IMMEDIATE statement. This enables a user
to modify transaction behaviors at run time.
Transaction names are fixed for all InterBase programs during preprocessing, and cannot
be dynamically assigned. A user can still modify DSQL transaction behavior at run time.
It is up to the programmer to anticipate possible transaction behavior modification and
plan for it. The following section describes how users can modify transaction behavior.

Modifying transaction behavior with “?”

The number and name of transactions available to a DSQL program is fixed when the
program is preprocessed with gpre, the InterBase preprocessor. The programmer
determines both the named transactions that control each DSQL statement in a program,
and the default behavior of those transactions. A user can change a named transaction’s
behavior at run time.
In DSQL programs, a user enters an SQL statement into a host-language string variable,
and then the host variable is processed in a PREPARE statement or EXECUTE IMMEDIATE
statement.

PREPARE
g Checks the statement in the variable for errors
g Loads the statement into an XSQLDA for a subsequent EXECUTE statement

EMBEDDED SQL GUIDE 83

 CHAPTER 4 WORKING WITH TRANSACTIONS

EXECUTE IMMEDIATE
g Checks the statement for errors
g Loads the statement into the XSQLDA
g Executes the statement
Both EXECUTE and EXECUTE IMMEDIATE operate within the context of a
programmer-specified transaction, which can be a named transaction. If the transaction
name is omitted, these statements are controlled by the default transaction, GDS__TRANS.
You can modify the transaction behavior for an EXECUTE and EXECUTE IMMEDIATE
statement by:
g Enabling a user to enter a SET TRANSACTION statement into a host variable
g Executing the SET TRANSACTION statement before the EXECUTE or EXECUTE IMMEDIATE
whose transaction context should be modified
In this context, a SET TRANSACTION statement changes the behavior of the next named or
default transaction until another SET TRANSACTION occurs.
The following C code fragment provides the user the option of specifying a new
transaction behavior, applies the behavior change, executes the next user statement in
the context of that changed transaction, then restores the transaction’s original behavior.
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char usertrans[512], query[1024];
char deftrans[] = {"SET TRANSACTION READ WRITE WAIT SNAPSHOT"};
EXEC SQL
END DECLARE SECTION;
. . .
printf("\nEnter SQL statement: ");
gets(query);
printf("\nChange transaction behavior (Y/N)? ");
gets(usertrans);
if (usertrans[0] == "Y" || usertrans[0] == "y")
{
printf("\nEnter \"SET TRANSACTION\" and desired behavior: ");
gets(usertrans);
EXEC SQL
COMMIT usertrans;
EXEC SQL
EXECUTE IMMEDIATE usertrans;
}

84 INTERBASE 6
WORKING WITH MULTIPLE TRANSACTIONS IN DSQL

else
{
EXEC SQL
EXECUTE IMMEDIATE deftrans;
}
EXEC SQL
EXECUTE IMMEDIATE query;
EXEC SQL
EXECUTE IMMEDIATE deftrans;
. . .

IMPORTANT As this example illustrates, you must commit or roll back any previous transactions
before you can execute SET TRANSACTION.

EMBEDDED SQL GUIDE 85

 CHAPTER 4 WORKING WITH TRANSACTIONS

86 INTERBASE 6
 CHAPTER

Working with Data

Chapter5
5
Definition Statements

This chapter discusses how to create, modify, and delete databases, tables, views, and
indexes in SQL applications. A database’s tables, views, and indexes make up most of its
underlying structure, or metadata.

IMPORTANT The discussion in this chapter applies equally to dynamic SQL (DSQL) applications,
except that users enter DSQL data definition statements at run time, and do not preface
those statements with EXEC SQL.

EMBEDDED SQL GUIDE 87

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

The preferred method for creating, modifying, and deleting metadata is through the
InterBase interactive SQL tool, isql, but in some instances, it may be necessary or desirable
to embed some data definition capabilities in an SQL application. Both SQL and DSQL
applications can use the following subset of data definition statements:

CREATE statement ALTER statement DROP statement

CREATE DATABASE ALTER DATABASE —

CREATE DOMAIN ALTER DOMAIN DROP DOMAIN

CREATE GENERATOR SET GENERATOR —

CREATE INDEX ALTER INDEX DROP INDEX

CREATE SHADOW ALTER SHADOW DROP SHADOW

CREATE TABLE ALTER TABLE DROP TABLE

CREATE VIEW — DROP VIEW

DECLARE EXTERNAL — DROP EXTERNAL

DECLARE FILTER — DROP FILTER

TABLE 5.1 Data definition statements supported for embedded applications

DSQL also supports creating, altering, and dropping stored procedures, triggers, and
exceptions. DSQL is especially powerful for data definition because it enables users to
enter any supported data definition statement at run time. For example, isql itself is a
DSQL application. For more information about using isql to define stored procedures,
triggers, and exceptions, see the Data Definition Guide. For a complete discussion of
DSQL programming, see Chapter 13, “Using Dynamic SQL.”

Creating metadata
SQL data definition statements are used in applications the sole purpose of which is to
create or modify databases or tables. Typically the expectation is that these applications
will be used only once by any given user, then discarded, or saved for later modification
by a database designer who can read the program code as a record of a database’s
structure. If data definition changes must be made, editing a copy of existing code is
easier than starting over.
Note Use the InterBase interactive SQL tool, isql, to create and alter data definitions
whenever possible. For more information about isql, see the Operations Guide

88 INTERBASE 6
CREATING METADATA

The SQL CREATE statement is used to make new databases, domains, tables, views, or
indexes. A COMMIT statement must follow every CREATE so that subsequent CREATE
statements can use previously defined metadata upon which they may rely. For example,
domain definitions must be committed before the domain can be referenced in
subsequent table definitions.

IMPORTANT Applications that mix data definition and data manipulation must be preprocessed using
the gpre -m switch. Such applications must explicitly start every transaction with SET
TRANSACTION.

Creating a database
CREATE DATABASE establishes a new database and its system tables, tables that describe the
internal structure of the database. InterBase uses the system tables whenever an
application accesses a database. SQL programs can read the data in most of these tables
just like any user-created table.
In its most elementary form, the syntax for CREATE DATABASE is:
EXEC SQL
CREATE DATABASE ’<filespec>’;

CREATE DATABASE must appear before any other CREATE statements. It requires one
parameter, the name of a database to create. For example, the following statement creates
a database named employee.gdb:
EXEC SQL
CREATE DATABASE ’employee.gdb’;

Note The database name can include a full file specification, including both host or node
names, and a directory path to the location where the database file should be created.
For information about file specifications for a particular operating system, see the
operating system manuals.

IMPORTANT Although InterBase enables access to remote databases, you should always create a
database directly on the machine where it is to reside.

4 Optional parameters
There are optional parameters for CREATE DATABASE. For example, when an application
running on a client attempts to connect to an InterBase server in order to create a
database, it may be expected to provide USER and PASSWORD parameters before the
connection is established. Other parameters specify the database page size, the number
and size of multi-file databases, and the default character set for the database.

EMBEDDED SQL GUIDE 89

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

For a complete discussion of all CREATE DATABASE parameters, see the Data Definition
Guide. For the complete syntax of CREATE DATABASE, see the Language Reference.

IMPORTANT An application that creates a database must be preprocessed with the gpre -m switch. It
must also create at least one table. If a database is created without a table, it cannot be
successfully opened by another program. Applications that perform both data definition
and data manipulation must declare tables with DECLARE TABLE before creating and
populating them. For more information about table creation, see “Creating a table” on
page 91.

4 Specifying a default character set

A database’s default character set designation specifies the character set the server uses
to transliterate and store CHAR, VARCHAR, and text Blob data in the database when no
other character set information is provided. A default character set should always be
specified for a database when it is created with CREATE DATABASE.
To specify a default character set, use the DEFAULT CHARACTER SET clause of CREATE
DATABASE. For example, the following statement creates a database that uses the
ISO8859_1 character set:
EXEC SQL
CREATE DATABASE ’europe.gdb’ DEFAULT CHARACTER SET ISO8859_1;

If you do not specify a character set, the character set defaults to NONE. Using character
set NONE means that there is no character set assumption for columns; data is stored and
retrieved just as you originally entered it. You can load any character set into a column
defined with NONE, but you cannot later move that data into another column that has
been defined with a different character set. In this case, no transliteration is performed
between the source and destination character sets, and errors may occur during
assignment.
For a complete description of the DEFAULT CHARACTER SET clause and a list of the character
sets supported by InterBase, see the Data Definition Guide.

Creating a domain
CREATE DOMAIN creates a column definition that is global to the database, and that can
be used to define columns in subsequent CREATE TABLE statements. CREATE DOMAIN is
especially useful when many tables in a database contain identical column definitions.
For example, in an employee database, several tables might define columns for
employees’ first and last names.
At its simplest, the syntax for CREATE DOMAIN is:

90 INTERBASE 6
CREATING METADATA

EXEC SQL
CREATE DOMAIN name AS <datatype>;

The following statements create two domains, FIRSTNAME, and LASTNAME.

EXEC SQL
CREATE DOMAIN FIRSTNAME AS VARCHAR(15);
EXEC SQL
CREATE DOMAIN LASTNAME AS VARCHAR(20);
EXEC SQL
COMMIT;

Once a domain is defined and committed, it can be used in CREATE TABLE statements to
define columns. For example, the following CREATE TABLE fragment illustrates how the
FIRSTNAME and LASTNAME domains can be used in place of column definitions in the
EMPLOYEE table definition.
EXEC SQL
CREATE TABLE EMPLOYEE
(
. . .
FIRST_NAME FIRSTNAME NOT NULL,
LAST_NAME LASTNAME NOT NULL;
. . .
);

A domain definition can also specify a default value, a NOT NULL attribute, a CHECK
constraint that limits inserts and updates to a range of values, a character set, and a
collation order.
For more information about creating domains and using them during table creation, see
the Data Definition Guide. For the complete syntax of CREATE DOMAIN, see the Language
Reference.

Creating a table
The CREATE TABLE statement defines a new database table and the columns and integrity
constraints within that table. Each column can include a character set specification and
a collation order specification. CREATE TABLE also automatically imposes a default SQL
security scheme on the table. The person who creates a table becomes its owner. A table’s
owner is assigned all privileges for it, including the right to grant privileges to other users.
A table can be created only for a database that already exists. At its simplest, the syntax
for CREATE TABLE is as follows:

EMBEDDED SQL GUIDE 91

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

EXEC SQL
CREATE TABLE name (<col_def> | <table_constraint>
[, <col_def> | <table_constraint> ...]);

col_def defines a column using the following syntax:

col {<datatype> | COMPUTED [BY] (<expr>) | domain}
<col_constraint> COLLATE collation

col must be a column name unique within the table definition.

datatype specifies the SQL datatype to use for column entries. COMPUTED BY can be used
to define a column whose value is computed from an expression when the column is
accessed at run time.
col_constraint is an optional integrity constraint that applies only to the associated
column.
tableconstraint is an optional integrity constraint that applies to an entire table.
Integrity constraints ensure that data entered in a table meets specific requirements, to
specify that data entered in a table or column is unique, or to enforce referential integrity
with other tables in the database.
The following code fragment contains SQL statements that create a database,
employee.gdb, and create a table, EMPLOYEE_PROJECT, with three columns, EMP_NO,
PROJ_ID, and DUTIES:
EXEC SQL
CREATE DATABASE ’employee.gdb’;
EXEC SQL
CREATE TABLE EMPLOYEE_PROJECT
(
EMP_NO SMALLINT NOT NULL,
PROJ_ID CHAR(5) NOT NULL,
DUTIES Blob SUBTYPE 1 SEGMENT SIZE 240
);
EXEC SQL
COMMIT;

An application can create multiple tables, but duplicating an existing table name is not
permitted.
For more information about SQL datatypes and integrity constraints, see the Data
Definition Guide. For more information about CREATE TABLE syntax, see the Language
Reference. For more information about changing or assigning table privileges, see the
security chapter in the Data Definition Guide.

92 INTERBASE 6
CREATING METADATA

4 Creating a computed column

A computed column is one whose value is calculated when the column is accessed at run
time. The value can be derived from any valid SQL expression that results in a single,
non-array value.
To create a computed column, use the following column declaration syntax in CREATE
TABLE:
col COMPUTED [BY] (<expr>)

The expression can reference previously defined columns in the table. For example, the
following statement creates a computed column, FULL_NAME, by concatenating two other
columns, LAST_NAME, and FIRST_NAME:
EXEC SQL
CREATE TABLE EMPLOYEE
(
. . .
FIRST_NAME VARCHAR(10) NOT NULL,
LAST_NAME VARCHAR(15) NOT NULL,
. . .
FULL_NAME COMPUTED BY (LAST_NAME || ’, ’ || FIRST_NAME)
);

For more information about COMPUTED BY, see the Data Definition Guide.

4 Declaring and creating a table

In programs that mix data definition and data manipulation, the DECLARE TABLE statement
must be used to describe a table’s structure to the InterBase preprocessor, gpre, before that
table can be created. During preprocessing, if gpre encounters a DECLARE TABLE statement,
it stores the table’s description for later reference. When gpre encounters a CREATE TABLE
statement for the previously declared table, it verifies that the column descriptions in the
CREATE statement match those in the DECLARE statement. If they do not match, gpre reports
the errors and cancels preprocessing so that the error can be fixed.
When used, DECLARE TABLE must come before the CREATE TABLE statement it describes.
For example, the following code fragment declares a table,
EMPLOYEE_PROJ, then creates it:
EXEC SQL
DECLARE EMPLOYEE_PROJECT TABLE
(
EMP_NO SMALLINT,
PROJ_ID CHAR(5),
DUTIES Blob(240, 1)

EMBEDDED SQL GUIDE 93

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

);
EXEC SQL
CREATE TABLE EMPLOYEE_PROJECT
(
EMP_NO SMALLINT,
PROJ_ID CHAR(5),
DUTIES Blob(240, 1)
);
EXEC SQL
COMMIT;

For more information about DECLARE TABLE, see the Language Reference.

Creating a view
A view is a virtual table that is based on a subset of one or more actual tables in a
database. Views are used to:
g Restrict user access to data by presenting only a subset of available data.
g Rearrange and present data from two or more tables in a manner especially useful to the
program.
Unlike a table, a view is not stored in the database as raw data. Instead, when a view is
created, the definition of the view is stored in the database. When a program uses the
view, InterBase reads the view definition and quickly generates the output as if it were a
table.
To make a view, use the following CREATE VIEW syntax:
EXEC SQL
CREATE VIEW name [(view_col [, view_col ...)] AS
<select> [WITH CHECK OPTION];

The name of the view, name, must be unique within the database.
To give each column displayed in the view its own name, independent of its column
name in an underlying table, enclose a list of view_col parameters in parentheses. Each
column of data returned by the view’s SELECT statement is assigned sequentially to a
corresponding view column name. If a list of view column names is omitted, column
names are assigned directly from the underlying table.
Listing independent names for columns in a view ensures that the appearance of a view
does not change if its underlying table structures are modified.

94 INTERBASE 6
CREATING METADATA

Note A view column name must be provided for each column of data returned by the
view’s SELECT statement, or else no view column names should be specified.
The select clause is a standard SELECT statement that specifies the selection criteria for
rows to include in the view. A SELECT in a view cannot include an ORDER BY clause. In
DSQL, it cannot include a UNION clause.
The optional WITH CHECK OPTION restricts inserts, updates, and deletes in a view that can
be updated.
To create a read-only view, a view’s creator must have SELECT privilege for the table or
tables underlying the view. To create a view for update requires ALL privilege for the table
or tables underlying the view. For more information about SQL privileges, see the security
chapter in the Data Definition Guide.

4 Creating a view for SELECT

Many views combine data from multiple tables or other views. A view based on multiple
tables or other views can be read, but not updated. For example, the following statement
creates a read-only view, PHONE_LIST, because it joins two tables, EMPLOYEE, and
DEPARTMENT:
EXEC SQL
CREATE VIEW PHONE_LIST AS
SELECT EMP_NO, FIRST_NAME, LAST_NAME, LOCATION, PHONE_NO
FROM EMPLOYEE, DEPARTMENT
WHERE EMPLOYEE.DEPT_NO = DEPARTMENT.DEPT_NO;
EXEC SQL
COMMIT;

IMPORTANT Only a view’s creator initially has access to it. To assign read access to others, use GRANT.
For more information about GRANT, see the security chapter of the Data Definition
Guide.

4 Creating a view for update

An updatable view is one that enables privileged users to insert, update, and delete
information in the view’s base table. To be updatable, a view must meet the following
conditions:
g It derives its columns from a single table or updatable view.
g It does not define a self-join of the base table.
g It does not reference columns derived from arithmetic expressions.
g The view’s SELECT statement does not contain:

EMBEDDED SQL GUIDE 95

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

· A WHERE clause that uses the DISTINCT predicate

· A HAVING clause
· Functions
· Nested queries
· Stored procedures
In the following view, HIGH_CITIES is an updatable view. It selects all cities in the CITIES
table with altitudes greater than or equal to a half mile.
EXEC SQL
CREATE VIEW HIGH_CITIES AS
SELECT CITY, COUNTRY_NAME, ALTITUDE FROM CITIES
WHERE ALTITUDE >= 2640;
EXEC SQL
COMMIT;

Users who have INSERT and UPDATE privileges for this view can change rows in or add
new rows to the view’s underlying table, CITIES. They can even insert or update rows that
cannot be displayed by the HIGH_CITIES view. The following INSERT adds a record for
Santa Cruz, California, altitude 23 feet, to the CITIES table:
EXEC SQL
INSERT INTO HIGH_CITIES (CITY, COUNTRY_NAME, ALTITUDE)
VALUES (’Santa Cruz’, ’United States’, ’23’);

To restrict inserts and updates through a view to only those rows that can be selected by
the view, use the WITH CHECK OPTION in the view definition. For example, the following
statement defines the view, HIGH_CITIES, to use the WITH CHECK OPTION. Users with INSERT
and UPDATE privileges will be able to enter rows only for cities with altitudes greater than
or equal to a half mile.
EXEC SQL
CREATE VIEW HIGH_CITIES AS
SELECT CITY, COUNTRY_NAME, ALTITUDE FROM CITIES
WHERE ALTITUDE > 2640 WITH CHECK OPTION;

Creating an index
SQL provides CREATE INDEX for establishing user-defined database indexes. An index,
based on one or more columns in a table, is used to speed data retrieval for queries that
access those columns. The syntax for CREATE INDEX is:

96 INTERBASE 6
CREATING METADATA

EXEC SQL
CREATE [UNIQUE] [ASC[ENDING] | DESC[ENDING]] INDEX <index> ON
table (col [, col ...]);

For example, the following statement defines an index, NAMEX, for the LAST_NAME and
FIRST_NAME columns in the EMPLOYEE table:
EXEC SQL
CREATE INDEX NAMEX ON EMPLOYEE (LAST_NAME, FIRST_NAME);

Note InterBase automatically generates system-level indexes when tables are defined
using UNIQUE and PRIMARY KEY constraints. For more information about constraints, see
the Data Definition Guide.
See the Language Reference for more information about CREATE INDEX syntax.

4 Preventing duplicate index entries

To define an index that eliminates duplicate entries, include the UNIQUE keyword in
CREATE INDEX. The following statement creates a unique index, PRODTYPEX, on the
PROJECT table:
EXEC SQL
CREATE UNIQUE INDEX PRODTYPEX ON PROJECT (PRODUCT, PROJ_NAME);

IMPORTANT After a unique index is defined, users cannot insert or update values in indexed columns
if those values already exist there. For unique indexes defined on multiple columns, like
PRODTYPEX in the previous example, the same value can be entered within individual
columns, but the combination of values entered in all columns defined for the index
must be unique.

4 Specifying index sort order

By default, SQL stores an index in ascending order. To make a descending sort on a
column or group of columns more efficient, use the DESCENDING keyword to define the
index. For example, the following statement creates an index, CHANGEX, based on the
CHANGE_DATE column in the SALARY_HISTORY table:
EXEC SQL
CREATE DESCENDING INDEX CHANGEX ON SALARY_HISTORY (CHANGE_DATE);

Note To retrieve indexed data in descending order, use ORDER BY in the SELECT statement
to specify retrieval order.

EMBEDDED SQL GUIDE 97

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

Creating generators
A generator is a monotonically increasing or decreasing numeric value that is inserted in
a field either directly by an SQL statement in an application or through a trigger.
Generators are often used to produce unique values to insert into a column used as a
primary key.
To create a generator for use in an application, use the following CREATE GENERATOR
syntax:
EXEC SQL
CREATE GENERATOR name;

The following statement creates a generator, EMP_NO_GEN, to specify a unique employee

number:
EXEC SQL
CREATE GENERATOR EMP_NO_GEN;
EXEC SQL
COMMIT;

Once a generator is created, the starting value for a generated number can be specified
with SET GENERATOR. To insert a generated number in a field, use the InterBase library
GEN_ID() function in an assignment statement. For more information about GEN_ID(),
CREATE GENERATOR, and SET GENERATOR, see the Data Definition Guide.

Dropping metadata
SQL supports several statements for deleting existing metadata:
g DROP TABLE, to delete a table from a database
g DROP VIEW, to delete a view definition from a database
g DROP INDEX, to delete a database index
g ALTER TABLE, to delete columns from a table
For more information about deleting columns with ALTER TABLE, see “Altering a table”
on page 101.

98 INTERBASE 6
DROPPING METADATA

Dropping an index
To delete an index, use DROP INDEX. An index can only be dropped by its creator, the
SYSDBA, or a user with root privileges. If an index is in use when the drop is attempted,
the drop is postponed until the index is no longer in use. The syntax of DROP INDEX is:
EXEC SQL
DROP INDEX name;

name is the name of the index to delete. For example, the following statement drops the
index, NEEDX:
EXEC SQL
DROP INDEX NEEDX;
EXEC SQL
COMMIT;
Deletion fails if the index is on a UNIQUE, PRIMARY KEY, or FOREIGN KEY integrity
constraint. To drop an index on a UNIQUE, PRIMARY KEY, or FOREIGN KEY integrity
constraint, first drop the constraints, the constrained columns, or the table.
For more information about DROP INDEX and dropping integrity constraints, see the Data
Definition Guide.

Dropping a view
To delete a view, use DROP VIEW. A view can only be dropped by its owner, the SYSDBA,
or a user with root privileges. If a view is in use when a drop is attempted, the drop is
postponed until the view is no longer in use. The syntax of DROP VIEW is:
EXEC SQL
DROP VIEW name;

The following statement drops the EMPLOYEE_SALARY view:

EXEC SQL
DROP VIEW EMPLOYEE_SALARY;
EXEC SQL
COMMIT;

Deleting a view fails if a view is used in another view, a trigger, or a computed column.
To delete a view that meets any of these conditions:
1. Delete the other view, trigger, or computed column.
2. Delete the view.

EMBEDDED SQL GUIDE 99

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

For more information about DROP VIEW, see the Data Definition Guide.

Dropping a table
Use DROP TABLE to remove a table from a database. A table can only be dropped by its
owner, the SYSDBA, or a user with root privileges. If a table is in use when a drop is
attempted, the drop is postponed until the table is no longer in use. The syntax of DROP
TABLE is:
EXEC SQL
DROP TABLE name;

name is the name of the table to drop. For example, the following statement drops the
EMPLOYEE table:
EXEC SQL
DROP TABLE EMPLOYEE;
EXEC SQL
COMMIT;

Deleting a table fails if a table is used in a view, a trigger, or a computed column. A table
cannot be deleted if a UNIQUE or PRIMARY KEY integrity constraint is defined for it, and
the constraint is also referenced by a FOREIGN KEY in another table. To drop the table, first
drop the FOREIGN KEY constraints in the other table, then drop the table.
Note Columns within a table can be dropped without dropping the rest of the table. For
more information, see “Dropping an existing column” on page 102.
For more information about DROP TABLE, see the Data Definition Guide.

Altering metadata
Most changes to data definitions are made at the table level, and involve adding new
columns to a table, or dropping obsolete columns from it. SQL provides ALTER TABLE to
add new columns to a table and to drop existing columns. A single ALTER TABLE can carry
out a single operation, or both operations.
Making changes to views and indexes always requires two separate statements:
1. Drop the existing definition.
2. Create a new definition.
If current metadata cannot be dropped, replacement definitions cannot be added.
Dropping metadata can fail for the following reasons:

100 INTERBASE 6
ALTERING METADATA

g The person attempting to drop metadata is not the metadata’s creator.

g SQL integrity constraints are defined for the metadata and referenced in other metadata.
g The metadata is used in another view, trigger, or computed column.
For more information about dropping metadata, see “Dropping metadata” on page 98.

Altering a table
ALTER TABLE enables the following changes to an existing table:
g Adding new column definitions
g Adding new table constraints
g Dropping existing column definitions
g Dropping existing table constraints
g Changing column definitions by dropping existing definitions, and adding new ones
g Changing existing table constraints by dropping existing definitions, and adding new
ones
g Modifying column names and datatypes
The simple syntax of ALTER TABLE is as follows:
EXEC SQL
ALTER TABLE name {ADD colname <datatype> [NOT NULL]
| ALTER [COLUMN] simple_column_name alter_rel_field
| DROP colname | ADD CONSTRAINT constraintname tableconstraint
| DROP CONSTRAINT constraintname};

Note For information about adding, dropping, and modifying constraints at the table
level, see the Data Definition Guide.
For the complete syntax of ALTER TABLE, see the Language Reference.

4 Adding a new column to a table

To add another column to an existing table, use ALTER TABLE. A table can only be modified
by its creator. The syntax for adding a column with ALTER TABLE is:
EXEC SQL
ALTER TABLE name ADD colname <datatype> colconstraint
[, ADD colname datatype colconstraint ...];

For example, the following statement adds a column, EMP_NO, to the EMPLOYEE table:

EMBEDDED SQL GUIDE 101

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

EXEC SQL
ALTER TABLE EMPLOYEE ADD EMP_NO EMPNO NOT NULL;
EXEC SQL
COMMIT;

This example makes use of a domain, EMPNO, to define a column. For more information
about domains, see the Data Definition Guide.
Multiple columns can be added to a table at the same time. Separate column definitions
with commas. For example, the following statement adds two columns, EMP_NO, and
FULL_NAME, to the EMPLOYEE table. FULL_NAME is a computed column, a column that
derives it values from calculations based on other columns:
EXEC SQL
ALTER TABLE EMPLOYEE
ADD EMP_NO EMPNO NOT NULL,
ADD FULL_NAME COMPUTED BY (LAST_NAME || ’, ’ || FIRST_NAME);
EXEC SQL
COMMIT;

This example creates a column using a value computed from two other columns already
defined for the EMPLOYEE table. For more information about creating computed columns,
see the Data Definition Guide.
New columns added to a table can be defined with integrity constraints. For more
information about adding columns with integrity constraints to a table, see the Data
Definition Guide.

4 Dropping an existing column

To delete a column definition and its data from a table, use ALTER TABLE. A column can
only be dropped by the owner of the table, the SYSDBA, or a user with root privileges. If
a table is in use when a column is dropped, the drop is postponed until the table is no
longer in use. The syntax for dropping a column with ALTER TABLE is:
EXEC SQL
ALTER TABLE name DROP colname [, colname ...];

For example, the following statement drops the EMP_NO column from the EMPLOYEE table:
EXEC SQL
ALTER TABLE EMPLOYEE DROP EMP_NO;
EXEC SQL
COMMIT;

Multiple columns can be dropped with a single ALTER TABLE. The following statement
drops the EMP_NO and FULL_NAME columns from the EMPLOYEE table:

102 INTERBASE 6
ALTERING METADATA

EXEC SQL
ALTER TABLE EMPLOYEE
DROP EMP_NO,
DROP FULL_NAME;
EXEC SQL
COMMIT;

Deleting a column fails if the column is part of a UNIQUE, PRIMARY KEY, or FOREIGN KEY
constraint. To drop the column, first drop the constraint, then the column.
Deleting a column also fails if the column is used by a CHECK constraint for another
column. To drop the column, first drop the CHECK constraint, then drop the column.
For more information about integrity constraints, see the Data Definition Guide.

4 Modifying a column
An existing column definition can be modified using ALTER TABLE, but if data already
stored in that column is not preserved before making changes, it will be lost.
Preserving data entered in a column and modifying the definition for a column, is a
five-step process:
1. Adding a new, temporary column to the table that mirrors the current
metadata of the column to be changed.
2. Copying the data from the column to be changed to the newly created
temporary column.
3. Modifying the column.
4. Copying data from the temporary column to the redefined column.
5. Dropping the temporary column.
For example, suppose the EMPLOYEE table contains a column, OFFICE_NO, defined to hold
a datatype of CHAR(3), and suppose that the size of the column needs to be increased by
one. The following numbered sequence describes each step and provides sample code:
1. First, create a temporary column to hold the data in OFFICE_NO during the
modification process:
EXEC SQL
ALTER TABLE EMPLOYEE ADD TEMP_NO CHAR(3);
EXEC SQL
COMMIT;
2. Move existing data from OFFICE_NO to TEMP_NO to preserve it:
EXEC SQL
UPDATE EMPLOYEE

EMBEDDED SQL GUIDE 103

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

SET TEMP_NO = OFFICE_NO;

3. Modify the new column definition for OFFICE_NO, specifying the datatype and
new size:
EXEC SQL
ALTER TABLE EMPLOYEE ALTER OFFICE_NO TYPE CHAR(4);
EXEC SQL
COMMIT;
4. Move the data from TEMP_NO to OFFICE_NO:
EXEC SQL
UPDATE EMPLOYEE
SET OFFICE_NO = TEMP_NO;
5. Finally, drop the TEMP_NO column:
EXEC SQL
ALTER TABLE DROP TEMP_NO;
EXEC SQL
COMMIT;

4 For more information about dropping column definitions, see “Dropping an

existing column” on page 102. For more information about adding column
definitions, see Modifying columns in a table
The ALTER TABLE ALTER command allows you to change the column position and name as
well.
For example, the following statement moves a column, EMP_NO, from the third position
to the second position in the EMPLOYEE table:
ALTER TABLE EMPLOYEE ALTER EMP_NO POSITION 2;

You could also change the name of the EMP_NO column to EMP_NUM as in the following
example:
ALTER TABLE EMPLOYEE ALTER EMP_NO TO EMP_NUM;

IMPORTANT Any changes to the field definitions may require the indexes to be rebuilt.
For the complete syntax of ALTER TABLE, see the Language Reference.

104 INTERBASE 6
ALTERING METADATA

Altering a view
To change the information provided by a view, follow these steps:
1. Drop the current view definition.
2. Create a new view definition and give it the same name as the dropped view.
For example, the following view is defined to select employee salary information:
EXEC SQL
CREATE VIEW EMPLOYEE_SALARY AS
SELECT EMP_NO, LAST_NAME, CURRENCY, SALARY
FROM EMPLOYEE, COUNTRY
WHERE EMPLOYEE.COUNTRY_CODE = COUNTRY.CODE;

Suppose the full name of each employee should be displayed instead of the last name.
First, drop the current view definition:
EXEC SQL
DROP EMPLOYEE_SALARY;
EXEC SQL
COMMIT;

Then create a new view definition that displays each employee’s full name:
EXEC SQL
CREATE VIEW EMPLOYEE_SALARY AS
SELECT EMP_NO, FULL_NAME, CURRENCY, SALARY
FROM EMPLOYEE, COUNTRY
WHERE EMPLOYEE.COUNTRY_CODE = COUNTRY.CODE;
EXEC SQL
COMMIT;

Altering an index
To change the definition of an index, follow these steps:
1. Use ALTER INDEX to make the current index inactive.
2. Drop the current index.
3. Create a new index and give it the same name as the dropped index.
An index is usually modified to change the combination of columns that are indexed, to
prevent or allow insertion of duplicate entries, or to specify index sort order. For example,
given the following definition of the NAMEX index:

EMBEDDED SQL GUIDE 105

 CHAPTER 5 WORKING WITH DATA DEFINITION STATEMENTS

EXEC SQL
CREATE INDEX NAMEX ON EMPLOYEE (LAST_NAME, FIRST_NAME);

Suppose there is an additional need to prevent duplicate entries with the UNIQUE
keyword. First, make the current index inactive, then drop it:
EXEC SQL
ALTER INDEX NAMEX INACTIVE;
EXEC SQL
DROP INDEX NAMEX;
EXEC SQL
COMMIT;
Then create a new index, NAMEX, based on the previous definition, that also includes the
UNIQUE keyword:
EXEC SQL
CREATE UNIQUE INDEX NAMEX ON EMPLOYEE (LAST_NAME, FIRST_NAME);
EXEC SQL
COMMIT

ALTER INDEX can be used directly to change an index’s sort order, or to add the ability to
handle unique or duplicate entries. For example, the following statement changes the
NAMEX index to permit duplicate entries:
EXEC SQL
ALTER INDEX NAMEX DUPLICATE;

IMPORTANT Be careful when altering an index directly. For example, changing an index from
supporting duplicate entries to one that requires unique entries without disabling the
index and recreating it can reduce index performance.
For more information about dropping an index, see “Dropping an index” on page 99.
For more information about creating an index, see “Creating an index” on page 96.

106 INTERBASE 6
 CHAPTER

Working with Data

Chapter6
6
The majority of SQL statements in an embedded program are devoted to reading or
modifying existing data, or adding new data to a database. This chapter describes the
types of data recognized by InterBase, and how to retrieve, modify, add, or delete data in
a database using SQL expressions and the following statements.
g SELECT statements query a database, that is, read or retrieve existing data from a database.
Variations of the SELECT statement make it possible to retrieve:
· A single row, or part of a row, from a table. This operation is referred to as a singleton
select.
· Multiple rows, or parts of rows, from a table using a SELECT within a DECLARE CURSOR
statement.
· Related rows, or parts of rows, from two or more tables into a virtual table, or results
table. This operation is referred to as a join.
· All rows, or parts of rows, from two or more tables into a virtual table. This operation
is referred to as a union.
g INSERT statements write new rows of data to a table.
g UPDATE statements modify existing rows of data in a table.
g DELETE statements remove existing rows of data from a table.

EMBEDDED SQL GUIDE 107

 CHAPTER 6 WORKING WITH DATA

To learn how to use the SELECT statement to retrieve data, see “Understanding data
retrieval with SELECT” on page 126. For information about retrieving a single row with
SELECT, see “Selecting a single row” on page 143. For information about retrieving
multiple rows, see “Selecting multiple rows” on page 144.
For information about using INSERT to write new data to a table, see “Inserting data” on
page 164. To modify data with UPDATE, see “Updating data” on page 170. To remove
data from a table with DELETE, see “Deleting data” on page 176.

Supported datatypes
To query or write to a table, it is necessary to know the structure of the table, what
columns it contains, and what datatypes are defined for those columns. InterBase
supports ten fundamental datatypes, described in the following table:

Name Size Range/Precision Description

BLOB Variable • None • Dynamically sizable dataype for
• Blob segment size is limited to 64K storing large data such as graphics,
text, and digitized voice
• Basic structural unit is the segment
• Blob subtype describes Blob contents
CHAR(n) n characters • 1 to 32,767 bytes • Fixed length CHAR or text string type
• Character set character size determines • Alternate keyword: CHARACTER
the maximum number of characters
that can fit in 32K
DATE 64 bits 1 Jan 100 a.d. to 29 Feb 32768 a.d. ISC_DATE

DECIMAL (precision, scale) Variable • precision = 1 to 18; specifies at least • Number with a decimal point scale
(16, 32, or precision digits of precision to store digits from the right
64 bits) • scale = 1 to 18; specifies number of • Example: DECIMAL(10, 3) holds
decimal places for storage numbers accurately in the following
• Must be less than or equal to precision format: ppppppp.sss
DOUBLE PRECISION 64 bitsa 2.225 x 10–308 to 1.797 x 10308 IEEE double precision: 15 digits
FLOAT 32 bits 1.175 x 10–38 to 3.402 x 1038 IEEE single precision: 7 digits
INTEGER 32 bits –2,147,483,648 to 2,147,483,647 Signed long (longword)
TABLE 6.1 Datatypes supported by InterBase

108 INTERBASE 6
SUPPORTED DATATYPES

Name Size Range/Precision Description

NUMERIC (precision, scale) Variable • precision = 1 to 18; specifies exactly • Number with a decimal point scale
(16, 32, or precision digits of precision to store digits from the right
64 bits) • scale = 1 to 18; specifies number of • Example: NUMERIC(10,3) holds
decimal places for storage numbers accurately in the following
• Must be less than or equal to precision format: ppppppp.sss
SMALLINT 16 bits –32,768 to 32,767 Signed short (word)
TIME 64 bits 00:00 AM to 23:59.9999 PM ISC_TIME

TIMESTAMP 64 bits 1 Jan 100 a.d. to 29 Feb 32768 a.d. ISC_TIMESTAMP; contains both date and
time information
VARCHAR (n) n characters • 1 to 32,765 bytes • Variable length CHAR or text string
• Character set character size determines type
the maximum number of characters • Alternate keywords: CHAR VARYING,
that can fit in 32K CHARACTER VARYING

TABLE 6.1 Datatypes supported by InterBase (continued)

a. Actual size of DOUBLE is platform-dependent. Most platforms support the 64-bit size.

The BLOB datatype can store large data objects of indeterminate and variable size, such
as bitmapped graphics images, vector drawings, sound files, chapter or book-length
documents, or any other kind of multimedia information. Because a Blob can hold
different kinds of information, it requires special processing for reading and writing. For
more information about Blob handling, see Chapter 8, “Working with Blob Data.”
The DATE, TIME, and TIMESTAMP datatypes may require conversion to and from InterBase when
entered or manipulated in a host-language program. For more information about
retrieving and writing dates, see Chapter 7, “Working with Dates.”
InterBase also supports arrays of most datatypes. An array is a matrix of individual items,
all of any single InterBase datatype, except Blob, that can be handled either as a single
entity, or manipulated item by item. To learn more about the flexible data access provided
by arrays, see Chapter 9, “Using Arrays.”
For a complete discussion of InterBase datatypes, see the Data Definition Guide.

EMBEDDED SQL GUIDE 109

 CHAPTER 6 WORKING WITH DATA

Understanding SQL expressions

All SQL data manipulation statements support SQL expressions, SQL syntax for comparing
and evaluating columns, constants, and host-language variables to produce a single
value.
In the SELECT statement, for example, the WHERE clause is used to specify a search
condition that determines if a row qualifies for retrieval. That search condition is an SQL
expression. DELETE and UPDATE also support search condition expressions. Typically,
when an expression is used as a search condition, the expression evaluates to a Boolean
value that is True, False, or Unknown.
SQL expressions can also appear in the INSERT statement VALUE clause and the UPDATE
statement SET clause to specify or calculate values to insert into a column. When inserting
or updating a numeric value via an expression, the expression is usually arithmetic, such
as multiplying one number by another to produce a new number which is then inserted
or updated in a column. When inserting or updating a string value, the expression may
concatenate, or combine, two strings to produce a single string for insertion or updating.
The following table describes the elements that can be used in expressions:

Element Description
Column names Columns from specified tables, against which to search or compare values,
or from which to calculate values.
Host-language variables Program variables containing changeable values. Host-language
variables must be preceded by a colon (:).
Constants Hard-coded numbers or quoted strings, like 507 or “Tokyo”.
Concatenation operator ||, used to combine character strings.
Arithmetic operators +, –, *, and /, used to calculate and evaluate values.
Logical operators Keywords, NOT, AND, and OR, used within simple search conditions, or to
combine simple search conditions to make complex searches. A logical
operation evaluates to true or false. Usually used only in search conditions.
TABLE 6.2 Elements of SQL expressions

110 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

Element Description
Comparison operators <, >, <=, >=, =, and <>, used to compare a value on the left side of the
operator to another on the right. A comparative operation evaluates to
true or false.
Other, more specialized comparison operators include ALL, ANY, BETWEEN,
CONTAINING, EXISTS, IN, IS, LIKE, NULL, SINGULAR, SOME, and STARTING WITH.
These operators can evaluate to True, False, or Unknown. They are usually
used only in search conditions.
COLLATE clause Comparisons of CHAR and VARCHAR values can sometimes take advantage
of a COLLATE clause to force the way text values are compared.
Stored procedures Reusable SQL statement blocks that can receive and return parameters,
and that are stored as part of a database’s metadata.
Subqueries SELECT statements, typically nested in WHERE clauses, that return values to
be compared with the result set of the main SELECT statement.
Parentheses Used to group expressions into hierarchies; operations inside parentheses
are performed before operations outside them. When parentheses are
nested, the contents of the innermost set is evaluated first and evaluation
proceeds outward.
Date literals String values that can be entered in quotes and be interpreted as date
values in SELECT, INSERT, and UPDATE operations. Possible strings are ‘ TODAY’,
‘NOW’, ‘YESTERDAY’, and ‘TOMORROW’.
The USER pseudocolumn References the name of the user who is currently logged in. For example,
USER can be used as a default in a column definition or to enter the current
user’s name in an INSERT. When a user name is present in a table, it can be
referenced with USER in SELECT and DELETE statements.
TABLE 6.2 Elements of SQL expressions (continued)

Complex expressions can be constructed by combining simple expressions in different

ways. For example the following WHERE clause uses a column name, three constants,
three comparison operators, and a set of grouping parentheses to retrieve only those rows
for employees with salaries between $60,000 and $120,000:
WHERE DEPARTMENT = ’Publications’ AND
(SALARY > 60000 AND SALARY < 120000)

EMBEDDED SQL GUIDE 111

 CHAPTER 6 WORKING WITH DATA

As another example, search conditions in WHERE clauses often contain nested SELECT
statements, or subqueries. In the following query, the WHERE clause contains a subquery
that uses the aggregate function, AVG(), to retrieve a list of all departments with bigger
than average salaries:
EXEC SQL
DECLARE WELL_PAID CURSOR FOR
SELECT DEPT_NO
INTO :wellpaid
FROM DEPARTMENT
WHERE SALARY > (SELECT AVG(SALARY) FROM DEPARTMENT);

For more information about using subqueries to specify search conditions, see “Using
subqueries” on page 161. For more information about aggregate functions, see
“Retrieving aggregate column information” on page 129.

Using the string operator in expressions

The string operator, ||, also referred to as a concatenation operator, enables a single
character string to be built from two or more character strings. Character strings can be
constants or values retrieved from a column. For example,
char strbuf[80];
. . .
EXEC SQL
SELECT LAST_NAME || ’ is the manager of publications.’
INTO :strbuf
FROM DEPARTMENT, EMPLOYEE
WHERE DEPT_NO = 5900 AND MNGR_NO = EMP_NO;

The string operator can also be used in INSERT or UPDATE statements:

EXEC SQL
INSERT INTO DEPARTMENT (MANAGER_NAME)
VALUES(:fname || :lname);

112 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

Using arithmetic operators in expressions

To calculate numeric values in expressions, InterBase recognizes four arithmetic
operators listed in the following table:

Operator Purpose Precedence Operator Purpose Precedence

* Multiplication 1 + Addition 3
/ Division 2 – Subtraction 4
TABLE 6.3 Arithmetic operators

Arithmetic operators are evaluated from left to right, except when ambiguities arise. In
these cases, InterBase evaluates operations according to the precedence specified in the
table (for example, multiplications are performed before divisions, and divisions are
performed before subtractions).
Arithmetic operations are always calculated before comparison and logical operations. To
change or force the order of evaluation, group operations in parentheses. InterBase
calculates operations within parentheses first. If parentheses are nested, the equation in
the innermost set is the first evaluated, and the outermost set is evaluated last. For more
information about precedence and using parentheses for grouping, see “Determining
precedence of operators” on page 122.
The following example illustrates a WHERE clause search condition that uses an arithmetic
operator to combine the values from two columns, then uses a comparison operator to
determine if that value is greater than 10:
DECLARE RAINCITIES CURSOR FOR
SELECT CITYNAME, COUNTRYNAME
INTO :cityname, :countryname
FROM CITIES
WHERE JANUARY_RAIN + FEBRUARY_RAIN > 10;

Using logical operators in expressions

Logical operators calculate a Boolean value, True, False, or Unknown, based on
comparing previously calculated simple search conditions immediately to the left and
right of the operator. InterBase recognizes three logical operators, NOT, AND, and OR.
NOT reverses the search condition in which it appears, while AND and OR are used to
combine simple search conditions. For example, the following query returns any
employee whose last name is not “Smith”:

EMBEDDED SQL GUIDE 113

 CHAPTER 6 WORKING WITH DATA

DECLARE NOSMITH CURSOR FOR

SELECT LAST_NAME
INTO :lname
FROM EMPLOYEE
WHERE NOT LNAME = ’Smith’;

When AND appears between search conditions, both search conditions must be true if a
row is to be retrieved. The following query returns any employee whose last name is
neither “Smith” nor “Jones”:
DECLARE NO_SMITH_OR_JONES CURSOR FOR
SELECT LAST_NAME
INTO :lname
FROM EMPLOYEE
WHERE NOT LNAME = ’Smith’ AND NOT LNAME = ’Jones’;

OR stipulates that one search condition or the other must be true. For example, the
following query returns any employee named “Smith” or “Jones”:
DECLARE ALL_SMITH_JONES CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
INTO :lname, :fname
FROM EMPLOYEE
WHERE LNAME = ’Smith’ OR LNAME = ’Jones’;

The order in which combined search conditions are evaluated is dictated by the
precedence of the operators that connect them. A NOT condition is evaluated before AND,
and AND is evaluated before OR. Parentheses can be used to change the order of
evaluation. For more information about precedence and using parentheses for grouping,
see “Determining precedence of operators” on page 122.

Using comparison operators in expressions

Comparison operators evaluate to a Boolean value: True, False, or Unknown, based on a
test for a specific relationship between a value to the left of the operator, and a value or
range of values to the right of the operator. Values compared must evaluate to the same
datatype, unless the CAST() function is used to translate one datatype to a different one
for comparison. Values can be columns, constants, or calculated values.
The following table lists operators that can be used in statements, describes how they are
used, and provides samples of their use:
Note Comparisons evaluate to Unknown if a NULL value is encountered.

114 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

For more information about CAST(), see “Using CAST( ) for datatype conversions” on
page 125.
InterBase also supports comparison operators that compare a value on the left of the
operator to the results of a subquery to the right of the operator. The following table lists
these operators, and describes how they are used:

Operator Purpose
ALL Determines if a value is equal to all values returned by a subquery
ANY and SOME Determines if a value is equal to any values returned by a subquery
EXISTS Determines if a value exists in at least one value returned by a subquery
SINGULAR Determines if a value exists in exactly one value returned by a subquery
TABLE 6.4 InterBase comparison operators requiring subqueries

For more information about using subqueries, see “Using subqueries” on page 161.

4 Using BETWEEN
BETWEEN tests whether a value falls within a range of values. The complete syntax for the
BETWEEN operator is:
<value> [NOT] BETWEEN <value> AND <value>

For example, the following cursor declaration retrieves LAST_NAME and FIRST_NAME
columns for employees with salaries between $100,000 and $250,000, inclusive:
EXEC SQL
DECLARE LARGE_SALARIES CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE SALARY BETWEEN 100000 AND 250000;

Use NOT BETWEEN to test whether a value falls outside a range of values. For example, the
following cursor declaration retrieves the names of employees with salaries less than
$30,000 and greater than $150,000:
EXEC SQL
DECLARE EXTREME_SALARIES CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE SALARY NOT BETWEEN 30000 AND 150000;

EMBEDDED SQL GUIDE 115

 CHAPTER 6 WORKING WITH DATA

4 Using CONTAINING
CONTAINING tests to see if an ASCII string value contains a quoted ASCII string supplied
by the program. String comparisons are case-insensitive; “String”, “STRING”, and “string”
are equivalent values for CONTAINING. Note that for Dialect 3 databases and clients, the
strings must be enclosed in single quotation marks. The complete syntax for CONTAINING
is:
<value> [NOT] CONTAINING ’<string>’

For example, the following cursor declaration retrieves the names of all employees whose
last names contain the three-letter combination, “las” (and “LAS” or “Las”):
EXEC SQL
DECLARE LAS_EMP CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE LAST_NAME CONTAINING ’las’;

Use NOT CONTAINING to test for strings that exclude a specified value. For example, the
following cursor declaration retrieves the names of all employees whose last names do
not contain “las” (also “LAS” or “Las”):
EXEC SQL
DECLARE NOT_LAS_EMP CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE LAST_NAME NOT CONTAINING ’las’;

Tip CONTAINING can be used to search a Blob segment by segment for an occurrence of a
quoted string.

4 Using IN
IN tests that a known value equals at least one value in a list of values. A list is a set of
values separated by commas and enclosed by parentheses. The values in the list must be
parenthesized and separated by commas. If the value being compared to a list of values
is NULL, IN returns Unknown.
The syntax for IN is:
<value> [NOT] IN (<value> [, <value> ...])

116 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

For example, the following cursor declaration retrieves the names of all employees in the
accounting, payroll, and human resources departments:
EXEC SQL
DECLARE ACCT_PAY_HR CURSOR FOR
SELECT DEPARTMENT, LAST_NAME, FIRST_NAME, EMP_NO
FROM EMPLOYEE EMP, DEPTARTMENT DEP
WHERE EMP.DEPT_NO = DEP.DEPT_NO AND
DEPARTMENT IN (’Accounting’, ’Payroll’, ’Human Resources’)
GROUP BY DEPARTMENT;

Use NOT IN to test that a value does not occur in a set of specified values. For example,
the following cursor declaration retrieves the names of all employees not in the
accounting, payroll, and human resources departments:
EXEC SQL
DECLARE NOT_ACCT_PAY_HR CURSOR FOR
SELECT DEPARTMENT, LAST_NAME, FIRST_NAME, EMP_NO
FROM EMPLOYEE EMP, DEPTARTMENT DEP
WHERE EMP.DEPT_NO = DEP.DEPT_NO AND
DEPARTMENT NOT IN (’Accounting’, ’Payroll’,
’Human Resources’)
GROUP BY DEPARTMENT;

IN can also be used to compare a value against the results of a subquery. For example,
the following cursor declaration retrieves all cities in Europe:
EXEC SQL
DECLARE NON_JFG_CITIES CURSOR FOR
SELECT C.COUNTRY, C.CITY, C.POPULATION
FROM CITIES C
WHERE C.COUNTRY NOT IN (SELECT O.COUNTRY FROM COUNTRIES O
WHERE O.CONTINENT <> ’Europe’)
GROUP BY C.COUNTRY;

For more information about subqueries, see “Using subqueries” on page 161.

4 Using LIKE
LIKE is a case-sensitive operator that tests a string value against a string containing
wildcards, symbols that substitute for a single, variable character, or a string of variable
characters. LIKE recognizes two wildcard symbols:
g % (percent) substitutes for a string of zero or more characters.
g _ (underscore) substitutes for a single character.

EMBEDDED SQL GUIDE 117

 CHAPTER 6 WORKING WITH DATA

The syntax for LIKE is:

<value> [NOT] LIKE <value> [ESCAPE ’symbol’]

For example, this cursor retrieves information about any employee whose last names
contain the three letter combination “ton” (but not “Ton”):
EXEC SQL
DECLARE TON_EMP CURSOR FOR
SELECT LAST_NAME, FIRST_NAME, EMP_NO
FROM EMPLOYEE
WHERE LAST_NAME LIKE ’%ton%’;

To test for a string that contains a percent or underscore character:

1. Precede the % or _ with another symbol (for example, @), in the quoted
comparison string.
2. Use the ESCAPE clause to identify the symbol (@, in this case) preceding % or
_ as a literal symbol. A literal symbol tells InterBase that the next character
should be included as is in the search string.
For example, this cursor retrieves all table names in RDB$RELATIONS that have underscores
in their names:
EXEC SQL
DECLARE UNDER_TABLE CURSOR FOR
SELECT RDB$RELATION_NAME
FROM RDB$RELATIONS
WHERE RDB$RELATION_NAME LIKE ’%@_%’ ESCAPE ’@’;

Use NOT LIKE to retrieve rows that do not contain strings matching those described. For
example, the following cursor retrieves all table names in RDB$RELATIONS that do not have
underscores in their names:
EXEC SQL
DECLARE NOT_UNDER_TABLE CURSOR FOR
SELECT RDB$RELATION_NAME
FROM RDB$RELATIONS
WHERE RDB$RELATION_NAME NOT LIKE ’%@_%’ ESCAPE ’@’;

4 Using IS NULL
IS NULL tests for the absence of a value in a column. The complete syntax of the IS NULL
clause is:
<value> IS [NOT] NULL

118 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

For example, the following cursor retrieves the names of employees who do not have
phone extensions:
EXEC SQL
DECLARE MISSING_PHONE CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE PHONE_EXT IS NULL;
Use IS NOT NULL to test that a column contains a value. For example, the following cursor
retrieves the phone numbers of all employees that have phone extensions:
EXEC SQL
DECLARE PHONE_LIST CURSOR FOR
SELECT LAST_NAME, FIRST_NAME, PHONE_EXT
FROM EMPLOYEE
WHERE PHONE_EXT IS NOT NULL
ORDER BY LAST_NAME, FIRST_NAME;

4 Using STARTING WITH

STARTING WITH is a case-sensitive operator that tests a string value to see if it begins with
a stipulated string of characters. To support international character set conversions,
STARTING WITH follows byte-matching rules for the specified collation order. The
complete syntax for STARTING WITH is:
<value> [NOT] STARTING WITH <value>

For example, the following cursor retrieves employee last names that start with “To”:
EXEC SQL
DECLARE TO_EMP CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE LAST_NAME STARTING WITH ’To’;

Use NOT STARTING WITH to retrieve information for columns that do not begin with the
stipulated string. For example, the following cursor retrieves all employees except those
whose last names start with “To”:
EXEC SQL
DECLARE NOT_TO_EMP CURSOR FOR
SELECT LAST_NAME, FIRST_NAME
FROM EMPLOYEE
WHERE LAST_NAME NOT STARTING WITH ’To’;

For more information about collation order and byte-matching rules, see the Data
Definition Guide.

EMBEDDED SQL GUIDE 119

 CHAPTER 6 WORKING WITH DATA

4 Using ALL
ALL tests that a value is true when compared to every value in a list returned by a
subquery. The complete syntax for ALL is:
<value> <comparison_operator> ALL (<subquery>)

For example, the following cursor retrieves information about employees whose salaries
are larger than that of the vice president of channel marketing:
EXEC SQL
DECLARE MORE_THAN_VP CURSOR FOR
SELECT LAST_NAME, FIRST_NAME, SALARY
FROM EMPLOYEE
WHERE SALARY > ALL (SELECT SALARY FROM EMPLOYEE
WHERE DEPT_NO = 7734);

ALL returns Unknown if the subquery returns a NULL value. It can also return Unknown
if the value to be compared is NULL and the subquery returns any non-NULL data. If the
value is NULL and the subquery returns an empty set, ALL evaluates to True.
For more information about subqueries, see “Using subqueries” on page 161.

4 Using ANY and SOME

ANY and SOME test that a value is true if it matches any value in a list returned by a
subquery. The complete syntax for ANY is:
<value> <comparison_operator> ANY | SOME (<subquery>)

For example, the following cursor retrieves information about salaries that are larger than
at least one salary in the channel marketing department:
EXEC SQL
DECLARE MORE_CHANNEL CURSOR FOR
SELECT LAST_NAME, FIRST_NAME, SALARY
FROM EMPLOYEE
WHERE SALARY > ANY (SELECT SALARY FROM EMPLOYEE
WHERE DEPT_NO = 7734);

ANY and SOME return Unknown if the subquery returns a NULL value. They can also return
Unknown if the value to be compared is NULL and the subquery returns any non-NULL
data. If the value is NULL and the subquery returns an empty set, ANY and SOME evaluate
to False.
For more information about subqueries, see “Using subqueries” on page 161.

120 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

4 Using EXISTS
EXISTS tests that for a given value there is at least one qualifying row meeting the search
condition specified in a subquery. The SELECT clause in the subquery must use the *
(asterisk) to select all columns. The complete syntax for EXISTS is:
[NOT] EXISTS (SELECT * FROM <tablelist> WHERE <search_condition>)

The following cursor retrieves all countries with rivers:

EXEC SQL
DECLARE RIVER_COUNTRIES CURSOR FOR
SELECT COUNTRY
FROM COUNTRIES C
WHERE EXISTS (SELECT * FROM RIVERS R
WHERE R.COUNTRY = C.COUNTRY);

Use NOT EXISTS to retrieve rows that do not meet the qualifying condition specified in the
subquery. The following cursor retrieves all countries without rivers:
EXEC SQL
DECLARE NON_RIVER_COUNTRIES COUNTRIES FOR
SELECT COUNTRY
FROM COUNTRIES C
WHERE NOT EXISTS (SELECT * FROM RIVERS R
WHERE R.COUNTRY = C.COUNTRY);

EXISTS always returns either True or False, even when handling NULL values.
For more information about subqueries, see “Using subqueries” on page 161.

4 Using SINGULAR
SINGULAR tests that for a given value there is exactly one qualifying row meeting the
search condition specified in a subquery. The SELECT clause in the subquery must use the
* (asterisk) to select all columns. The complete syntax for SINGULAR is:
[NOT] SINGULAR (SELECT * FROM <tablelist> WHERE <search_condition>)

The following cursor retrieves all countries with a single capital:

EXEC SQL
DECLARE SINGLE_CAPITAL CURSOR FOR
SELECT COUNTRY
FROM COUNTRIES COU
WHERE SINGULAR (SELECT * FROM CITIES CIT
WHERE CIT.CITY = COU.CAPITAL);

EMBEDDED SQL GUIDE 121

 CHAPTER 6 WORKING WITH DATA

Use NOT SINGULAR to retrieve rows that do not meet the qualifying condition specified in
the subquery. For example, the following cursor retrieves all countries with more than
one capital:
EXEC SQL
DECLARE MULTI_CAPITAL CURSOR FOR
SELECT COUNTRY
FROM COUNTRIES COU
WHERE NOT SINGULAR (SELECT * FROM CITIES CIT
WHERE CIT.CITY = COU.CAPITAL);

For more information about subqueries, see “Using subqueries” on page 161.

Determining precedence of operators

The order in which operators and the values they affect are evaluated in a statement is
called precedence. There are two levels of precedence for SQL operators:
g Precedence among operators of different types.
g Precedence among operators of the same type.

4 Precedence among operators

AMONG OPERATORS OF DIFFERENT TYPES
The following table lists the evaluation order of different InterBase operator types, from
first evaluated (highest precedence) to last evaluated (lowest precedence):

Operator type Precedence Explanation

String Highest Strings are always concatenated before all other operations
take place.
Mathematical ? Math is performed after string concatenation, but before
comparison and logical operations.
Comparison ? Comparison operations are evaluated after string
concatenation and math, but before logical operations.
Logical Lowest Logical operations are evaluated after all other operations.
TABLE 6.5 Operator precedence by operator type

122 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

AMONG OPERATORS OF THE SAME TYPE

When an expression contains several operators of the same type, those operators are
evaluated from left to right unless there is a conflict where two operators of the same type
affect the same values.
For example, in the mathematical equation, 3 + 2 * 6, both the addition and
multiplication operators work with the same value, 2. Evaluated from left to right, the
equation evaluates to 30: 3+ 2 = 5; 5 * 6 = 30. InterBase follows standard mathematical
rules for evaluating mathematical expressions, that stipulate multiplication is performed
before addition: 2 *6 = 12; 3 + 12 = 15.
The following table lists the evaluation order for all mathematical operators, from highest
to lowest:

Operator Precedence Explanation

* Highest Multiplication is performed before all other mathematical
operations.
/ ? Division is performed before addition and subtraction.
+ ? Addition is performed before subtraction.
– Lowest Subtraction is performed after all other mathematical operations.
TABLE 6.6 Mathematical operator precedence

InterBase also follows rules for determining the order in which comparison operators are
evaluated when conflicts arise during normal left to right evaluation. The next table
describes the evaluation order for comparison operators, from highest to lowest:

Operator Precedence Explanation

=, == Highest Equality operations are evaluated before all other
comparison operations.
<>, !=, ~=, ^= ?
> ?
TABLE 6.7 Comparison operator precedence

EMBEDDED SQL GUIDE 123

 CHAPTER 6 WORKING WITH DATA

Operator Precedence Explanation

< ?
>= ?
<= ?
!>, ~>, ^> ?
!<, ~<, ^< Lowest Not less than operations are evaluated after all other
comparison operations.
TABLE 6.7 Comparison operator precedence (continued)

ALL, ANY, BETWEEN, CONTAINING, EXISTS, IN, LIKE, NULL, SINGULAR, SOME, and STARTING
WITH are evaluated after all listed comparison operators when they conflict with other
comparison operators during normal left to right evaluation. When they conflict with one
another they are evaluated strictly from left to right.
When logical operators conflict during normal left to right processing, they, too, are
evaluated according to a hierarchy, detailed in the following table:

Operator Precedence Explanation

NOT Highest NOT operations are evaluated before all other logical operations.
AND ? AND operations are evaluated after NOT operations, and before OR
operations.
OR Lowest OR operations are evaluated after all other logical operations.
TABLE 6.8 Logical operator precedence

4 Changing evaluation order of operators

To change the evaluation order of operations in an expression, use parentheses to group
operations that should be evaluated as a unit, or that should derive a single value for use
in other operations. For example, without parenthetical grouping, 3 + 2 * 6 evaluates to
15. To cause the addition to be performed before the multiplication, use parentheses:
(3 + 2) * 6 = 30

Tip Always use parentheses to group operations in complex expressions, even when default
order of evaluation is desired. Explicitly grouped expressions are easier to understand
and debug.

124 INTERBASE 6
UNDERSTANDING SQL EXPRESSIONS

Using CAST( ) for datatype conversions

Normally, only similar datatypes can be compared or evaluated in expressions. The CAST()
function can be used in expressions to translate one datatype into another for comparison
purposes. The syntax for CAST() is:
CAST (<value> | NULL AS datatype)

For example, in the following WHERE clause, CAST() is used to translate a CHAR datatype,
INTERVIEW_DATE, to a DATE datatype to compare against a DATE datatype, HIRE_DATE:
WHERE HIRE_DATE = CAST(INTERVIEW_DATE AS DATE);

CAST() can be used to compare columns with different datatypes in the same table, or
across tables. You can convert one datatype to another as shown in the following table:

From datatype To datatype

NUMERIC CHARACTER

CHARACTER DATE, TIME, TIMESTAMP

TIMESTAMP DATE, TIME, CHARACTER

DATE TIMESTAMP, CHARACTER

TIME TIMESTAMP, CHARACTER

TABLE 6.9 Compatible datatypes for CAST()

An error results if a given datatype cannot be converted into the datatype specified in
CAST().

Using UPPER() on text data

The UPPER() function can be used in SELECT, INSERT, UPDATE, or DELETE operations to force
character and Blob text data to uppercase. For example, an application that prompts a
user for a department name might want to ensure that all department names are stored
in uppercase to simplify data retrieval later. The following code illustrates how UPPER()
would be used in the INSERT statement to guarantee a user’s entry is uppercase:
EXEC SQL
BEGIN DECLARE SECTION;
char response[26];
EXEC SQL
END DECLARE SECTION;

EMBEDDED SQL GUIDE 125

 CHAPTER 6 WORKING WITH DATA

. . .
printf("Enter new department name: ");
response[0] = ’\0’;
gets(response);
if (response)
EXEC SQL
INSERT INTO DEPARTMENT(DEPT_NO, DEPARTMENT)
VALUES(GEN_ID(GDEPT_NO, 1), UPPER(:response));
. . .

The next statement illustrates how UPPER() can be used in a SELECT statement to affect
both the appearance of values retrieved, and to affect its search condition:
EXEC SQL
SELECT DEPT_NO, UPPER(DEPARTMENT)
FROM DEPARTMENT
WHERE UPPER(DEPARTMENT) STARTING WITH ’A’;

Understanding data retrieval with SELECT

The SELECT statement handles all queries in SQL. SELECT can retrieve one or more rows
from a table, and can return entire rows, or a subset of columns from each row, often
referred to as a projection. Optional SELECT syntax can be used to specify search criteria
that restrict the number of rows returned, to select rows with unknown values, to select
rows through a view, and to combine rows from two or more tables.
At a minimum, every SELECT statement must:
g List which columns to retrieve from a table. The column list immediately follows the
SELECT keyword.
g Name the table to search in a FROM clause.
Singleton selects must also include both an INTO clause to specify the host variables into
which retrieved values should be stored, and a WHERE clause to specify the search
conditions that cause only a single row to be returned.
The following SELECT retrieves three columns from a table and stores the values in three
host-language variables:
EXEC SQL
SELECT EMP_NO, FIRSTNAME, LASTNAME
INTO :emp_no, :fname, :lname
FROM EMPLOYEE WHERE EMP_NO = 1888;

126 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

Tip Host variables must be declared in a program before they can be used in SQL
statements. For more information about declaring host variables, see Chapter 2,
“Application Requirements.”
The following table lists all SELECT statement clauses, in the order that they are used, and
prescribes their use in singleton and multi-row selects:

Singleton Multi-row
Clause Purpose SELECT SELECT
SELECT Lists columns to retrieve. Required Required
INTO Lists host variables for storing retrieved columns. Required Not allowed
FROM Identifies the tables to search for values. Required Required
WHERE Specifies the search conditions used to restrict retrieved Optional Optional
rows to a subset of all available rows. A WHERE clause can
contain its own SELECT statement, referred to as a subquery.
GROUP BY Groups related rows based on common column values. Optional Optional
Used in conjunction with HAVING.
HAVING Restricts rows generated by GROUP BY to a subset of those Optional Optional
rows.
UNION Combines the results of two or more SELECT statements to Optional Optional
produce a single, dynamic table without duplicate rows.
PLAN Specifies the query plan that should be used by the query Optional Optional
optimizer instead of one it would normally choose.
ORDER BY Specifies the sort order of rows returned by a SELECT, either Optional Optional
ascending (ASC), the default, or descending (DESC).
FOR UPDATE Specifies columns listed after the SELECT clause of a DECLARE Not allowed Optional
CURSOR statement that can be updated using a WHERE
CURRENT OF clause.

TABLE 6.10 SELECT statement clauses

Using each of these clauses with SELECT is described in the following sections, after which
using SELECT directly to return a single row, and using SELECT within a DECLARE CURSOR
statement to return multiple rows are described in detail. For a complete overview of
SELECT syntax, see the Language Reference.

EMBEDDED SQL GUIDE 127

 CHAPTER 6 WORKING WITH DATA

Listing columns to retrieve with SELECT

A list of columns to retrieve must always follow the SELECT keyword in a SELECT statement.
The SELECT keyword and its column list is called a SELECT clause.

4 Retrieving a list of columns

To retrieve a subset of columns for a row of data, list each column by name, in the order
of desired retrieval, and separate each column name from the next by a comma.
Operations that retrieve a subset of columns are often called projections.
For example, the following SELECT retrieves three columns:
EXEC SQL
SELECT EMP_NO, FIRSTNAME, LASTNAME
INTO :emp_no, :fname, :lname
FROM EMPLOYEE WHERE EMP_NO = 2220;

4 Retrieving all columns

To retrieve all columns of data, use an asterisk (*) instead of listing any columns by name.
For example, the following SELECT retrieves every column of data for a single row in the
EMPLOYEE table:
EXEC SQL
SELECT *
INTO :emp_no, :fname, :lname, :phone_ext, :hire, :dept_no,
:job_code, :job_grade, :job_country, :salary, :full_name
FROM EMPLOEE WHERE EMP_NO = 1888;

IMPORTANT You must provide one host variable for each column returned by a query.

ELIMINATING DUPLICATE COLUMNS WITH DISTINCT

In a query returning multiple rows, it may be desirable to eliminate duplicate columns.
For example, the following query, meant to determine if the EMPLOYEE table contains
employees with the last name, SMITH, might locate many such rows:
EXEC SQL
DECLARE SMITH CURSOR FOR
SELECT LAST_NAME
FROM EMPLOYEE
WHERE LAST_NAME = ’Smith’;

To eliminate duplicate columns in such a query, use the DISTINCT keyword with SELECT.
For example, the following SELECT yields only a single instance of “Smith”:

128 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

EXEC SQL
DECLARE SMITH CURSOR FOR
SELECT DISTINCT LAST_NAME
FROM EMPLOYEE
WHERE LAST_NAME = ’Smith’;

DISTINCT affects all columns listed in a SELECT statement.

4 Retrieving aggregate column information

SELECT can include aggregate functions, functions that calculate or retrieve a single,
collective numeric value for a column or expression based on each qualifying row in a
query rather than retrieving each value separately. The following table lists the aggregate
functions supported by InterBase:

Function Purpose
AVG() Calculates the average numeric value for a set of values.
MIN() Retrieves the minimum value in a set of values.
MAX() Retrieves the maximum value in a set of values.
SUM() Calculates the total of numeric values in a set of values.
COUNT() Calculates the number of rows that satisfy the query’s search condition
(specified in the WHERE clause).
TABLE 6.11 Aggregate functions in SQL

For example, the following query returns the average salary for all employees in the
EMPLOYEE table:
EXEC SQL
SELECT AVG(SALARY)
INTO :avg_sal
FROM EMPLOYEE;

The following SELECT returns the number of qualifying rows it encounters in the
EMPLOYEE table, both the maximum and minimum employee number of employees in the
table, and the total salary of all employees in the table:
EXEC SQL
SELECT COUNT(*), MAX(EMP_NO), MIN(EMP_NO), SUM(SALARY)
INTO :counter, :maxno, :minno, :total_salary
FROM EMPLOYEE;

EMBEDDED SQL GUIDE 129

 CHAPTER 6 WORKING WITH DATA

If a field value involved in an aggregate calculation is NULL or unknown, the entire row
is automatically excluded from the calculation. Automatic exclusion prevents averages
from being skewed by meaningless data.
Note Aggregate functions can also be used to calculate values for groups of rows. The
resulting value is called a group aggregate. For more information about using group
aggregates, see “Grouping rows with GROUP BY” on page 139.

4 Multi-table SELECT statements

When data is retrieved from multiple tables, views, and select procedures, the same
column name may appear in more than one table. In these cases, the SELECT statement
must contain enough information to distinguish like-named columns from one another.
To distinguish column names in multiple tables, precede those columns with one of the
following qualifiers in the SELECT clause:
g The name of the table, followed by a period. For example,
EMPLOYEE.EMP_NO identifies a column named EMP_NO in the EMPLOYEE table.
g A table correlation name (alias) followed by a period. For example, if the correlation
name for the EMPLOYEE table is EMP, then EMP.EMP_NO identifies a column named EMP_NO
in the EMPLOYEES table.
Correlation names can be declared for tables, views, and select procedures in the FROM
clause of the SELECT statement. For more information about declaring correlation names,
and for examples of their use, see “Declaring and using correlation names” on
page 134.

4 Specifying transaction names

InterBase enables an SQL application to run many simultaneous transactions if:
g Each transaction is first named with a SET TRANSACTION statement.
g Each data manipulation statement (SELECT, INSERT, UPDATE, DELETE) specifies a
TRANSACTION clause that identifies the name of the transaction under which it operates.
g SQL statements are not dynamic.
In SELECT, the TRANSACTION clause intervenes between the SELECT keyword and the
column list, as in the following syntax fragment:
SELECT TRANSACTION name <col> [, <col> ...]

The TRANSACTION clause is optional in single-transaction programs or in programs where

only one transaction is open at a time. It must be used in a multi-transaction program.
For example, the following SELECT is controlled by the transaction, T1:
EXEC SQL

130 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

SELECT TRANSACTION T1:

COUNT(*), MAX(EMP_NO), MIN(EMP_NO), SUM(SALARY)
INTO :counter, :maxno, :minno, :total_salary
FROM EMPLOYEE;

For a complete discussion of transaction handling and naming, see Chapter 4, “Working
with Transactions.”

Specifying host variables with INTO

A singleton select returns data to a list of host-language variables specified by an INTO
clause in the SELECT statement. The INTO clause immediately follows the list of table
columns from which data is to be extracted. Each host variable in the list must be
preceded by a colon (:) and separated from the next by a comma.
The host-language variables in the INTO clause must already have been declared before
they can be used. The number, order, and datatype of host-language variables must
correspond to the number, order, and datatype of the columns retrieved. Otherwise,
overflow or data conversion errors may occur.
For example, the following C program fragment declares three host variables, lname,
fname, and salary. Two, lname, and fname, are declared as character arrays; salary is
declared as a long integer. The SELECT statement specifies that three columns of data are
to be retrieved, while the INTO clause specifies the host variables into which the data
should be read.
. . .
EXEC SQL
BEGIN DECLARE SECTION;
long salary;
char lname[20], fname[15];
EXEC SQL
END DECLARE SECTION;
. . .
EXEC SQL
SELECT LAST_NAME, FIRST_NAME, SALARY
INTO :lanem, :fname, :salary
FROM EMPLOYEE
WHERE LNAME = ’Smith’;
. . .

EMBEDDED SQL GUIDE 131

 CHAPTER 6 WORKING WITH DATA

Note In a multi-row select, the INTO clause is part of the FETCH statement, not the SELECT
statement. For more information about the INTO clause in FETCH, see “Fetching rows
with a cursor” on page 147.

Listing tables to search with FROM

The FROM clause is required in a SELECT statement. It identifies the tables, views, or select
procedures from which data is to be retrieved. The complete syntax of the FROM clause is:
FROM table | view | procedure [alias] [, table | view | procedure
[alias] ...]

There must be at least one table, view, or select procedure name following the FROM
keyword. When retrieving data from multiple sources, each source must be listed,
assigned an alias, and separated from the next with a comma. For more information
about select procedures, see Chapter 10, “Working with Stored Procedures.”

4 Listing a single table or view

The FROM clause in the following SELECT specifies a single table, EMPLOYEE, from which
to retrieve data:
EXEC SQL
SELECT LAST_NAME, FIRST_NAME, SALARY
INTO :lanem, :fname, :salary
FROM EMPLOYEE
WHERE LNAME = ’Smith’;

Use the same INTO clause syntax to specify a view or select procedure as the source for
data retrieval instead of a table. For example, the following SELECT specifies a select
procedure, MVIEW, from which to retrieve data. MVIEW returns information for all
managers whose last names begin with the letter “M,” and the WHERE clause narrows the
rows returned to a single row where the DEPT_NO column is 430:
EXEC SQL
SELECT DEPT_NO, LAST_NAME, FIRST_NAME, SALARY
INTO :lname, :fname, :salary
FROM MVIEW
WHERE DEPT_NO = 430;

For more information about select procedures, see Chapter 10, “Working with
Stored Procedures.”

132 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

4 Listing multiple tables

To retrieve data from multiple tables, views, or select procedures, include all sources in
the FROM clause, separating sources from one another by commas.
There are two different possibilities to consider when working with multiple data sources:
1. The name of each referenced column is unique across all tables.
2. The names of one or more referenced columns exist in two or more tables.
In the first case, just use the column names themselves to reference the columns. For
example, the following query returns data from two tables, DEPARTMENT, and EMPLOYEE:
EXEC SQL
SELECT DEPARTMENT, DEPT_NO, LAST_NAME, FIRST_NAME, EMP_NO
INTO :dept_name, :dept_no, :lname, :fname, :empno
FROM DEPARTMENT, EMPLOYEE
WHERE DEPT_NO = ’Publications’ AND MNGR_NO = EMP_NO;

In the second case, column names that occur in two or more tables must be distinguished
from one another by preceding each column name with its table name and a period in
the SELECT clause. For example, if an EMP_NO column exists in both the DEPARTMENT and
EMPLOYEE then the previous query must be recast as follows:
EXEC SQL
SELECT DEPARTMENT, DEPT_NO, LAST_NAME, FIRST_NAME,
EMLOYEE.EMP_NO
INTO :dept_name, :dept_no, :lname, :fname, :empno
FROM DEPARTMENT, EMPLOYEE
WHERE DEPT_NO = ’Publications’ AND
DEPARTMENT.EMP_NO = EMPLOYEE.EMP_NO;

For more information about the SELECT clause, see “Listing columns to retrieve with
SELECT” on page 128.

IMPORTANT For queries involving joins, column names can be qualified by correlation names, brief
alternate names, or aliases, that are assigned to each table in a FROM clause and
substituted for them in other SELECT statement clauses when qualifying column names.
Even when joins are not involved, assigning and using correlation names can reduce the
length of complex queries.

EMBEDDED SQL GUIDE 133

 CHAPTER 6 WORKING WITH DATA

4 Declaring and using correlation names

A correlation name, or alias, is a temporary variable that represents a table name. It can
contain up to 31 alphanumeric characters, dollar signs ($), and underscores (_), but must
always start with an alphabetic character. Using brief correlation names reduces typing
of long queries. Correlation names must be substituted for actual table names in joins,
and can be substituted for them in complex queries.
A correlation name is associated with a table in the FROM clause; it replaces table names
to qualify column names everywhere else in the statement. For example, to associate the
correlation name, DEPT with the DEPARTMENT table, and EMP, with the EMPLOYEES table, a
FROM clause might appear as:
FROM DEPARTMENT DEPT, EMPLOYEE EMP

Like an actual table name, a correlation name is used to qualify column names wherever
they appear in a SELECT statement. For example, the following query employs the
correlation names, DEPT, and EMP, previously described:
EXEC SQL
SELECT DEPARTMENT, DEPT_NO, LAST_NAME, FIRST_NAME,
EMLOYEE.EMP_NO
INTO :dept_name, :dept_no, :lname, :fname, :empno
FROM DEPARTMENT DEPT, EMPLOYEE EMP
WHERE DEPT_NO = ’Publications’ AND DEPT.EMP_NO = EMP.EMP_NO;

For more information about the SELECT clause, see “Listing columns to retrieve with
SELECT” on page 128.

Restricting row retrieval with WHERE

In a query, the WHERE clause specifies the data a row must (or must not) contain to be
retrieved.
In singleton selects, where a query must only return one row, WHERE is mandatory unless
a select procedure specified in the FROM clause returns only one row itself.
In SELECT statements within DECLARE CURSOR statements, the WHERE clause is optional. If
the WHERE clause is omitted, a query returns all rows in the table. To retrieve a subset of
rows in a table, a cursor declaration must include a WHERE clause.
The simple syntax for WHERE is:
WHERE <search_condition>

For example, the following simple WHERE clause tests a row to see if the
DEPARTMENT column is “Publications”:

134 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

WHERE DEPARTMENT = ’Publications’

4 What is a search condition?

Because the WHERE clause specifies the type of data a query is searching for it is often
called a search condition. A query examines each row in a table to see if it meets the
criteria specified in the search condition. If it does, the row qualifies for retrieval.
When a row is compared to a search condition, one of three values is returned:
g True: A row meets the conditions specified in the WHERE clause.
g False: A row fails to meet the conditions specified in the WHERE clause.
g Unknown: A column tested in the WHERE clause contains an unknown value that could
not be evaluated because of a NULL comparison.
Most search conditions, no matter how complex, evaluate to True or False. An expression
that evaluates to True or False—like the search condition in the WHERE clause—is called
a Boolean expression.

4 Structure of a search condition

A typical simple search condition compares a value in one column against a constant or
a value in another column. For example, the following WHERE clause tests a row to see if
a field equals a hard-coded constant:
WHERE DEPARTMENT = ’Publications’

This search condition has three elements: a column name, a comparison operator (the
equal sign), and a constant. Most search conditions are more complex than this. They
involve additional elements and combinations of simple search conditions. The following
table describes expression elements that can be used in search conditions:

EMBEDDED SQL GUIDE 135

 CHAPTER 6 WORKING WITH DATA

Element Description
Column names Columns from tables listed in the FROM clause, against which to search or
compare values.
Host-language variables Program variables containing changeable values. When used in a SELECT,
host-language variables must be preceded by a colon (:).
Constants Hard-coded numbers or quoted strings, like 507 or “Tokyo”.
Concatenation operators ||, used to combine character strings.
Arithmetic operators +, –, *, and /, used to calculate and evaluate search condition values.
Logical operators Keywords, NOT, AND, and OR, used within simple search conditions, or to
combine simple search conditions to make complex searches. A logical
operation evaluates to true or false.
Comparison operators <, >, <=, >=, =, and <>, used to compare a value on the left side of the
operator to another on the right. A comparative operation evaluates to
True or False.
Other, more specialized comparison operators include ALL, ANY, BETWEEN,
CONTAINING, EXISTS, IN, IS, LIKE, NULL, SINGULAR, SOME, and STARTING WITH.
These operators can evaluate to True, False, or Unknown.
COLLATE clause Comparisons of CHAR and VARCHAR values can sometimes take advantage
of a COLLATE clause to force the way text values are compared.
Stored procedures Reusable SQL statement blocks that can receive and return parameters,
and that are stored as part of a database’s metadata. For more information
about stored procedures in queries, see Chapter 10, “Working with
Stored Procedures.”
Subqueries A SELECT statement nested within the WHERE clause to return or calculate
values against which rows searched by the main SELECT statement are
compared. For more information about subqueries, see “Using
subqueries” on page 161.
Parentheses Group related parts of search conditions which should be processed
separately to produce a single value which is then used to evaluate the
search condition. Parenthetical expressions can be nested.
TABLE 6.12 Elements of WHERE clause SEARCH conditions

136 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

Complex search conditions can be constructed by combining simple search conditions in

different ways. For example, the following WHERE clause uses a column name, three
constants, three comparison operators, and a set of grouping parentheses to retrieve only
those rows for employees with salaries between $60,000 and $120,000:
WHERE DEPARTMENT = ’Publications’ AND
(SALARY > 60000 AND SALARY < 120000)

Search conditions in WHERE clauses often contain nested SELECT statements, or

subqueries. For example, in the following query, the WHERE clause contains a subquery
that uses the aggregate function, AVG(), to retrieve a list of all departments with
bigger-than-average salaries:
EXEC SQL
DECLARE WELL_PAID CURSOR FOR
SELECT DEPT_NO
INTO :wellpaid
FROM DEPARTMENT
WHERE SALARY > (SELECT AVG(SALARY) FROM DEPARTMENT);

For a general discussion of building search conditions from SQL expressions, see
“Understanding SQL expressions” on page 110. For more information about using
subqueries to specify search conditions, see “Using subqueries” on page 161. For more
information about aggregate functions, see “Retrieving aggregate column
information” on page 129.

4 Collation order in comparisons

When CHAR or VARCHAR values are compared in a WHERE clause, it can be necessary to
specify a collation order for the comparisons if the values being compared use different
collation orders.
To specify the collation order to use for a value during a comparison, include a COLLATE
clause after the value. For example, in the following WHERE clause fragment from an
embedded application, the value to the left of the comparison operator is forced to be
compared using a specific collation:
WHERE LNAME COLLATE FR_CA = :lname_search;

For more information about collation order and a list of collations available to InterBase,
see the Data Definition Guide.

EMBEDDED SQL GUIDE 137

 CHAPTER 6 WORKING WITH DATA

Sorting rows with ORDER BY

By default, a query retrieves rows in the exact order it finds them in a table, and because
internal table storage is unordered, retrieval, too, is likely to be unordered. To specify the
order in which rows are returned by a query, use the optional ORDER BY clause at the end
of a SELECT statement.
ORDER BY retrieves rows based on a column list. Every column in the ORDER BY clause
must also appear somewhere in the SELECT clause at the start of the statement. Each
column can optionally be ordered in ascending order (ASC, the default), or descending
order (DESC). The complete syntax of ORDER BY is:
ORDER BY col [COLLATE collation] [ASC | DESC]
[,col [COLLATE collation] [ASC | DESC] ...];

For example, the following cursor declaration orders output based on the LAST_NAME
column. Because DESC is specified in the ORDER BY clause, employees are retrieved from
Z to A:
EXEC SQL
DECLARE PHONE_LIST CURSOR FOR
SELECT LAST_NAME, FIRST_NAME, PHONE_EXT
FROM EMPLOYEE
WHERE PHONE_EXT IS NOT NULL
ORDER BY LAST_NAME DESC, FIRST_NAME;

4 ORDER BY with multiple columns

If more than one column is specified in an ORDER BY clause, rows are first arranged by
the values in the first column. Then rows that contain the same first-column value are
arranged according to the values in the second column, and so on. Each ORDER BY
column can include its own sort order specification.

IMPORTANT In multi-column sorts, after a sort order is specified, it applies to all subsequent columns
until another sort order is specified, as in the previous example. This attribute is
sometimes called sticky sort order. For example, the following cursor declaration orders
retrieval by LAST_NAME in descending order, then refines it alphabetically within
LAST_NAME groups by FIRST_NAME in ascending order:
EXEC SQL
DECLARE PHONE_LIST CURSOR FOR
SELECT LAST_NAME, FIRST_NAME, PHONE_EXT
FROM EMPLOYEE
WHERE PHONE_EXT IS NOT NULL
ORDER BY LAST_NAME DESC, FIRST_NAME ASC;

138 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

4 Collation order in an ORDER BY clause

When CHAR or VARCHAR columns are ordered in a SELECT statement, it can be necessary
to specify a collation order for the ordering, especially if columns used for ordering use
different collation orders.
To specify the collation order to use for ordering a column in the ORDER BY clause,
include a COLLATE clause after the column name. For example, in the following ORDER BY
clause, a different collation order for each of two columns is specified:
. . .
ORDER BY LNAME COLLATE FR_CA, FNAME COLLATE FR_FR;

For more information about collation order and a list of available collations in InterBase,
see the Data Definition Guide.

Grouping rows with GROUP BY

The optional GROUP BY clause enables a query to return summary information about
groups of rows that share column values instead of returning each qualifying row. The
complete syntax of GROUP BY is:
GROUP BY col [COLLATE collation] [, col [COLLATE collation] ...]

For example, consider two cursor declarations. The first declaration returns the names of
all employees each department, and arranges retrieval in ascending alphabetic order by
department and employee name.
EXEC SQL
DECLARE DEPT_EMP CURSOR FOR
SELECT DEPARTMENT, LAST_NAME, FIRST_NAME
FROM DEPARTMENT D, EMPLOYEE E
WHERE D.DEPT_NO = E.DEPT_NO
ORDER BY DEPARTMENT, LAST_NAME, FIRST_NAME;

In contrast, the next cursor illustrates the use of aggregate functions with GROUP BY to
return results known as group aggregates. It returns the average salary of all employees
in each department. The GROUP BY clause assures that average salaries are calculated and
retrieved based on department names, while the ORDER BY clause arranges retrieved rows
alphabetically by department name.
EXEC SQL
DECLARE AVG_DEPT_SAL CURSOR FOR
SELECT DEPARTMENT, AVG(SALARY)
FROM DEPARTMENT D, EMPLOYEE E
WHERE D.DEPT_NO = E.DEPT_NO

EMBEDDED SQL GUIDE 139

 CHAPTER 6 WORKING WITH DATA

GROUP BY DEPARTMENT
ORDER BY DEPARTMENT;

4 Collation order in a GROUP BY clause

When CHAR or VARCHAR columns are grouped in a SELECT statement, it can be necessary
to specify a collation order for the grouping, especially if columns used for grouping use
different collation orders.
To specify the collation order to use for grouping columns in the GROUP BY clause,
include a COLLATE clause after the column name. For example, in the following GROUP
BY clause, the collation order for two columns is specified:
. . .
GROUP BY LNAME COLLATE FR_CA, FNAME COLLATE FR_CA;

For more information about collation order and a list of collation orders available in
InterBase, see the Data Definition Guide.

4 Limitations of GROUP BY
When using GROUP BY, be aware of the following limitations:
g Each column name that appears in a GROUP BY clause must also be specified in the SELECT
clause.
g GROUP BY cannot specify a column whose values are derived from a mathematical,
aggregate, or user-defined function.
g GROUP BY cannot be used in SELECT statements that:
· Contain an INTO clause (singleton selects).
· Use a subquery with a FROM clause which references a view whose definition contains
a GROUP BY or HAVING clause.
g For each SELECT clause in a query, including subqueries, there can only be one GROUP BY
clause.

Restricting grouped rows with HAVING

Just as a WHERE clause reduces the number of rows returned by a SELECT clause, the
HAVING clause can be used to reduce the number of rows returned by a GROUP BY clause.
The syntax of HAVING is:
HAVING <search_condition>

140 INTERBASE 6
UNDERSTANDING DATA RETRIEVAL WITH SELECT

HAVING uses search conditions that are like the search conditions that can appear in the
WHERE clause, but with the following restrictions:
g Each search condition usually corresponds to an aggregate function used in the SELECT
clause.
g The FROM clause of a subquery appearing in a HAVING clause cannot name any table or
view specified in the main query’s FROM clause.
g A correlated subquery cannot be used in a HAVING clause.
For example, the following cursor declaration returns the average salary for all employees
in each department. The GROUP BY clause assures that average salaries are calculated and
retrieved based on department names. The HAVING clause restricts retrieval to those
groups where the average salary is greater than 60,000, while the ORDER BY clause
arranges retrieved rows alphabetically by department name.
EXEC SQL
DECLARE SIXTY_THOU CURSOR FOR
SELECT DEPARTMENT, AVG(SALARY)
FROM DEPARTMENT D, EMPLOYEE E
WHERE D.DEPT_NO = E.DEPT_NO
GROUP BY DEPARTMENT
HAVING AVG(SALARY) > 60000
ORDER BY DEPARTMENT;

Note HAVING can also be used without GROUP BY. In this case, all rows retrieved by a
SELECT are treated as a single group, and each column named in the SELECT clause is
normally operated on by an aggregate function.
For more information about search conditions, see “Restricting row retrieval with
WHERE” on page 134. For more information about subqueries, see “Using subqueries”
on page 161.

Appending tables with UNION

Sometimes two or more tables in a database are identically structured, or have columns
that contain similar data. Where table structures overlap, information from those tables
can be combined to produce a single results table that returns a projection for every
qualifying row in both tables. The UNION clause retrieves all rows from each table,
appends one table to the end of another, and eliminates duplicate rows.
Unions are commonly used to perform aggregate operations on tables.

EMBEDDED SQL GUIDE 141

 CHAPTER 6 WORKING WITH DATA

The syntax for UNION is:

UNION SELECT col [, col ...] | * FROM <tableref> [, <tableref> ...]

For example, three tables, CITIES, COUNTRIES, and NATIONAL_PARKS, each contain the
names of cities. Assuming triggers have not been created that ensure that a city entered
in one table is also entered in the others to which it also applies, UNION can be used to
retrieve the names of all cities that appear in any of these tables.
EXEC SQL
DECLARE ALLCITIES CURSOR FOR
SELECT CIT.CITY FROM CITIES CIT
UNION SELECT COU.CAPITAL FROM COUNTRIES COU
UNION SELECT N.PARKCITY FROM NATIONAL_PARKS N;

Tip If two or more tables share entirely identical structures—similarly named columns,
identical datatypes, and similar data values in each column—UNION can return all rows
for each table by substituting an asterisk (*) for specific column names in the SELECT
clauses of the UNION.

Specifying a query plan with PLAN

To process a SELECT statement, InterBase uses an internal algorithm, called the query
optimizer, to determine the most efficient plan for retrieving data. The most efficient
retrieval plan also results in the fastest retrieval time. Occasionally the optimizer may
choose a plan that is less efficient. For example, when the number of rows in a table
grows sufficiently large, or when many duplicate rows are inserted or deleted from
indexed columns in a table, but the index’s selectivity is not recomputed, the optimizer
might choose a less efficient plan.
For these occasions, SELECT provided an optional PLAN clause that enables a
knowledgeable programmer to specify a retrieval plan. A query plan is built around the
availability of indexes, the way indexes are joined or merged, and a chosen access
method.
To specify a query plan, use the following PLAN syntax:
PLAN <plan_expr>
<plan_expr> =
[JOIN | [SORT] MERGE] (<plan_item> | <plan_expr>
[, <plan_item> | <plan_expr> ...])
<plan_item> = {table | alias}
NATURAL | INDEX ( <index> [, <index> ...]) | ORDER <index>

142 INTERBASE 6
SELECTING A SINGLE ROW

The PLAN syntax enables specifying a single table, or a join of two or more tables in a
single pass. Plan expressions can be nested in parentheses to specify any combination of
joins.
During retrieval, information from different tables is joined to speed retrieval. If indexes
are defined for the information to be joined, then these indexes are used to perform a
join. The optional JOIN keyword can be used to document this type of operation. When
no indexes exist for the information to join, retrieval speed can be improved by specifying
SORT MERGE instead of JOIN.
A plan_item is the name of a table to search for data. If a table is used more than once
in a query, aliases must be used to distinguish them in the PLAN clause. Part of the
plan_item specification indicates the way that rows should be accessed. The following
choices are possible:
g NATURAL, the default order, specifies that rows are accessed sequentially in no defined
order. For unindexed items, this is the only option.
g INDEX specifies that one or more indexes should be used to access items. All indexes to
be used must be specified. If any Boolean or join terms remain after all indexes are used,
they will be evaluated without benefit of an index. If any indexes are specified that cannot
be used, an error is returned.
g ORDER specifies that items are to be sorted based on a specified index.

Selecting a single row

An operation that retrieves a single row of data is called a singleton select. To retrieve a
single row from a table, to retrieve a column defined with a unique index, or to select an
aggregate value like COUNT() or AVG() from a table, use the following SELECT statement
syntax:
SELECT <col> [, <col> ...]
INTO :variable [, :variable ...]
FROM table
WHERE <search_condition>;

The mandatory INTO clause specifies the host variables where retrieved data is copied for
use in the program. Each host variable’s name must be preceded by a colon (:). For each
column retrieved, there must be one host variable of a corresponding datatype. Columns
are retrieved in the order they are listed in the SELECT clause, and are copied into host
variables in the order the variables are listed in the INTO clause.

EMBEDDED SQL GUIDE 143

 CHAPTER 6 WORKING WITH DATA

The WHERE clause must specify a search condition that guarantees that only one row is
retrieved. If the WHERE clause does not reduce the number of rows returned to a single
row, the SELECT fails.

IMPORTANT To select data from a table, a user must have SELECT privilege for a table, or a stored
procedure invoked by the user’s application must have SELECT privileges for the table.
For example, the following SELECT retrieves information from the
DEPARTMENT table for the department, Publications:
EXEC SQL
SELECT DEPARTMENT, DEPT_NO, HEAD_DEPT, BUDGET, LOCATION, PHONE_NO
INTO :deptname, :dept_no, :manager, :budget, :location, :phone
FROM DEPARTMENT
WHERE DEPARTMENT = ’Publications’;

When SQL retrieves the specified row, it copies the value in DEPARTMENT to the host
variable, deptname, copies the value in DEPT_NO to :dept_no, copies the value in
HEAD_DEPT to :manager, and so on.

Selecting multiple rows

Most queries specify search conditions that retrieve more than one row. For example, a
query that asks to see all employees in a company that make more than $60,000 can
retrieve many employees.
Because host variables can only hold a single column value at a time, a query that returns
multiple rows must build a temporary table in memory, called a results table, from which
rows can then be extracted and processed, one at a time, in sequential order. SQL keeps
track of the next row to process in the results table by establishing a pointer to it, called
a cursor.

IMPORTANT In dynamic SQL (DSQL), the process for creating a query and retrieving data is
somewhat different. For more information about multi-row selection in DSQL, see
“Selecting multiple rows in DSQL” on page 152.
To retrieve multiple rows into a results table, establish a cursor into the table, and process
individual rows in the table, SQL provides the following sequence of statements:
1. DECLARE CURSOR establishes a name for the cursor and specifies the query to
perform.
2. OPEN executes the query, builds the results table, and positions the cursor at
the start of the table.

144 INTERBASE 6
SELECTING MULTIPLE ROWS

3. FETCH retrieves a single row at a time from the results table into host variables
for program processing.
4. CLOSE releases system resources when all rows are retrieved.

IMPORTANT To select data from a table, a user must have SELECT privilege for a table, or a stored
procedure invoked by the user’s application must have SELECT privilege for it.

Declaring a cursor
To declare a cursor and specify rows of data to retrieve, use the DECLARE
CURSOR statement. DECLARE CURSOR is a descriptive, non-executable statement. InterBase
uses the information in the statement to prepare system resources for the cursor when it
is opened, but does not actually perform the query. Because DECLARE CURSOR is
non-executable, SQLCODE is not assigned when this statement is used.
The syntax for DECLARE CURSOR is:
DECLARE cursorname CURSOR FOR
SELECT <col> [, <col> ...]
FROM table [, <table> ...]
WHERE <search_condition>
[GROUP BY col [, col ...]]
[HAVING <search_condition>]
[ORDER BY col [ASC | DESC] [, col ...] [ASC | DESC]
| FOR UPDATE OF col [, col ...]];

The cursorname is used in subsequent OPEN, FETCH, and CLOSE statements to identify the
active cursor.
With the following exceptions, the SELECT statement inside a DECLARE
CURSOR is similar to a stand-alone SELECT:
g A SELECT in a DECLARE CURSOR cannot include an INTO clause.
g A SELECT in a DECLARE CURSOR can optionally include either an ORDER BY clause or a FOR
UPDATE clause.
For example, the following statement declares a cursor:
EXEC SQL
DECLARE TO_BE_HIRED CURSOR FOR
SELECT D.DEPARTMENT, D.LOCATION, P.DEPARTMENT
FROM DEPARTMENT D, DEPARTMENT P
WHERE D.MNGR_NO IS NULL
AND D.HEAD_DEPT = P.DEPT_NO;

EMBEDDED SQL GUIDE 145

 CHAPTER 6 WORKING WITH DATA

4 Updating through cursors

In many applications, data retrieval and update may be interdependent. DECLARE CURSOR
supports an optional FOR UPDATE clause that optionally lists columns in retrieved rows
that can be modified. For example, the following statement declares such a cursor:
EXEC SQL
DECLARE H CURSOR FOR
SELECT CUST_NO
FROM CUSTOMER
WHERE ON_HOLD = ’*’
FOR UPDATE OF ON_HOLD;

If a column list after FOR UPDATE is omitted, all columns retrieved for each row may be
updated. For example, the following query enables updating for two columns:
EXEC SQL
DECLARE H CURSOR FOR
SELECT CUST_NAME CUST_NO
FROM CUSTOMER
WHERE ON_HOLD = ’*’;

For more information about updating columns through a cursor, see “Updating multiple
rows” on page 171.

Opening a cursor
Before data selected by a cursor can be accessed, the cursor must be opened with the
OPEN statement. OPEN activates the cursor and builds a results table. It builds the results
table based on the selection criteria specified in the DECLARE CURSOR statement. The rows
in the results table comprise the active set of the cursor.
For example, the following statement opens a previously declared cursor called
DEPT_EMP:
EXEC SQL
OPEN DEPT_EMP;

When InterBase executes the OPEN statement, the cursor is positioned at the start of the
first row in the results table.

146 INTERBASE 6
SELECTING MULTIPLE ROWS

Fetching rows with a cursor

Once a cursor is opened, rows can be retrieved, one at a time, from the results table by
using the FETCH statement. FETCH:
1. Retrieves the next available row from the results table.
2. Copies those rows into the host variables specified in the INTO clause of the
FETCH statement.
3. Advances the cursor to the start of the next available row or sets
SQLCODE to 100, indicating the cursor is at the end of the results table and
there are no more rows to retrieve.
The complete syntax of the FETCH statement in SQL is:
FETCH <cursorname> INTO :variable [[INDICATOR] :variable]
[, :variable [[INDICATOR] :variable>] ...];

IMPORTANT In dynamic SQL (DSQL) multi-row select processing, a different FETCH syntax is used.
For more information about retrieving multiple rows in DSQL, see “Fetching rows with
a DSQL cursor” on page 154.
For example, the following statement retrieves a row from the results table for the
DEPT_EMP cursor, and copies its column values into the host-language variables,
deptname, lname, and fname:
EXEC SQL
FETCH DEPT_EMP
INTO :deptname, :lname, :fname;

To process each row in a results table in the same manner, enclose the FETCH statement
in a host-language looping construct. For example, the following C code fetches and
prints each row defined for the DEPT_EMP cursor:
. . .
EXEC SQL
FETCH DEPT_EMP
INTO :deptname, :lname, :fname;
while (!SQLCODE)
{
printf("%s %s works in the %s department.\n", fname,
lname, deptname);
EXEC SQL
FETCH DEPT_EMP
INTO :deptname, :lname, :fname;
}

EMBEDDED SQL GUIDE 147

 CHAPTER 6 WORKING WITH DATA

EXEC SQL
CLOSE DEPT_EMP;
. . .

Every FETCH statement should be tested to see if the end of the active set is reached. The
previous example operates in the context of a while loop that continues processing as
long as SQLCODE is zero. If SQLCODE is 100, it indicates that there are no more rows to
retrieve. If SQLCODE is less than zero, it indicates that an error occurred.

4 Retrieving indicator status

Any column can have a NULL value, except those defined with the NOT NULL or UNIQUE
integrity constraints. Rather than store a value for the column, InterBase sets a flag
indicating the column has no assigned value.
To determine if a value returned for a column is NULL, follow each variable named in the
INTO clause with the INDICATOR keyword and the name of a short integer variable, called
an indicator variable, where InterBase should store the status of the NULL value flag for
the column. If the value retrieved is:
g NULL, the indicator variable is set to –1.
g Not NULL, the indicator parameter is set to 0.
For example, the following C code declares three host-language variables, department,
manager, and missing_manager, then retrieves column values into
department, manager, and a status flag for the column retrieved into manager,
missing_manager, with a FETCH from a previously declared cursor, GETCITY:
. . .
char department[26];
char manager[36];
short missing_manager;
. . .
FETCH GETCITY INTO :department, :manager INDICATOR :missing_manager;

The optional INDICATOR keyword can be omitted:

FETCH GETCITY INTO :department, :manager :missing_manager;

Often, the space between the variable that receives the actual contents of a
column and the variable that holds the status of the NULL value flag is also omitted:
FETCH GETCITY INTO :department, :manager:missing_manager;

Note While InterBase enforces the SQL requirement that the number of host variables in
a FETCH must equal the number of columns specified in DECLARE CURSOR, indicator
variables in a FETCH statement are not counted toward the column count.

148 INTERBASE 6
SELECTING MULTIPLE ROWS

4 Refetching rows with a cursor

The only supported cursor movement is forward in sequential order through the active
set.
To revisit previously fetched rows, close the cursor and then reopen it with another OPEN
statement. For example, the following statements close the DEPT_EMP cursor, then
recreate it, effectively repositioning the cursor at the start of the DEPT_EMP results table:
EXEC SQL
CLOSE DEPT_EMP;
EXEC SQL
OPEN DEPT_EMP;

Closing the cursor

When the end of a cursor’s active set is reached, a cursor should be closed to free up
system resources. To close a cursor, use the CLOSE statement. For example, the following
statement closes the DEPT_EMP cursor:
EXEC SQL
CLOSE DEPT_EMP;

Programs can check for the end of the active set by examining SQLCODE, which is set to
100 to indicate there are no more rows to retrieve.

A complete cursor example

The following program declares a cursor, opens the cursor, and then loops through the
cursor’s active set, fetching and printing values. The program closes the cursor when all
processing is finished or an error occurs.
#include <stdio.h>
EXEC SQL
BEGIN DECLARE SECTION;
char deptname[26];
char lname[16];
char fname[11];
EXEC SQL
END DECLARE SECTION;

main ()
{

EMBEDDED SQL GUIDE 149

 CHAPTER 6 WORKING WITH DATA

EXEC SQL
WHENEVER SQLERROR GO TO abend;
EXEC SQL
DECLARE DEPT_EMP CURSOR FOR
SELECT DEPARTMENT, LAST_NAME, FIRST_NAME
FROM DEPARTMENT D, EMPLOYEE E
WHERE D.DEPT_NO = E.DEPT_NO
ORDER BY DEPARTMENT, LAST_NAME, FIRST_NAME;
EXEC SQL
OPEN DEPT_EMP;
EXEC SQL
FETCH DEPT_EMP
INTO :deptname, :lname, :fname;
while (!SQLCODE)
{
printf("%s %s works in the %s department.\n",fname,
lname, deptname);
EXEC SQL
FETCH DEPT_EMP
INTO :deptname, :lname, :fname;
}
EXEC SQL
CLOSE DEPT_EMP;
exit();

abend:
if (SQLCODE)
{
isc_print_sqlerror();
EXEC SQL
ROLLBACK;
EXEC SQL
CLOSE_DEPT_EMP;
EXEC SQL
DISCONNECT ALL;
exit(1)
}
else
{
EXEC SQL
COMMIT;
EXEC SQL

150 INTERBASE 6
SELECTING MULTIPLE ROWS

DISCONNECT ALL;
exit()
}
}

Selecting rows with NULL values

Any column can have NULL values, except those defined with the NOT NULL or UNIQUE
integrity constraints. Rather than store a value for the column, InterBase sets a flag
indicating the column has no assigned value.
Use IS NULL in a WHERE clause search condition to query for NULL values. For example,
some rows in the DEPARTMENT table do not have a value for the
BUDGET column. Departments with no stored budget have the NULL value flag set for that
column. The following cursor declaration retrieves rows for departments without budgets
for possible update:
EXEC SQL
DECLARE NO_BUDGET CURSOR FOR
SELECT DEPARTMENT, BUDGET
FROM DEPARTMENT
WHERE BUDGET IS NULL
FOR UPDATE OF BUDGET;

Note To determine if a column has a NULL value, use an indicator variable. For more
information about indicator variables, see “Retrieving indicator status” on page 148.
A direct query on a column containing a NULL value returns zero for numbers, blanks for
characters, and 17 November 1858 for dates. For example, the following cursor
declaration retrieves all department budgets, even those with NULL values, which are
reported as zero:
EXEC SQL
DECLARE ALL_BUDGETS CURSOR FOR
SELECT DEPARTMENT, BUDGET
FROM DEPARTMENT
ORDER BY BUDGET DESCENDING;

4 Limitations on NULL values

Because InterBase treats NULL values as non-values, the following limitations on NULL
values in queries should be noted:
g Rows with NULL values are sorted after all other rows.

EMBEDDED SQL GUIDE 151

 CHAPTER 6 WORKING WITH DATA

g NULL values are skipped by all aggregate operations, except for COUNT(*).
g NULL values cannot be elicited by a negated test in a search condition.
g NULL values cannot satisfy a join condition.
NULL values can be tested in comparisons. If a value on either side of a comparison
operator is NULL, the result of the comparison is Unknown.
For the Boolean operators (NOT, AND, and OR), the following considerations are made:
g NULL values with NOT always returns Unknown.
g NULL values with AND return Unknown unless one operand for AND is false. In this latter
case, False is returned.
g NULL values with OR return Unknown unless one operand for OR is true. In this latter case,
True is returned.
For information about defining alternate NULL values, see the Data Definition Guide.

Selecting rows through a view

To select a subset of rows available through a view, substitute the name of the view for a
table name in the FROM clause of a SELECT. For example, the following cursor produces a
list of employee phone numbers based on the PHONE_VIEW view:
EXEC SQL
DECLARE PHONE_LIST CURSOR FOR
SELECT FIRST_NAME, LAST_NAME, PHONE_EXT
FROM PHONE_VIEW
WHERE EMPLOYEE.DEPT_NO = DEPARTMENT.DEPT_NO;

A view can be a join. Views can also be used in joins, themselves, in place of tables. For
more information about views in joins, see “Joining tables” on page 155.

Selecting multiple rows in DSQL

In DSQL users are usually permitted to specify queries at run time. To accommodate any
type of query the user supplies, DSQL requires the use of extended SQL descriptor areas
(XSQLDAs) where a query’s input and output can be prepared and described. For queries
returning multiple rows, DSQL supports variations of the DECLARE CURSOR, OPEN, and
FETCH statements that make use of the XSQLDA.
To retrieve multiple rows into a results table, establish a cursor into the table, and process
individual rows in the table. DSQL provides the following sequence of statements:

152 INTERBASE 6
SELECTING MULTIPLE ROWS IN DSQL

1. PREPARE establishes the user-defined query specification in the XSQLDA

structure used for output.
2. DECLARE CURSOR establishes a name for the cursor and specifies the query to
perform.
3. OPEN executes the query, builds the results table, and positions the cursor at
the start of the table.
4. FETCH retrieves a single row at a time from the results table for program
processing.
5. CLOSE releases system resources when all rows are retrieved.
The following three sections describe how to declare a DSQL cursor, how to open it, and
how to fetch rows using the cursor. For more information about creating and filling
XSQLDA structures, and preparing DSQL queries with PREPARE, see Chapter 13, “Using
Dynamic SQL.”For more information about closing a cursor, see “Closing the cursor”
on page 149.

Declaring a DSQL cursor

DSQL must declare a cursor based on a user-defined SELECT statement. Usually, DSQL
programs:
g Prompt the user for a query (SELECT).
g Store the query in a host-language variable.
g Issue a PREPARE statement that uses the host-language variable to describe the query
results in an XSQLDA.
g Declare a cursor using the query alias.
The complete syntax for DECLARE CURSOR in DSQL is:
DECLARE cursorname CURSOR FOR queryname;

For example, the following C code fragment declares a string variable,

querystring, to hold the user-defined query, gets a query from the user and stores it in
querystring, uses querystring to PREPARE a query called QUERY, then declares a cursor, C,
that uses QUERY:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char querystring [512];
XSQLDA *InputSqlda, *OutputSqlda;

EMBEDDED SQL GUIDE 153

 CHAPTER 6 WORKING WITH DATA

EXEC SQL
END DECLARE SECTION;
. . .
printf("Enter query: "); /* prompt for query from user */
gets(querystring); /* get the string, store in querystring */
. . .
EXEC SQL
PREPARE QUERY INTO OutputSqlda FROM :querystring;
. . .
EXEC SQL
DECLARE C CURSOR FOR QUERY;

For more information about creating and filling XSQLDA structures, and preparing DSQL
queries with PREPARE, see Chapter 13, “Using Dynamic SQL.”

Opening a DSQL cursor

The OPEN statement in DSQL establishes a results table from the input parameters
specified in a previously declared and populated XSQLDA. A cursor must be opened before
data can be retrieved. The syntax for a DSQL OPEN is:
OPEN cursorname USING DESCRIPTOR sqldaname;

For example, the following statement opens the cursor, C, using the XSQLDA, InputSqlda:
EXEC SQL
OPEN C USING DESCRIPTOR InputSqlda;

Fetching rows with a DSQL cursor

DSQL uses the FETCH statement to retrieve rows from a results table. The rows are
retrieved according to specifications provided in a previously established and populated
extended SQL descriptor area (XSQLDA) that describes the user’s request. The syntax for
the DSQL FETCH statement is:
FETCH cursorname USING DESCRIPTOR descriptorname;

For example, the following C code fragment declares XSQLDA structures for input and
output, and illustrates how the output structure is used in a FETCH statement:
. . .
XSQLDA *InputSqlda, *OutputSqlda;
. . .

154 INTERBASE 6
JOINING TABLES

EXEC SQL
FETCH C USING DESCRIPTOR OutputSqlda;
. . .

For more information about creating and filling XSQLDA structures, and preparing DSQL
queries with PREPARE, see Chapter 13, “Using Dynamic SQL.”

Joining tables
Joins enable retrieval of data from two or more tables in a database with a single SELECT.
The tables from which data is to be extracted are listed in the FROM clause. Optional
syntax in the FROM clause can reduce the number of rows returned, and additional WHERE
clause syntax can further reduce the number of rows returned.
From the information in a SELECT that describes a join, InterBase builds a table that
contains the results of the join operation, the results table, sometimes also called a
dynamic or virtual table.
InterBase supports two types of joins:
g Inner joins link rows in tables based on specified join conditions, and return only those
rows that match the join conditions. There are three types of inner joins:
· Equi-joins link rows based on common values or equality relationships in the join
columns.
· Joins that link rows based on comparisons other than equality in the join columns.
There is not an officially recognized name for these types of joins, but for simplicity’s
sake they may be categorized as comparative joins, or non-equi-joins.
· Reflexive or self-joins, compare values within a column of a single table.
g Outer joins link rows in tables based on specified join conditions and return both rows
that match the join conditions, and all other rows from one or more tables even if they
do not match the join condition.
The most commonly used joins are inner joins, because they both restrict the data
returned, and show a clear relationship between two or more tables. Outer joins,
however, are useful for viewing joined rows against a background of rows that do not
meet the join conditions.

EMBEDDED SQL GUIDE 155

 CHAPTER 6 WORKING WITH DATA

Choosing join columns

How do you choose which columns to join? At a minimum, they must be of compatible
datatypes and of similar content. You cannot, for example, join a CHAR column to an
INTEGER column. A common and reliable criterion is to join the foreign key of one table
to its referenced primary key. Often, joins are made between identical columns in two
tables. For example, you might join the Job and Employee tables on their respective
job_code columns.
INTEGER, DECIMAL, NUMERIC, and FLOAT datatypes can be compared to one another
because they are all numbers. String values, like CHAR and VARCHAR, can only be
compared to other string values unless they contain ASCII values that are all numbers. The
CAST() function can be used to force translation of one InterBase datatype to another for
comparisons. For more information about CAST(), see “Using CAST( ) for datatype
conversions” on page 125.

IMPORTANT If a joined column contains a NULL value for a given row, InterBase does not include that
row in the results table unless performing an outer join.

Using inner joins

InterBase supports two methods for creating inner joins. For portability and compatibility
with existing SQL applications, InterBase continues to support the old SQL method for
specifying joins. In older versions of SQL, there is no explicit join language. An inner join
is specified by listing tables to join in the FROM clause of a SELECT, and the columns to
compare in the WHERE clause.
For example, the following join returns the department name, manager number, and
salary for any manager whose salary accounts for one third or more of the total salaries
of employees in that department.
EXEC SQL
DECLARE BIG_SAL CURSOR FOR
SELECT D.DEPARTMENT, D.MNGR_NO, E.SALARY
FROM DEPARTMENT D, EMPLOYEE E
WHERE D.MNGR_NO = E.EMP_NO
AND E.SALARY*2 >= (SELECT SUM(S.SALARY) FROM EMPLOYEE S
WHERE D.DEPT_NO = S.DEPT_NO)
ORDER BY D.DEPARTMENT;

InterBase also implements new, explicit join syntax based on SQL-92:

SELECT col [, col ...] | *

156 INTERBASE 6
JOINING TABLES

FROM <tablerefleft> [INNER] JOIN <tablerefright>

[ON <searchcondition>]
[WHERE <searchcondition>];

The join is explicitly declared in the FROM clause using the JOIN keyword. The table
reference appearing to the left of the JOIN keyword is called the left table, while the table
to the right of the JOIN is called the right table. The conditions of the join—the columns
from each table—are stated in the ON clause. The WHERE clause contains search
conditions that limit the number of rows returned. For example, using the new join
syntax, the previously described query can be rewritten as:
EXEC SQL
DECLARE BIG_SAL CURSOR FOR
SELECT D.DEPARTMENT, D.MNGR_NO, E.SALARY
FROM DEPARTMENT D INNER JOIN EMPLOYEE E
ON D.MNGR_NO = E.EMP_NO
WHERE E.SALARY*2 > (SELECT SUM(S.SALARY) FROM EMPLOYEE S
WHERE D.DEPT_NO = S.DEPT_NO)
ORDER BY D.DEPARTMENT;

The new join syntax offers several advantages. An explicit join declaration makes the
intention of the program clear when reading its source code.
The ON clause contains join conditions. The WHERE clause can contains conditions that
restrict which rows are returned.
The FROM clause also permits the use of table references, which can be used to construct
joins between three or more tables. For more information about nested joins, see “Using
nested joins” on page 161.

4 Creating equi-joins
An inner join that matches values in join columns is called an equi-join. Equi-joins are
among the most common join operations. The ON clause in an equi-join always takes the
form:
ON t1.column = t2.column

For example, the following join returns a list of cities around the world if the capital cities
also appear in the CITIES table, and also returns the populations of those cities:
EXEC SQL
DECLARE CAPPOP CURSOR FOR
SELECT COU.NAME, COU.CAPITAL, CIT.POPULATION
FROM COUNTRIES COU JOIN CITIES CIT ON CIT.NAME = COU.CAPITAL
WHERE COU.CAPITAL NOT NULL
ORDER BY COU.NAME;

EMBEDDED SQL GUIDE 157

 CHAPTER 6 WORKING WITH DATA

In this example, the ON clause specifies that the CITIES table must contain a city name
that matches a capital name in the COUNTRIES table if a row is to be returned. Note that
the WHERE clause restricts rows retrieved from the COUNTRIES table to those where the
CAPITAL column contains a value.

4 Joins based on comparison operators

Inner joins can compare values in join columns using other comparison operators besides
the equality operator. For example, a join might be based on a column in one table
having a value less than the value in a column in another table. The ON clause in a
comparison join always takes the form:
ON t1.column <operator> t2.column

where operator is a valid comparison operator. For a list of valid comparison operators,
see “Using comparison operators in expressions” on page 114.
For example, the following join returns information about provinces in Canada that are
larger than the state of Alaska in the United States:
EXEC SQL
DECLARE BIGPROVINCE CURSOR FOR
SELECT S.STATE_NAME, S.AREA, P.PROVINCE_NAME, P.AREA
FROM STATES S JOIN PROVINCE P ON P.AREA > S.AREA AND
P.COUNTRY = ’Canada’
WHERE S.STATE_NAME = ’Alaska’;

In this example, the first comparison operator in the ON clause tests to see if the area of
a province is greater than the area of any state (the WHERE clause restricts final output to
display only information for provinces that are larger in area than the state of Alaska).

4 Creating self-joins
A self-join is an inner join where a table is joined to itself to correlate columns of data.
For example, the RIVERS table lists rivers by name, and, for each river, lists the river into
which it flows. Not all rivers, of course, flow into other rivers. To discover which rivers
flow into other rivers, and what their names are, the
RIVERS table must be joined to itself:
EXEC SQL
DECLARE RIVERSTORIVERS CURSOR FOR
SELECT R1.RIVER, R2.RIVER
FROM RIVERS R1 JOIN RIVERS R2 ON R2.OUTFLOW = R1.RIVER
ORDER BY R1.RIVER, R2.SOURCE;

158 INTERBASE 6
JOINING TABLES

As this example illustrates, when a table is joined to itself, each invocation of the table
must be assigned a unique correlation name (R1 and R2 are correlation names in the
example). For more information about assigning and using correlation names, see
“Declaring and using correlation names” on page 134.

Using outer joins

Outer joins produce a results table that contains columns from every row in one table,
and a subset of rows from another table. Actually, one type of outer join returns all rows
from each table, but this type of join is used less frequently than other types. Outer join
syntax is very similar to that of inner joins:
SELECT col [, col ...] | *
FROM <tablerefleft> {LEFT | RIGHT | FULL} [OUTER] JOIN
<tablerefright> [ON <searchcondition>]
[WHERE <searchcondition>];

Outer join syntax requires that you specify the type of join to perform. There are three
possibilities:
g A left outer join retrieves all rows from the left table in a join, and retrieves any rows from
the right table that match the search condition specified in the ON clause.
g A right outer join retrieves all rows from the right table in a join, and retrieves any rows
from the left table that match the search condition specified in the ON clause.
g A full outer join retrieves all rows from both the left and right tables in a join regardless
of the search condition specified in the ON clause.
Outer joins are useful for comparing a subset of data to the background of all data from
which it is retrieved. For example, when listing those countries which contain the sources
of rivers, it may be interesting to see those countries which are not the sources of rivers
as well.

4 Using a left outer join

The left outer join is more commonly used than other types of outer joins. The following
left outer join retrieves those countries that contain the sources of rivers, and identifies
those countries that do not have NULL values in the R.RIVERS column:
EXEC SQL
DECLARE RIVSOURCE CURSOR FOR
SELECT C.COUNTRY, R.RIVER
FROM COUNTRIES C LEFT JOIN RIVERS R ON R.SOURCE = C.COUNTRY
ORDER BY C.COUNTRY;

EMBEDDED SQL GUIDE 159

 CHAPTER 6 WORKING WITH DATA

The ON clause enables join search conditions to be expressed in the FROM clause. The
search condition that follows the ON clause is the only place where retrieval of rows can
be restricted based on columns appearing in the right table. The WHERE clause can be
used to further restrict rows based solely on columns in the left (outer) table.

4 Using a right outer join

A right outer join retrieves all rows from the right table in a join, and only those rows
from the left table that match the search condition specified in the ON clause. The
following right outer join retrieves a list of rivers and their countries of origin, but also
reports those countries that are not the source of any river:
EXEC SQL
DECLARE RIVSOURCE CURSOR FOR
SELECT R.RIVER, C.COUNTRY
FROM RIVERS.R RIGHT JOIN COUNTRIES C ON C.COUNTRY = R.SOURCE
ORDER BY C.COUNTRY;

Tip Most right outer joins can be rewritten as left outer joins by reversing the order in which
tables are listed.

4 Using a full outer join

A full outer join returns all selected columns that do not contain NULL values from each
table in the FROM clause without regard to search conditions. It is useful to consolidate
similar data from disparate tables.
For example, several tables in a database may contain city names. Assuming triggers have
not been created that ensure that a city entered in one table is also entered in the others
to which it also applies, one of the only ways to see a list of all cities in the database is
to use full outer joins. The following example uses two full outer joins to retrieve the
name of every city listed in three tables, COUNTRIES, CITIES, and NATIONAL_PARKS:
EXEC SQL
DECLARE ALLCITIES CURSOR FOR
SELECT DISTINCT CIT.CITY, COU.CAPITAL, N.PARKCITY
FROM (CITIES CIT FULL JOIN COUNTRIES COU) FULL
JOIN NATIONAL_PARKS N;
This example uses a nested full outer join to process all rows from the CITIES and
COUNTRIES tables. The result table produced by that operation is then used as the left
table of the full outer join with the NATIONAL_PARKS table. For more information about
using nested joins, see “Using nested joins” on page 161.

160 INTERBASE 6
USING SUBQUERIES

Note In most databases where tables share similar or related information, triggers are
usually created to ensure that all tables are updated with shared information. For more
information about triggers, see the Data Definition Guide.

Using nested joins

The SELECT statement FROM clause can be used to specify any combination of available
tables or table references, parenthetical, nested joins whose results tables are created and
then processed as if they were actual tables stored in the database. Table references are
flexible and powerful, enabling the succinct creation of complex joins in a single location
in a SELECT.
For example, the following statement contains a parenthetical outer join that creates a
results table with the names of every city in the CITIES table even if the city is not
associated with a country in the COUNTRIES table. The results table is then processed as
the left table of an inner join that returns only those cities that have professional sports
teams of any kind, the name of the team, and the sport the team plays.
DECLARE SPORTSCITIES CURSOR FOR
SELECT COU.COUNTRY, C.CITY, T.TEAM, T.SPORT
FROM (CITIES CIT LEFT JOIN COUNTRIES COU ON COU.COUNTRY =
CIT.COUNTRY) INNER JOIN TEAMS T ON T.CITY = C.CITY
ORDER BY COU.COUNTRY;

For more information about left joins, see “Using outer joins” on page 159.

Using subqueries
A subquery is a parenthetical SELECT statement nested inside the WHERE clause of another
SELECT statement, where it functions as a search condition to restrict the number of rows
returned by the outer, or parent, query. A subquery can refer to the same table or tables
as its parent query, or to other tables.
The elementary syntax for a subquery is:
SELECT [DISTINCT] col [, col ...]
FROM <tableref> [, <tableref> ...]
WHERE {expression {[NOT] IN | comparison_operator}
| [NOT] EXISTS} (SELECT [DISTINCT] col [, col ...]
FROM <tableref> [, <tableref> ...]
WHERE <search_condition>);

EMBEDDED SQL GUIDE 161

 CHAPTER 6 WORKING WITH DATA

Because a subquery is a search condition, it is usually evaluated before its parent query,
which then uses the result to determine whether or not a row qualifies for retrieval. The
only exception is the correlated subquery, where the parent query provides values for the
subquery to evaluate. For more information about correlated subqueries, see
“Correlated subqueries” on page 163.
A subquery determines the search condition for a parent’s WHERE clause in one of the
following ways:
g Produces a list of values for evaluation by an IN operator in the parent query’s WHERE
clause, or where a comparison operator is modified by the ALL, ANY, or SOME operators.
g Returns a single value for use with a comparison operator.
g Tests whether or not data meets conditions specified by an EXISTS operator in the parent
query’s WHERE clause.
Subqueries can be nested within other subqueries as search conditions, establishing a
chain of parent/child queries.

Simple subqueries
A subquery is especially useful for extracting data from a single table when a self-join is
inadequate. For example, it is impossible to retrieve a list of those countries with a larger
than average area by joining the COUNTRIES table to itself. A subquery, however, can easily
return that information.
EXEC SQL
DECLARE LARGECOUNTRIES CURSOR FOR
SELECT COUNTRY, AREA
FROM COUNTRIES
WHERE AREA > (SELECT AVG(AREA) FROM COUNTRIES);
ORDER BY AREA;

In this example, both the query and subquery refer to the same table. Queries and
subqueries can refer to different tables, too. For example, the following query refers to
the CITIES table, and includes a subquery that refers to the COUNTRIES table:
EXEC SQL
DECLARE EUROCAPPOP CURSOR FOR
SELECT CIT.CITY, CIT.POPULATION
FROM CITIES CIT
WHERE CIT.CITY IN (SELECT COU.CAPITAL FROM COUNTRIES COU
WHERE COU.CONTINENT = ’Europe’)
ORDER BY CIT.CITY;

162 INTERBASE 6
USING SUBQUERIES

This example uses correlation names to distinguish between tables even though the
query and subquery reference separate tables. Correlation names are only necessary
when both a query and subquery refer to the same tables and those tables share column
names, but it is good programming practice to use them. For more information about
using correlation names, see “Declaring and using correlation names” on page 134.

Correlated subqueries
A correlated subquery is a subquery that depends on its parent query for the values it
evaluates. Because each row evaluated by the parent query is potentially different, the
subquery is executed once for each row presented to it by the parent query.
For example, the following query lists each country for which there are three or more
cities stored in the CITIES table. For each row in the COUNTRIES table, a country name is
retrieved in the parent query, then used in the comparison operation in the subquery’s
WHERE clause to verify if a city in the CITIES table should be counted by the COUNT()
function. If COUNT() exceeds 2 for a row, the row is retrieved.
EXEC SQL
DECLARE TRICITIES CURSOR FOR
SELECT COUNTRY
FROM COUNTRIES COU
WHERE 3 <= (SELECT COUNT (*)
FROM CITIES CIT
WHERE CIT.CITY = COU.CAPITAL);

Simple and correlated subqueries can be nested and mixed to build complex queries. For
example, the following query retrieves the country name, capital city, and largest city of
countries whose areas are larger than the average area of countries that have at least one
city within 30 meters of sea level:
EXEC SQL
DECLARE SEACOUNTRIES CURSOR FOR
SELECT CO1.COUNTRY, C01.CAPITAL, CI1.CITY
FROM COUNTRIES C01, CITIES CI1
WHERE CO1.COUNTRY = CI1.COUNTRY AND CI1.POPULATION =
(SELECT MAX(CI2.POPULATION)
FROM CITIES CI2 WHERE CI2.COUNTRY = CI1.COUNTRY)
AND CO1.AREA >
(SELECT AVG (CO2.AREA)
FROM COUNTRIES C02 WHERE EXISTS

EMBEDDED SQL GUIDE 163

 CHAPTER 6 WORKING WITH DATA

(SELECT *
FROM CITIES CI3 WHERE CI3.COUNTRY = CO2.COUNTRY
AND CI3.ALTITUDE <= 30));

When a table is separately searched by queries and subqueries, as in this example, each
invocation of the table must establish a separate correlation name for the table. Using
correlation names is the only method to assure that column references are associated
with appropriate instances of their tables. For more information about correlation names,
see “Declaring and using correlation names” on page 134.

Inserting data
New rows of data are added to one table at a time with the INSERT statement. To insert
data, a user or stored procedure must have INSERT privilege for a table.
The INSERT statement enables data insertion from two different sources:
g A VALUES clause that contains a list of values to add, either through hard-coded values, or
host-language variables.
g A SELECT statement that retrieves values from one table to add to another.
The syntax of INSERT is as follows:
INSERT [TRANSACTION name] INTO table [(col [, col ...])]
{VALUES (<val>[:ind] [, <val>[:ind] ...])
| SELECT <clause>};

The list of columns into which to insert values is optional in DSQL applications. If it is
omitted, then values are inserted into a table’s columns according to the order in which
the columns were created. If there are more columns than values, the remaining columns
are filled with zeros.

Using VALUES to insert columns

Use the VALUES clause to add a row of specific values to a table, or to add values entered
by a user at run time. The list of values that follows the keyword can come from either
from host-language variables, or from hard-coded assignments.
For example, the following statement adds a new row to the DEPARTMENT table using
hard-coded value assignments:
EXEC SQL

164 INTERBASE 6
INSERTING DATA

INSERT INTO DEPARTMENT (DEPT_NO, DEPARTMENT)

VALUES (7734, ’Marketing’);

Because the DEPARTMENT table contains additional columns not specified in the INSERT,
NULL values are assigned to the missing fields.
The following C code example prompts a user for information to add to the DEPARTMENT
table, and inserts those values from host variables:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char department[26], dept_no[16];
int dept_num;
EXEC SQL
END DECLARE SECTION;
. . .
printf("Enter name of department: ");
gets(department);
printf("\nEnter department number: ");
dept_num = atoi(gets(dept_no));
EXEC SQL
INSERT INTO COUNTRIES (DEPT_NO, DEPARTMENT)
VALUES (:dept_num, :department);

When host variables are used in the values list, they must be preceded by colons (:) so
that SQL can distinguish them from table column names.

Using SELECT to insert columns

To insert values from one table into another row in the same table or into a row in
another table, use a SELECT statement to specify a list of insertion values. For example,
the following INSERT statement copies DEPARTMENT and BUDGET information about the
publications department from the OLDDEPT table to the DEPARTMENT table. It also
illustrates how values can be hard-coded into a SELECT statement to substitute actual
column data.
EXEC SQL
INSERT INTO DEPARTMENTS (DEPT_NO, DEPARTMENT, BUDGET)
SELECT DEPT_NO, ’Publications’, BUDGET
FROM OLDDEPT
WHERE DEPARTMENT = ’Documentation’;

EMBEDDED SQL GUIDE 165

 CHAPTER 6 WORKING WITH DATA

The assignments in the SELECT can include arithmetic operations. For example, suppose
an application keeps track of employees by using an employee number. When a new
employee is hired, the following statement inserts a new employee row into the EMPLOYEE
table, and assigns a new employee number to the row by using a SELECT statement to find
the current maximum employee number and adding one to it. It also reads values for
LAST_NAME and FIRST_NAME from the host variables, lastname, and firstname.
EXEC SQL
INSERT INTO EMPLOYEE (EMP_NO, LAST_NAME, FIRST_NAME)
SELECT (MAX(EMP_NO) + 1, :lastname, :firstname)
FROM EMPLOYEE;

Inserting rows with NULL column values

Sometimes when a new row is added to a table, values are not necessary or available for
all its columns. In these cases, a NULL value should be assigned to those columns when
the row is inserted. There are three ways to assign a NULL value to a column on insertion:
g Ignore the column.
g Assign a NULL value to the column. This is standard SQL practice.
g Use indicator variables.

4 Ignoring a column
A NULL value is assigned to any column that is not explicitly specified in an INTO clause.
When InterBase encounters an unreferenced column during insertion, it sets a flag for
the column indicating that its value is unknown. For example, the DEPARTMENT table
contains several columns, among them HEAD_DEPT, MNGR_NO, and BUDGET. The
following INSERT does not provide values for these columns:
EXEC SQL
INSERT INTO DEPARTMENT (DEPT_NO, DEPARTMENT)
VALUES (:newdept_no, :newdept_name);

Because HEAD_DEPT, MNGR_NO, and BUDGET are not specified, InterBase sets the NULL
value flag for each of these columns.
Note If a column is added to an existing table, InterBase sets a NULL value flag for all
existing rows in the table.

166 INTERBASE 6
INSERTING DATA

4 Assigning a NULL value to a column

When a specific value is not provided for a column on insertion, it is standard SQL
practice to assign a NULL value to that column. In InterBase a column is set to NULL by
specifying NULL for the column in the INSERT statement.
For example, the following statement stores a row into the DEPARTMENT table, assigns the
values of host variables to some columns, and assigns a NULL value to other columns:
EXEC SQL
INSERT INTO DEPARTMENT
(DEPT_NO, DEPARTMENT, HEAD_DEPT, MNGR_NO, BUDGET,
LOCATION, PHONE_NO)
VALUES (:dept_no, :dept_name, NULL, NULL, 1500000, NULL, NULL);

4 Using indicator variables

Another method for trapping and assigning NULL values—through indicator variables—
is necessary in applications that prompt users for data, where users can choose not to
enter values. By default, when InterBase stores new data, it stores zeroes for NULL
numeric data, and spaces for NULL character data. Because zeroes and spaces may be
valid data, it becomes impossible to distinguish missing data in the new row from actual
zeroes and spaces.
To trap missing data with indicator variables, and store NULL value flags, follow these
steps:
1. Declare a host-language variable to use as an indicator variable.
2. Test a value entered by the user and set the indicator variable to one of the
following values:

0 The host-language variable contains data.

–1 The host-language variable does not contain data.

3. Associate the indicator variable with the host variable in the INSERT statement
using the following syntax:
INSERT INTO table (<col> [, <col> ...])
VALUES (:variable [INDICATOR] :indicator
[, :variable [INDICATOR] :indicator ...]);

Note The INDICATOR keyword is optional.

EMBEDDED SQL GUIDE 167

 CHAPTER 6 WORKING WITH DATA

For example, the following C code fragment prompts the user for the name of a
department, the department number, and a budget for the department. It tests that the
user has entered a budget. If not, it sets the indicator variable, bi, to –1. Otherwise, it sets
bi to 0. Finally, the program INSERTS the information into the DEPARTMENT table. If the
indicator variable is –1, then no actual data is stored in the BUDGET column, but a flag is
set for the column indicating that the value is NULL
. . .
EXEC SQL
BEGIN DECLARE SECTION;
short bi; /* indicator variable declaration */
char department[26], dept_no_ascii[26], budget_ascii[26];
long num_val; /* host variable for inserting budget */
short dept_no;
EXEC SQL
END DECLARE SECTION;
. . .
printf("Enter new department name: ");
gets(cidepartment);
printf("\nEnter department number: ");
gets(dept_no_ascii);
printf("\nEnter department’s budget: ");
gets(budget_ascii);
if (budget_ascii = "")
{
bi = -1; num_val = 0;
}
else
{
bi = 0;
num_val = atoi(budget_ascii);
}
dept_no = atoi(dept_no_ascii);
EXEC SQL
INSERT INTO DEPARTMENT (DEPARTMENT, DEPT_NO, BUDGET)
VALUES (:department, :dept_no, :num_val INDICATOR :bi);
. . .

Indicator status can also be determined for data retrieved from a table. For information
about trapping NULL values retrieved from a table, see “Retrieving indicator status” on
page 148.

168 INTERBASE 6
INSERTING DATA

Inserting data through a view

New rows can be inserted through a view if the following conditions are met:
g The view is updatable. For a complete discussion of updatable views, see the Data
Definition Guide.
g The view is created using the WITH CHECK OPTION.
g A user or stored procedure has INSERT privilege for the view.
Values can only be inserted through a view for those columns named in the view.
InterBase stores NULL values for unreferenced columns. For example, suppose the view,
PART_DEPT, is defined as follows:
EXEC SQL
CREATE VIEW PART_DEPT
(DEPARTMENT, DEPT_NO, BUDGET)
AS SELECT DEPARTMENT, DEPT_NO, BUDGET
FROM DEPARTMENT
WHERE DEPT_NO NOT NULL AND BUDGET > 50000
WITH CHECK OPTION;

Because PART_DEPT references a single table, DEPARTMENT, new data can be inserted for
the DEPARTMENT, DEPT_NO, and BUDGET columns. The WITH CHECK OPTION assures that
all values entered through the view fall within ranges of values that can be selected by
this view. For example, the following statement inserts a new row for the Publications
department through the PART_DEPT view:
EXEC SQL
INSERT INTO PART_DEPT (DEPARTMENT, DEPT_NO, BUDGET)
VALUES (’Publications’, ’7735’, 1500000);

InterBase inserts NULL values for all other columns in the DEPARTMENT table that are not
available directly through the view.
For information about creating a view, see Chapter 5, “Working with Data Definition
Statements.” For the complete syntax of CREATE VIEW, see the Language Reference.
Note See the chapter on triggers in the Data Definition Guide for tips on using triggers
to update non-updatable views.

Specifying transaction names in an INSERT

InterBase enables an SQL application to run simultaneous transactions if:

EMBEDDED SQL GUIDE 169

 CHAPTER 6 WORKING WITH DATA

g Each transaction is first named with a SET TRANSACTION statement. For a complete
discussion of transaction handling and naming, see Chapter 4, “Working with
Transactions.”
g Each data manipulation statement (SELECT, INSERT, UPDATE, DELETE, DECLARE, OPEN,
FETCH, and CLOSE) specifies a TRANSACTION clause that identifies the name of the
transaction under which it operates.
g SQL statements are not dynamic (DSQL). DSQL does not support user-specified
transaction names.
With INSERT, the TRANSACTION clause intervenes between the INSERT keyword and the list
of columns to insert, as in the following syntax fragment:
INSERT TRANSACTION name INTO table (col [, col ...])

The TRANSACTION clause is optional in single-transaction programs. It must be used in a

multi-transaction program unless a statement operates under control of the default
transaction, GDS__TRANS. For example, the following INSERT is controlled by the
transaction, T1:
EXEC SQL
INSERT TRANSACTION T1 INTO DEPARTMENT (DEPARTMENT, DEPT_NO, BUDGET)
VALUES (:deptname, :deptno, :budget INDICATOR :bi);

Updating data
To change values for existing rows of data in a table, use the UPDATE statement. To update
a table, a user or procedure must have UPDATE privilege for it. The syntax of UPDATE is:
UPDATE [TRANSACTION name] table
SET col = <assignment> [, col = <assignment> ...]
WHERE <search_condition> | WHERE CURRENT OF cursorname;

UPDATE changes values for columns specified in the SET clause; columns not listed in the
SET clause are not changed. A single UPDATE statement can be used to modify any
number of rows in a table. For example, the following statement modifies a single row:
EXEC SQL
UPDATE DEPARTMENT
SET DEPARTMENT = ’Publications’
WHERE DEPARTMENT = ’Documentation’;

170 INTERBASE 6
UPDATING DATA

The WHERE clause in this example targets a single row for update. If the same change
should be propagated to a number of rows in a table, the WHERE clause can be more
general. For example, to change all occurrences of “Documentation” to “Publications” for
all departments in the DEPARTMENT table where DEPARTMENT equals “Documentation,”
the UPDATE statement would be as follows:
EXEC SQL
UPDATE DEPARTMENT
SET DEPARTMENT = ’Publications’
WHERE DEPARTMENT = ’Documentation’;

Using UPDATE to make the same modification to a number of rows is sometimes called a
mass update, or a searched update.
The WHERE clause in an UPDATE statement can contain a subquery that references one or
more other tables. For a complete discussion of subqueries, see “Using subqueries” on
page 161.

Updating multiple rows

There are two basic methods for modifying rows:
g The searched update method, where the same changes are applied to a number of rows,
is most useful for automated updating of rows without a cursor.
g The positioned update method, where rows are retrieved through a cursor and updated
row by row, is most useful for enabling users to enter different changes for each row
retrieved.
A searched update is easier to program than a positioned update, but also more limited
in what it can accomplish.

4 Using a searched update

Use a searched update to make the same changes to a number of rows. The UPDATE SET
clause specifies the actual changes that are to be made to columns for each row that
matches the search condition specified in the WHERE clause. Values to set can be specified
as constants or variables.
For example, the following C code fragment prompts for a country name and a
percentage change in population, then updates all cities in that country with the new
population:
. . .
EXEC SQL
BEGIN DECLARE SECTION;

EMBEDDED SQL GUIDE 171

 CHAPTER 6 WORKING WITH DATA

char country[26], asciimult[10];

int multiplier;
EXEC SQL
END DECLARE SECTION;
. . .
main ()
{
printf("Enter country with city populations needing adjustment: ");
gets(country);
printf("\nPercent change (100%% to -100%%:");
gets(asciimult);
multiplier = atoi(asciimult);
EXEC SQL
UPDATE CITIES
SET POPULATION = POPULATION * (1 + :multiplier / 100)
WHERE COUNTRY = :country;

if (SQLCODE && (SQLCODE != 100))

{
isc_print_sqlerr(SQLCODE, isc_status);
EXEC SQL
ROLLBACK RELEASE;
}
else
{
EXEC SQL
COMMIT RELEASE;
}
}

IMPORTANT Searched updates cannot be performed on arrays of datatypes.

4 Using a positioned update

Use cursors to select rows for update when prompting users for changes on a row-by-row
basis, and displaying pre- or post-modification values between row updates. Updating
through a cursor is a seven-step process:
1. Declare host-language variables needed for the update operation.
2. Declare a cursor describing the rows to retrieve for update, and include the
FOR UPDATE clause in DSQL. For more information about declaring and using
cursors, see “Selecting multiple rows” on page 144.

172 INTERBASE 6
UPDATING DATA

3. Open the cursor.

4. Fetch a row.
5. Display current values and prompt for new values.
6. Update the currently selected row using the WHERE CURRENT OF clause.
7. Repeat steps 3 to 7 until all selected rows are updated.
For example, the following C code fragment updates the POPULATION column by
user-specified amounts for cities in the CITIES table that are in a country also specified by
the user:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char country[26], asciimult[10];
int multiplier;
EXEC SQL
END DECLARE SECTION;
. . .
main ()
{
EXEC SQL
DECLARE CHANGEPOP CURSOR FOR
SELECT CITY, POPULATION
FROM CITIES
WHERE COUNTRY = :country;

printf("Enter country with city populations needing adjustment: ");

gets(country);
EXEC SQL
OPEN CHANGEPOP;
EXEC SQL
FETCH CHANGEPOP INTO :country;
while(!SQLCODE)
{
printf("\nPercent change (100%% to -100%%:");
gets(asciimult);
multiplier = atoi(asciimult);
EXEC SQL
UPDATE CITIES
SET POPULATION = POPULATION * (1 + :multiplier / 100)
WHERE CURRENT OF CHANGEPOP;

EMBEDDED SQL GUIDE 173

 CHAPTER 6 WORKING WITH DATA

EXEC SQL
FETCH CHANGEPOP INTO :country;
if (SQLCODE && (SQLCODE != 100))
{
isc_print_sqlerr(SQLCODE, isc_status);
EXEC SQL
ROLLBACK RELEASE;
exit(1);
}
}
EXEC SQL
COMMIT RELEASE;
}

IMPORTANT Using FOR UPDATE with a cursor causes rows to be fetched from the database one at a
time. If FOR UPDATE is omitted, rows are fetched in batches.

NULLing columns with UPDATE

To set a column’s value to NULL during update, specify a NULL value for the column in the
SET clause. For example, the following UPDATE sets the budget of all departments without
managers to NULL:
EXEC SQL
UPDATE DEPARTMENT
SET BUDGET = NULL
WHERE MNGR_NO = NULL;

Updating through a view

Existing rows can be updated through a view if the following conditions are met:
g The view is updatable. For a complete discussion of updatable views, see the Data
Definition Guide.
g The view is created using the WITH CHECK OPTION.
g A user or stored procedure has UPDATE privilege for the view.
Values can only be updated through a view for those columns named in the view. For
example, suppose the view, PART_DEPT, is defined as follows:
EXEC SQL

174 INTERBASE 6
UPDATING DATA

CREATE VIEW PART_DEPT

(DEPARTMENT, NUMBER, BUDGET)
AS SELECT DEPARTMENT, DEPT_NO, BUDGET
FROM DEPARTMENT
WITH CHECK OPTION;

Because PART_DEPT references a single table, data can be updated for the columns named
in the view. The WITH CHECK OPTION assures that all values entered through the view fall
within ranges prescribed for each column when the DEPARTMENT table was created. For
example, the following statement updates the budget of the Publications department
through the PART_DEPT view:
EXEC SQL
UPDATE PART_DEPT
SET BUDGET = 2505700
WHERE DEPARTMENT = ’Publications’;

For information about creating a view, see Chapter 5, “Working with Data Definition
Statements.” For the complete syntax of CREATE VIEW, see the Language Reference.
Note See the chapter on triggers in the Data Definition Guide for tips on using triggers
to update non-updatable views.

Specifying transaction names in UPDATE

InterBase enables an SQL application to run simultaneous transactions if:
g Each transaction is first named with a SET TRANSACTION statement. For a complete
discussion of transaction handling and naming, see Chapter 4, “Working with
Transactions.”
g Each data manipulation statement (SELECT, INSERT, UPDATE, DELETE, DECLARE, OPEN,
FETCH, and CLOSE) specifies a TRANSACTION clause that identifies the name of the
transaction under which it operates.
g SQL statements are not dynamic (DSQL). DSQL does not support multiple simultaneous
transactions.
In UPDATE, the TRANSACTION clause intervenes between the UPDATE keyword and the
name of the table to update, as in the following syntax:
UPDATE [TRANSACTION name] table
SET col = <assignment> [, col = <assignment> ...]
WHERE <search_condition> | WHERE CURRENT OF cursorname;

EMBEDDED SQL GUIDE 175

 CHAPTER 6 WORKING WITH DATA

The TRANSACTION clause must be used in multi-transaction programs, but is optional in

single-transaction programs or in programs where only one transaction is open at a time.
For example, the following UPDATE is controlled by the transaction, T1:
EXEC SQL
UPDATE TRANSACTION T1 DEPARTMENT
SET BUDGET = 2505700
WHERE DEPARTMENT = ’Publications’;

Deleting data
To remove rows of data from a table, use the DELETE statement. To delete rows a user or
procedure must have DELETE privilege for the table.
The syntax of DELETE is:
DELETE [TRANSACTION name] FROM table
WHERE <search_condition> | WHERE CURRENT OF cursorname;

DELETE irretrievably removes entire rows from the table specified in the FROM clause,
regardless of each column’s datatype.
A single DELETE can be used to remove any number of rows in a table. For example, the
following statement removes the single row containing “Channel Marketing” from the
DEPARTMENT table:
EXEC SQL
DELETE FROM DEPARTMENT
WHERE DEPARTMENT = ’Channel Marketing’;

The WHERE clause in this example targets a single row for update. If the same deletion
criteria apply to a number of rows in a table, the WHERE clause can be more general. For
example, to remove all rows from the DEPARTMENT table with BUDGET values
< $1,000,000, the DELETE statement would be as follows:
EXEC SQL
DELETE FROM DEPARTMENT
WHERE BUDGET < 1000000;

Using DELETE to remove a number of rows is sometimes called a mass delete.

The WHERE clause in a DELETE statement can contain a subquery that references one or
more other tables. For a discussion of subqueries, see “Using subqueries” on page 161.

176 INTERBASE 6
DELETING DATA

Deleting multiple rows

There are two methods for modifying rows:
g The searched delete method, where the same deletion condition applies to a number of
rows, is most useful for automated removal of rows.
g The positioned delete method, where rows are retrieved through a cursor and deleted row
by row, is most useful for enabling users to choose which rows that meet certain
conditions should be removed.
A searched delete is easier to program than a positioned delete, but less flexible.

4 Using a searched delete

Use a searched delete to remove a number of rows that match a condition specified in
the WHERE clause. For example, the following C code fragment prompts for a country
name, then deletes all rows that have cities in that country:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char country[26];
EXEC SQL
END DECLARE SECTION;
. . .
main ()
{
printf("Enter country with cities to delete: ");
gets(country);
EXEC SQL
DELETE FROM CITIES
WHERE COUNTRY = :country;

if(SQLCODE && (SQLCODE != 100))

{
isc_print_sqlerr(SQLCODE, isc_status);
EXEC SQL
ROLLBACK RELEASE;
}
else
{
EXEC SQL
COMMIT RELEASE;

EMBEDDED SQL GUIDE 177

 CHAPTER 6 WORKING WITH DATA

}
}

4 Using a positioned delete

Use cursors to select rows for deletion when users should decide deletion on a
row-by-row basis, and displaying pre- or post-modification values between row updates.
Updating through a cursor is a seven-step process:
1. Declare host-language variables needed for the delete operation.
2. Declare a cursor describing the rows to retrieve for possible deletion, and
include the FOR UPDATE clause. For more information about declaring and
using cursors, see “Selecting multiple rows” on page 144.
3. Open the cursor.
4. Fetch a row.
5. Display current values and prompt for permission to delete.
6. Delete the currently selected row using the WHERE CURRENT OF clause to
specify the name of the cursor.
7. Repeat steps 3 to 7 until all selected rows are deleted.
For example, the following C code deletes rows in the CITIES table that are in North
America only if a user types Y when prompted:
. . .
EXEC SQL
BEGIN DECLARE SECTION;
char cityname[26];
EXEC SQL
END DECLARE SECTION;
char response[5];
. . .
main ()
{
EXEC SQL
DECLARE DELETECITY CURSOR FOR
SELECT CITY,
FROM CITIES
WHERE CONTINENT = ’North America’;
EXEC SQL
OPEN DELETECITY;
while (!SQLCODE)
{

178 INTERBASE 6
DELETING DATA

EXEC SQL
FETCH DELETECITY INTO :cityname;
if (SQLCODE)
{
if (SQLCODE == 100)
{
printf(’Deletions complete.’);
EXEC SQL
COMMIT;
EXEC SQL
CLOSE DELETECITY;
EXEC SQL
DISCONNECT ALL:
}
isc_print_sqlerr(SQLCODE, isc_status);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT ALL;
exit(1);
}
printf("\nDelete %s (Y/N)?", cityname);
gets(response);
if(response[0] == ’Y’ || response == ’y’)
{
EXEC SQL
DELETE FROM CITIES
WHERE CURRENT OF DELETECITY;
if(SQLCODE && (SQLCODE != 100))
{
isc_print_sqlerr(SQLCODE, isc_status);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT;
exit(1);
}
}
}

EMBEDDED SQL GUIDE 179

 CHAPTER 6 WORKING WITH DATA

Deleting through a view

You can delete entire rows through a view if the following conditions are met:
g The view is updatable. For a complete discussion of updatable views, see the Data
Definition Guide.
g A user or stored procedure has DELETE privilege for the view.
For example, the following statement deletes all departments with budgets under
$1,000,000, from the DEPARTMENT table through the PART_DEPT view:
EXEC SQL
DELETE FROM PART_DEPT
WHERE BUDGET < 1000000;

For information about creating a view, see Chapter 5, “Working with Data Definition
Statements.” For CREATE VIEW syntax, see the Language Reference.
Note See the chapter on triggers in the Data Definition Guide for tips on using triggers
to delete through non-updatable views.

Specifying transaction names in a DELETE

InterBase enables an SQL application to run simultaneous transactions if:
g Each transaction is first named with a SET TRANSACTION statement. For a complete
discussion of transaction handling and naming, see Chapter 4, “Working with
Transactions.”
g Each data manipulation statement (SELECT, INSERT, UPDATE, DELETE, DECLARE, OPEN,
FETCH, and CLOSE) specifies a TRANSACTION clause that identifies the name of the
transaction under which it operates.
g SQL statements are not dynamic (DSQL). DSQL does not support multiple simultaneous
transactions.
For DELETE, the TRANSACTION clause intervenes between the DELETE keyword and the
FROM clause specifying the table from which to delete:
DELETE TRANSACTION name FROM table ...

180 INTERBASE 6
DELETING DATA

The TRANSACTION clause is optional in single-transaction programs or in programs where

only one transaction is open at a time. It must be used in a multi-transaction program.
For example, the following DELETE is controlled by the transaction, T1:
EXEC SQL
DELETE TRANSACTION T1 FROM PART_DEPT
WHERE BUDGET < 1000000;

EMBEDDED SQL GUIDE 181

 CHAPTER 6 WORKING WITH DATA

182 INTERBASE 6
 CHAPTER

Working with Dates

Chapter7
7
Most host languages do not support the DATE, TIME, and TIMESTAMP datatypes. Instead,
they treat dates as strings or structures. InterBase supports DATE and TIME datatypes that
are stored as a single long integers, and a TIMESTAMP datatype that is stored in tables as
two long integers. An InterBase DATE datatype includes information about year, month,
and day of the month, the TIME datatype includes information about time, and the
TIMESTAMP datatype is a combination of the two.
This chapter discusses how to SELECT, INSERT, and UPDATE dates from tables in SQL
applications using the following isc call interface routines:
g isc_decode_sql_date() converts the InterBase internal date format to the C time structure
g isc_encode_sql_date() converts the C time structure to the internal InterBase date format
g isc_decode_sql_time() toconverts the InterBase internal time format to the C time
structure
g sql_encode_sql_time() converts the C time structure to the internal InterBase time format
g isc_decode_timestamp() converts the InterBase internal timestamp format to the C time
structure; this was formerly isc_decode_date()
g isc_encode_timestamp() converts the C time structure to the InterBase internal
timestamp format; this was formerly isc_encode_date()
See the API Guide for a description of each of these functions.

EMBEDDED SQL GUIDE 183

 CHAPTER 7 WORKING WITH DATES

The chapter also discusses how to use the CAST() function to translate DATE, TIME, and
TIMESTAMP datatypes into each other or into CHAR datatypes and back again, and how to
use the DATE literals, NOW and TODAY when selecting and inserting dates.

Selecting dates and times

To select a date and time (timestamp) from a table, and convert it to a form usable in a
C language program, follow these steps:
1. Create a host variable for a C time structure. Most C and C++ compilers
provide a typedef declaration, struct tm, for the C time structure in the time.h
header file. The following C code includes that header file, and declares a
variable of type struct tm:
#include <time.h>;
. . .
struct tm hire_time;
. . .

To create host-language time structures in languages other than C and C++, see the
host-language reference manual.
2. Create a host variable of type ISC_TIMESTAMP. For example, the host-variable
declaration might look like this:
ISC_TIMESTAMP hire_date;

The ISC_TIMESTAMP structure is automatically declared for programs when they are
preprocessed with gpre, but the programmer must declare actual host-language
variables of type ISC_TIMESTAMP.
3. Retrieve a timestamp from a table into the ISC_TIMESTAMP variable. For
example,
EXEC SQL
SELECT LAST_NAME, FIRST_NAME, DATE_OF_HIRE
INTO :lname, :fname, :hire_date
FROM EMPLOYEE
WHERE LAST_NAME = ’Smith’ AND FIRST_NAME = ’Margaret’;

184 INTERBASE 6
INSERTING DATES AND TIMES

Convert the isc_timestamp variable into a numeric Unix format with the InterBase
function, isc_decode_timestamp(). This function is automatically declared for
programs when they are preprocessed with gpre. isc_decode_timestamp() requires
two parameters: the address of the isc_timestamp host-language variable, and the
address of the struct tm host-language variable. For example, the following code
fragment coverts hire_date to hire_time:
isc_decode_timestamp(&hire_date, &hire_time);

Inserting dates and times

To insert a date and time (timestamp) in a table, it must be converted from the
host-language format into InterBase format, and then stored. To perform the conversion
and insertion in a C program, follow these steps:
1. Create a host variable for a C time structure. Most C and C++ compilers
provide a typedef declaration, struct tm, for the C time structure in the time.h
header file. The following C code includes that header file, and declares a
struct tm variable, hire_time:
#include <time.h>;
. . .
struct tm hire_time;
. . .

To create host-language time structures in languages other than C and C++, see the
host-language reference manual.
2. Create a host variable of type ISC_TIMESTAMP, for use by InterBase. For
example, the host-variable declaration might look like this:
ISC_TIMESTAMP mydate;

The ISC_TIMESTAMP structure is automatically declared for programs when they are
preprocessed with gpre, but the programmer must declare actual host-language
variables of type ISC_TIMESTAMP.
3. Put date information into hire_time.

EMBEDDED SQL GUIDE 185

 CHAPTER 7 WORKING WITH DATES

4. Use the InterBase isc_encode_timestamp() function to convert the

information in hire_time into InterBase internal format and store that
formatted information in the ISC_TIMESTAMP host variable (hire_date in the
example). This function is automatically declared for programs when they
are preprocessed with gpre. isc_encode_timestamp() requires two
parameters, the address of the Unix time structure, and the address of the
ISC_TIMESTAMP host-language variable.
For example, the following code converts hire_time to hire_date:
isc_encode_timestamp(&hire_time, &hire_date);

5. Insert the date into a table. For example,

EXEC SQL
INSERT INTO EMPLOYEE (EMP_NO, DEPARTMENT, DATE_OF_HIRE)
VALUES (:emp_no, :deptname, :hire_date);

Updating dates and times

To update a DATE, TIME, or TIMESTAMP datatype in a table, you must convert it from the
host-language format into InterBase format, and then store it. To convert a host variable
into InterBase format, see “Inserting dates and times” on page 185. The actual update
is performed using an UPDATE statement. For example,
EXEC SQL
UPDATE EMPLOYEE
SET DATE_OF_HIRE = :hire_date
WHERE DATE_OF_HIRE < ’1 JAN 1994’

Using CAST( ) to convert dates and times

The built-in CAST() function can be used in SELECT statements to translate between the
following types:
g DATE, TIME, or TIMESTAMP datatype into a CHAR datatype
g CHAR datatype into a DATE, TIME, or TIMESTAMP datatype
g DATE or TIME datatype into a TIMESTAMP datatype
g TIMESTAMP datatype into a DATE or TIME datatype

186 INTERBASE 6
USING CAST( ) TO CONVERT DATES AND TIMES

Typically, CAST() is used in the WHERE clause to compare different datatypes. The syntax
for CAST() is:
CAST (<value> AS <datatype>)

In the following WHERE clause, CAST() is translates a CHAR datatype, INTERVIEW_DATE, to

a DATE datatype to compare against a DATE datatype, HIRE_DATE:
… WHERE HIRE_DATE = CAST(INTERVIEW_DATE AS DATE);

In the next example, CAST() translates a DATE datatype into a CHAR datatype:
… WHERE CAST(HIRE_DATE AS CHAR) = INTERVIEW_DATE;

CAST() also can be used to compare columns with different datatypes in the same table,
or across tables.
The following two sections show the possible conversions to and from datetime (DATE,
TIME, and TIMESTAMP) datatypes and other SQL datatypes.
For more information about CAST(), see Chapter 6, “Working with Data.”

Casting from SQL datatypes to datetime datatypes

The following table shows the SQL datatypes from which the DATE, TIME, and TIMESTAMP
datatypes can be cast.

CAST datatype below TIMESTAMP DATE TIME

TO datatype at right
SMALLINT Error Error Error
INTEGER
FLOAT
DOUBLE PRECISION
NUMERIC
DECIMAL

VARCHAR(n) Succeeds it the string is in Succeeds it the string is in Succeeds it the string is in
CHAR(n) the following format: the following format: the following format:
CSTRING(n)
YYYY-MM-DD YYYY-MM-DD HH:MM:SS.thou
HH:MM:SS.thou
BLOB Error Error Error
TABLE 7.1 Casting from SQL datatypes to datetime datatypes

EMBEDDED SQL GUIDE 187

 CHAPTER 7 WORKING WITH DATES

CAST datatype below TIMESTAMP DATE TIME

TO datatype at right
TIMESTAMP Always succeeds Succeeds: date portion of Succeeds: time portion of
TIMESTAMP TIMESTAMP

DATE Succeeds: time portion of Always succeeds Error

TIMESTAMP set to 0:0:0.0000

TIME Succeeds: date portion of Error Always succeeds

TIMESTAMP set to the base-0
date (17 November, 1858)
TABLE 7.1 Casting from SQL datatypes to datetime datatypes

Casting from datetime datatypes to other SQL datatypes

The following table shows the SQL datatypes into which the DATE, TIME, and TIMESTAMP
datatypes can be cast.

CAST datatype at right TIMESTAMP DATE TIME

TO datatype below
SMALLINT Error Error Error
INTEGER
FLOAT
DOUBLE PRECISION
NUMERIC
DECIMAL

VARCHAR(n) Succeeds if n is 24 characters Succeeds if n is 10 characters Succeeds if n is 10 characters

CHAR(n) or more; resulting string is in or more; resulting string is in or more; resulting string is in
CSTRING(n) format: format: format:
YYYY-MM-DD YYYY-MM-DD HH:MM:SS.thou
HH:MM:SS.thou
BLOB Error Error Error
TABLE 7.2 Casting from datetime datatypes to other SQL datatypes

188 INTERBASE 6
USING DATE LITERALS

CAST datatype at right TIMESTAMP DATE TIME

TO datatype below
TIMESTAMP Always succeeds Succeeds: time portion set Succeeds: date portion set to
to 0:0:0.0000 17 November, 1858
DATE Succeeds: date portion of Always succeeds Error
TIMESTAMP is the result

TIME Succeeds: time portion of Error Always succeeds

TIMESTAMP is the result

TABLE 7.2 Casting from datetime datatypes to other SQL datatypes

Using date literals

InterBase supports five date literals, ’TIMESTAMP’, ’DATE’, ‘TIME’, ’YESTERDAY’, and
’TOMORROW’. Date literals are string values entered between quotation marks that can be
interpreted as date values for EXTRACT, SELECT, INSERT, and UPDATE operations.
’TIMESTAMP’ is a date literal that combines today’s date and time in InterBase format. ’DATE’
is today’s date with time information set to zero. Similarly, ’YESTERDAY’ and ’TOMORROW’
are the expected dates with the time information set to zero.
In EXTRACT and SELECT, ’TIMESTAMP’ and ’DATE’ can be used in the search condition of a
WHERE clause to restrict the data retrieved:
EXEC SQL
SELECT * FROM CROSS_RATE WHERE UPDATE_DATE = ’DATE’;

In INSERT and UPDATE, ’TIMESTAMP’ and ’DATE’ can be used to enter date and time values
instead of relying on isc calls to convert C dates to InterBase dates:
EXEC SQL
INSERT INTO CROSS_RATE VALUES(:from, :to, :rate, ’NOW’);
EXEC SQL
UPDATE CROSS_RATE
SET CONV_RATE = 1.75,
SET UPDATE_DATE = ’TIMESTAMP’
WHERE FROM_CURRENCY = ’POUND’ AND TO_CURRENCT = ’DOLLAR’
AND UPDATE_DATE < ’TODAY’;

EMBEDDED SQL GUIDE 189

 CHAPTER 7 WORKING WITH DATES

190 INTERBASE 6
 CHAPTER

Working with Blob Data

Chapter8
8
This chapter describes the BLOB datatype and its subtypes, how to store Blobs, how to
access them with SQL, DSQL, and API calls, and how to filter Blobs. It also includes
information on writing Blob filters.

What is a Blob?
A Blob is a dynamically sizable datatype that has no specified size and encoding. You can
use a Blob to store large amounts of data of various types, including:
g Bitmapped images
g Vector drawings
g Sounds, video segments, and other multimedia information
g Text and data, including book-length documents
Data stored in the Blob datatype can be manipulated in most of the same ways as data
stored in any other datatype. InterBase stores Blob data inside the database, in contrast
to similar other systems that store pointers to non-database files. For each Blob, there is
a unique identification handle in the appropriate table to point to the database location
of the Blob. By maintaining the Blob data within the database, InterBase improves data
management and access.

EMBEDDED SQL GUIDE 191

 CHAPTER 8 WORKING WITH BLOB DATA

The combination of true database management of Blob data and support for a variety of
datatypes makes InterBase Blob support ideal for transaction-intensive multimedia
applications. For example, InterBase is an excellent platform for interactive kiosk
applications that might provide hundreds or thousands of product descriptions,
photographs, and video clips, in addition to point-of-sale and order processing
capabilities.

How are Blob data stored?

Blob is the InterBase datatype that represents various objects, such as bitmapped images,
sound, video, and text. Before you store these items in the database, you create or
manage them as platform- or product-specific files or data structures, such as:
g TIFF, PICT, BMP, WMF, GEM, TARGA or other bitmapped or vector-graphic files.
g MIDI or WAV sound files.
g Audio Video Interleaved format (.AVI) or QuickTime video files.
g ASCII, MIF, DOC, RTF, WPx or other text files.
g CAD files.
You must load these files from memory into the database programmatically, as you do
any other host-language data items or records you intend to store in InterBase.

192 INTERBASE 6
HOW ARE BLOB DATA STORED?

Blob subtypes
Although you manage Blob data in the same way as other datatypes, InterBase provides
more flexible datatyping rules for Blob data. Because there are many native datatypes that
you can define as Blob data, InterBase treats them somewhat generically and allows you
to define your own datatype, known as a subtype. Also, InterBase provides seven standard
subtypes with which you can characterize Blob data:

Blob
subtype Description
0 Unstructured, generally applied to binary data or data of an indeterminate type
1 Text
2 Binary language representation (BLR)
3 Access control list
4 (Reserved for future use)
5 Encoded description of a table’s current metadata
6 Description of multi-database transaction that finished irregularly
TABLE 8.1 Blob subtypes defined by InterBase

You can specify user-defined subtypes as negative numbers between –1 and

–32,678. Positive integers are reserved for InterBase subtypes.
For example, the following statement defines three Blob columns: Blob1 with subtype 0
(the default), Blob2 with subtype 1 (TEXT), and Blob3 with user-defined subtype –1:
EXEC SQL CREATE TABLE TABLE2
(
BLOB1 BLOB,
BLOB2 BLOB SUB_TYPE 1,
BLOB3 BLOB SUB_TYPE -1
);

To specify both a default segment length and a subtype when creating a Blob column,
use the SEGMENT SIZE option after the SUB_TYPE option. For example:
EXEC SQL CREATE TABLE TABLE2
(
BLOB1 BLOB SUB_TYPE 1 SEGMENT SIZE 100;
);

EMBEDDED SQL GUIDE 193

 CHAPTER 8 WORKING WITH BLOB DATA

The only rule InterBase enforces over these user-defined subtypes is that, when
converting a Blob from one subtype to another, those subtypes must be compatible.
InterBase does not otherwise enforce subtype integrity.

Blob database storage

Because Blob data are typically large, variably-sized objects of binary or text data,
InterBase stores them most efficiently using a method of segmentation. It would be an
inefficient use of disk space to store each Blob as one contiguous mass. Instead, InterBase
stores each Blob in segments that are indexed by a handle that InterBase generates when
you create the Blob. This handle is known as the Blob ID and is a quadword (64-bit)
containing a unique combination of table identifier and Blob identifier.
The Blob ID for each Blob is stored in its appropriate field in the table record. The Blob
ID points to the first segment of the Blob, or to a page of pointers, each of which points
to a segment of one or more Blob fields. You can retrieve the Blob ID by executing a
SELECT statement that specifies the Blob as the target, as in the following example:
EXEC SQL
DECLARE BLOBDESC CURSOR FOR
SELECT GUIDEBOOK
FROM TOURISM
WHERE STATE = ’CA’;

You define Blob columns the same way you define non-Blob columns.
The following SQL code creates a table with a Blob column called PROJ_DESC. It sets the
subtype parameter to 1, which denotes a TEXT Blob, and sets the segment size to 80 bytes:
CREATE TABLE PROJECT
(
PROJ_ID PROJNO NOT NULL,
PROJ_NAME VARCHAR(20) NOT NULL UNIQUE,
PROJ_DESC BLOB SUBTYPE 1 SEGMENT SIZE 80,
TEAM_LEADER EMPNO,
PRODUCT PRODTYPE,
...
);

The following diagram shows the relationship between a Blob column containing a Blob
ID and the Blob data referenced by the Blob ID:

194 INTERBASE 6
HOW ARE BLOB DATA STORED?

FIGURE 8.1 Relationship of a Blob ID to Blob segments in a database

Blob
column
Table row … Blob ID …

Blob data segment segment segment …

Rather than store Blob data directly in the table, InterBase stores a Blob ID in each row
of the table. The Blob ID, a unique number, points to the first segment of the Blob data
that is stored elsewhere in the database, in a series of segments. When an application
creates a Blob, it must write data to that Blob a segment at a time. Similarly, when an
application reads of Blob, it reads a segment at a time. Because most Blob data are large
objects, most Blob management is performed with loops in the application code.

Blob segment length

When you define a Blob in a table, you specify the expected size of Blob segments that
are to be written to the column in the Blob definition statement. The segment length you
define for a Blob column specifies the maximum number of bytes that an application is
expected to write to or read from any Blob in the column. The default segment length is
80. For example, the following column declaration creates a Blob with a segment length
of 120:
EXEC SQL CREATE TABLE TABLE2
(
Blob1 Blob SEGMENT SIZE 120;
);

InterBase uses the segment length setting to determine the size of an internal buffer to
which it writes Blob segment data. Normally, you should not attempt to write segments
larger than the segment length you defined in the table; doing so may result in a buffer
overflow and possible memory corruption.
Specifying a segment size of n guarantees that no more than n number of bytes are read
or written in a single Blob operation. With some types of operations, for instance, with
SELECT, INSERT, and UPDATE operations, you can read or write Blob segments of varying
length.
In the following example of an INSERT CURSOR statement, specify the segment length in
a host language variable, segment_length, as follows:

EMBEDDED SQL GUIDE 195

 CHAPTER 8 WORKING WITH BLOB DATA

EXEC SQL
INSERT CURSOR BCINS VALUES (:write_segment_buffer
INDICATOR :segment_length);

For more information about the syntax of the INSERT CURSOR statement, see the Language
Reference.

Overriding segment length

You can override the segment length setting by including the MAXIMUM_SEGMENT option
in a DECLARE CURSOR statement. For example, the following Blob INSERT cursor
declaration overrides the segment length that was defined for the field, Blob2, increasing
it to 1024:
EXEC SQL
DECLARE BCINS CURSOR FOR INSERT Blob Blob2 INTO TABLE 2
MAXIMUM_SEGMENT 1024;

Note By overriding the segment length setting, you affect only the segment size for the
cursor, not for the column, or for other cursors. Other cursors using the same Blob
column maintain the original segment size that was defined in the column definition, or
can specify their own overrides.
The segment length setting does not affect InterBase system performance. Choose the
segment length most convenient for the specific application. The largest possible segment
length is 65,535 bytes (64K).

Accessing Blob data with SQL

InterBase supports SELECT, INSERT, UPDATE, and DELETE operations on Blob data. The
following sections contain brief discussions of example programs. These programs
illustrate how to perform standard SQL operations on Blob data.

Selecting Blob data

The following example program selects Blob data from the GUIDEBOOK column of the
TOURISM table:
1. Declare host-language variables to store the Blob ID, the Blob segment data,
and the length of segment data:
EXEC SQL

196 INTERBASE 6
ACCESSING BLOB DATA WITH SQL

BEGIN DECLARE SECTION;

BASED ON TOURISM.GUIDEBOOK blob_id;
BASED ON TOURISM.GUIDEBOOK.SEGMENT blob_segment_buf;
BASED ON TOURISM.STATE state;
unsigned short blob_seg_len;
EXEC SQL
END DECLARE SECTION;

The BASED ON … SEGMENT syntax declares a host-language variable,

blob_segment_buf, that is large enough to hold a Blob segment during a FETCH
operation. For more information about the BASED ON statement, see the Language
Reference.
2. Declare a table cursor to select the desired Blob column, in this case the
GUIDEBOOK column:
EXEC SQL
DECLARE TC CURSOR FOR
SELECT STATE, GUIDEBOOK
FROM TOURISM
WHERE STATE = ’CA’;

3. Declare a Blob read cursor. A Blob read cursor is a special cursor used for
reading Blob segments:
EXEC SQL
DECLARE BC CURSOR FOR
READ Blob GUIDEBOOK
FROM TOURISM;

The segment length of the GUIDEBOOK Blob column is defined as 60, so Blob cursor,
BC, reads a maximum of 60 bytes at a time.
To override the segment length specified in the database schema for GUIDEBOOK,
use the MAXIMUM_SEGMENT option. For example, the following code restricts each
Blob read operation to a maximum of 40 bytes, and SQLCODE is set to 101 to indicate
when only a portion of a segment has been read:
EXEC SQL
DECLARE BC CURSOR FOR
READ Blob GUIDEBOOK
FROM TOURISM
MAXIMUM_SEGMENT 40;

No matter what the segment length setting is, only one segment is read at a time.
4. Open the table cursor and fetch a row of data containing a Blob:

EMBEDDED SQL GUIDE 197

 CHAPTER 8 WORKING WITH BLOB DATA

EXEC SQL
OPEN TC;
EXEC SQL
FETCH TC INTO :state, :blob_id;

The FETCH statement fetches the STATE and GUIDEBOOK columns into host variables
state and blob_id, respectively.
5. Open the Blob read cursor using the Blob ID stored in the blob_id variable,
and fetch the first segment of Blob data:
EXEC SQL
OPEN BC USING :blob_id;
EXEC SQL
FETCH BC INTO :blob_segment_buf:blob_seg_len;

When the FETCH operation completes, blob_segment_buf contains the first segment
of the Blob, and blob_seg_len contains the segment’s length, which is the number of
bytes copied into blob_segment_buf.
6. Fetch the remaining segments in a loop. SQLCODE should be checked each
time a fetch is performed. An error code of 100 indicates that all of the Blob
data has been fetched. An error code of 101 indicates that the segment
contains additional data:
while (SQLCODE != 100 || SQLCODE == 101)
{
printf("%*.*s", blob_seg_len, blob_seg_len, blob_segment_buf);
EXEC SQL
FETCH BC INTO :blob_segment_buf:blob_seg_len;
}

InterBase produces an error code of 101 when the length of the segment buffer is less
than the length of a particular segment.
For example, if the length of the segment buffer is 40 and the length of a particular
segment is 60, the first FETCH produces an error code of 101 indicating that data
remains in the segment. The second FETCH reads the remaining 20 bytes of data, and
produces an SQLCODE of 0, indicating that the next segment is ready to be read, or 100
if this was the last segment in the Blob.
1. Close the Blob read cursor:
EXEC SQL
CLOSE BC;

2. Close the table cursor:

198 INTERBASE 6
ACCESSING BLOB DATA WITH SQL

EXEC SQL
CLOSE TC;

Inserting Blob data

The following program inserts Blob data into the GUIDEBOOK column of the TOURISM
table:
1. Declare host-language variables to store the Blob ID, Blob segment data, and
the length of segment data:
EXEC SQL
BEGIN DECLARE SECTION;
BASED ON TOURISM.GUIDEBOOK blob_id;
BASED ON TOURISM.GUIDEBOOK.SEGMENT blob_segment_buf;
BASED ON TOURISM.STATE state;
unsigned short blob_seg_len;
EXEC SQL
END DECLARE SECTION;

· The BASED ON … SEGMENT syntax declares a host-language variable, blob_segment_buf,

that is large enough to hold a Blob segment during a FETCH operation. For more
information about the BASED ON directive, see the Language Reference.
2. Declare a Blob insert cursor:
EXEC SQL
DECLARE BC CURSOR FOR INSERT Blob GUIDEBOOK INTO TOURISM;

3. Open the Blob insert cursor and specify the host variable in which to store
the Blob ID:
EXEC SQL
OPEN BC INTO :blob_id;

4. Store the segment data in the segment buffer, blob_segment_buf, calculate

the length of the segment data, and use an INSERT CURSOR statement to write
the segment:
sprintf(blob_segment_buf, ’We hold these truths to be self
evident’);
blob_segment_len = strlen(blob_segment_buf);

EXEC SQL
INSERT CURSOR BC VALUES (:blob_segment_buf:blob_segment_len);

EMBEDDED SQL GUIDE 199

 CHAPTER 8 WORKING WITH BLOB DATA

Repeat these steps in a loop until you have written all Blob segments.
5. Close the Blob insert cursor:
EXEC SQL
CLOSE BC;

6. Use an INSERT statement to insert a new row containing the Blob into the
TOURISM table:
EXEC SQL
INSERT INTO TOURISM (STATE,GUIDEBOOK) VALUES (’CA’,:blob_id);

7. Commit the changes to the database:

EXEC SQL
COMMIT;

Updating Blob data

You cannot update a Blob directly. You must create a new Blob, read the old Blob data
into a buffer where you can edit or modify it, then write the modified data to the new
Blob.
Create a new Blob by following these steps:
1. Declare a Blob insert cursor:
EXEC SQL
DECLARE BC CURSOR FOR INSERT BLOB GUIDEBOOK INTO TOURISM;

2. Open the Blob insert cursor and specify the host variable in which to store
the Blob ID:
EXEC SQL
OPEN BC INTO :blob_id;

3. Store the old Blob segment data in the segment buffer blob_segment_buf,
calculate the length of the segment data, perform any modifications to the
data, and use an INSERT CURSOR statement to write the segment:
/* Programmatically read the first/next segment of the old Blob
* segment data into blob_segment_buf; */
EXEC SQL
INSERT CURSOR BC VALUES (:blob_segment_buf:blob_segment_len);

Repeat these steps in a loop until you have written all Blob segments.
4. Close the Blob insert cursor:

200 INTERBASE 6
ACCESSING BLOB DATA WITH SQL

EXEC SQL
CLOSE BC;

5. When you have completed creating the new Blob, issue an UPDATE statement
to replace the old Blob in the table with the new one, as in the following
example:
EXEC SQL UPDATE TOURISM
SET
GUIDEBOOK = :blob_id;
WHERE CURRENT OF TC;

Note The TC table cursor points to a target row established by declaring the cursor and
then fetching the row to update.
To modify a text Blob using this technique, you might read an existing Blob field into a
host-language buffer, modify the data, then write the modified buffer over the existing
field data with an UPDATE statement.

Deleting Blob data

There are two methods for deleting a Blob. The first is to delete the row containing the
Blob. The second is to update the row and set the Blob column to NULL or to the Blob ID
of a different Blob (for example, the new Blob created to update the data of an existing
Blob).
The following statement deletes current Blob data in the GUIDEBOOK column of the
TOURISM table by setting it to NULL:
EXEC SQL UPDATE TOURISM
SET
GUIDEBOOK = NULL;
WHERE CURRENT OF TC;

Blob data is not immediately deleted when DELETE is specified. The actual delete
operation occurs when InterBase performs version cleanup. The following code fragment
illustrates how to recover space after deleting a Blob:
EXEC SQL
UPDATE TABLE SET Blob_COLUMN = NULL WHERE ROW = :myrow;
EXEC SQL
COMMIT;
/* wait for all active transactions to finish */
/* force a sweep of the database */

EMBEDDED SQL GUIDE 201

 CHAPTER 8 WORKING WITH BLOB DATA

When InterBase performs garbage collection on old versions of a record, it verifies

whether or not recent versions of the record reference the Blob ID. If the record does not
reference the Blob ID, InterBase cleans up the Blob.

Accessing Blob data with API calls

In addition to accessing Blob data using SQL as described in this chapter, the InterBase
API provides routines for accessing Blob data. The following API calls are provided for
accessing and managing Blob data:

Function Description
isc_blob_default_desc() Loads a Blob descriptor data structure with default information
about a Blob.
isc_blob_gen_bpb() Generates a Blob parameter buffer (BPB) from source and target
Blob descriptors to allow dynamic access to Blob subtype and
character set information.
isc_blob_info() Returns information about an open Blob.
isc_blob_lookup_desc() Looks up and stores into a Blob descriptor the subtype, character
set, and segment size of a Blob.
isc_blob_set_desc() Sets the fields of a Blob descriptor to values specified in parameters
to isc_blob_set_desc().
isc_cancel_blob() Discards a Blob and frees internal storage.
isc_close_blob() Closes an open Blob.
isc_create_blob2() Creates a context for storing a Blob, opens the Blob for write access,
and optionally specifies a filter to be used to translate the Blob data
from one subtype to another.
isc_get_segment() Reads a segment from an open Blob.
isc_open_blob2() Opens an existing Blob for retrieval and optional filtering.
isc_put_segment() Writes a Blob segment.
TABLE 8.2 API Blob calls

For details on using the API calls to access Blob data, see the API Guide.

202 INTERBASE 6
FILTERING BLOB DATA

Filtering Blob data

An understanding of Blob subtypes is particularly important when working with Blob
filters. A Blob filter is a routine that translates Blob data from one subtype to another.
InterBase includes a set of special internal Blob filters that convert from subtype 0 to
subtype 1 (TEXT), and from subtype 1 (TEXT) to subtype 0. In addition to using these
standard filters, you can write your own external filters to provide special data translation.
For example, you might develop a filter to translate bitmapped images from one format
to another.

IMPORTANT Blob filters are available for databases residing on all InterBase server platforms except
NetWare, where Blob filters cannot be created or used.

Using the standard InterBase text filters

The standard InterBase filters convert Blob data of subtype 0, or any InterBase system
type, to subtype 1 (TEXT).
When a text filter is being used to read data from a Blob column, it modifies the standard
InterBase behavior for supplying segments. Regardless of the actual nature of the
segments in the Blob column, the text filter enforces the rule that segments must end with
a newline character (\n).
The text filter returns all the characters up to and including the first newline as the first
segment, the next characters up to and including the second newline as the second
segment, and so on.

Tip To convert any non-text subtype to TEXT, declare its FROM subtype as subtype 0 and its
TO subtype as subtype 1.

Using an external Blob filter

Unlike the standard InterBase filters that convert between subtype 0 and subtype 1, an
external Blob filter is generally part of a library of routines you create and link to your
application.
To use an external filter, you must first write it, compile and link it, then declare it to the
database that contains the Blob data you want processed.

EMBEDDED SQL GUIDE 203

 CHAPTER 8 WORKING WITH BLOB DATA

4 Declaring an external filter to the database

To declare an external filter to a database, use the DECLARE FILTER statement. For example,
the following statement declares the filter, SAMPLE:
EXEC SQL
DECLARE FILTER SAMPLE
INPUT_TYPE -1 OUTPUT_TYPE -2
ENTRY_POINT ’FilterFunction’
MODULE_NAME ’filter.dll’;

In the example, the filter’s input subtype is defined as -1 and its output subtype as -2. In
this example, INPUT_TYPE specifies lowercase text and OUTPUT_TYPE specifies uppercase
text. The purpose of filter, SAMPLE, therefore, is to translate Blob data from lowercase text
to uppercase text.
The ENTRY_POINT and MODULE_NAME parameters specify the external routine that
InterBase calls when the filter is invoked. The MODULE_NAME parameter specifies filter.dll,
the dynamic link library containing the filter’s executable code. The ENTRY_POINT
parameter specifies the entry point into the DLL. The example shows only a simple file
name. It is good practice to specify a fully-qualified path name, since users of your
application need to load the file.

4 Using a filter to read and write Blob data

The following illustration shows the default behavior of the SAMPLE filter that translates
from lowercase text to uppercase text.

FIGURE 8.2 Filtering from lowercase to uppercase

Application Blob
Filter:
abcdef SAMPLE ABCDEF

Similarly, when reading data, the SAMPLE filter can easily read Blob data of subtype -2,
and translate it to data of subtype -1.

204 INTERBASE 6
WRITING AN EXTERNAL BLOB FILTER

FIGURE 8.3 Filtering from uppercase to lowercase

Blob Application
Filter:
ABCDEF SAMPLE abcdef

4 Invoking a filter in an application

To invoke a filter in an application, use the FILTER option when declaring a Blob cursor.
Then, when the application performs operations using the cursor, InterBase
automatically invokes the filter.
For example, the following INSERT cursor definition specifies that the filter, SAMPLE, is to
be used in any operations involving the cursor, BCINS1:
EXEC SQL
DECLARE BCINS1 CURSOR FOR
INSERT Blob Blob1 INTO TABLE1
FILTER FROM -1 TO -2;

When InterBase processes this declaration, it searches a list of filters defined in the
current database for a filter with matching FROM and TO subtypes. If such a filter exists,
InterBase invokes it during Blob operations that use the cursor, BCINS1. If InterBase
cannot locate a filter with matching FROM and TO subtypes, it returns an error to the
application.

Writing an external Blob filter

If you choose to write your own filters, you must have a detailed understanding of the
datatypes you plan to translate. As mentioned elsewhere in this chapter, InterBase does
not do strict datatype checking on Blob data, but does enforce the rule that Blob source
and target subtypes must be compatible. Maintaining and enforcing this compatibility is
your responsibility.

EMBEDDED SQL GUIDE 205

 CHAPTER 8 WORKING WITH BLOB DATA

Filter types
Filters can be divided into two types: filters that convert data one segment at a time, and
filters that convert data many segments at a time.
The first type of filter reads a segment of data, converts it, and supplies it to the
application a segment at a time.
The second type of filter might read all the data and do all the conversion when the Blob
read cursor is first opened, and then simulate supplying data a segment at a time to the
application.
If timing is an issue for your application, you should carefully consider these two types
of filters and which might better serve your purpose.

Read-only and write-only filters

Some filters support only reading from or only writing to a Blob, but not both operations.
If you attempt to use a Blob filter for an operation that it does not support, InterBase
returns an error to the application.

Defining the filter function

When writing your filter, you must include an entry point, known as a filter function, in
the declaration section of the program. InterBase calls the filter function when your
application performs a Blob access operation. All communication between InterBase and
the filter is through the filter function. The filter function itself may call other functions
that comprise the filter executable.

206 INTERBASE 6
WRITING AN EXTERNAL BLOB FILTER

FIGURE 8.4 Filter interaction with an application and a database

APPLICATION

INTERBASE

FILTER

Declare the name of the filter function and the name of the filter executable with the
ENTRY_POINT and MODULE_NAME parameters of the DECLARE FILTER statement.
A filter function must have the following declaration calling sequence:
filter_function_name(short action, isc_blob_ctl control);

The parameter, action, is one of eight possible action macro definitions and the
parameter, control, is an instance of the isc_blob_ctl Blob control structure, defined in
the InterBase header file ibase.h. These parameters are discussed later in this chapter.
The following listing of a skeleton filter declares the filter function, jpeg_filter:
#include <ibase.h>

#define SUCCESS 0
#define FAILURE 1

ISC_STATUS jpeg_filter(short action, isc_blob_ctl control)

{
ISC_STATUS status = SUCCESS;

switch (action)
{
case isc_blob_filter_open:
. . .
break;
case isc_blob_filter_get_segment:
. . .
break;
case isc_blob_filter_create:
. . .

EMBEDDED SQL GUIDE 207

 CHAPTER 8 WORKING WITH BLOB DATA

break;
case isc_blob_filter_put_segment:
. . .
break;
case isc_blob_filter_close:
. . .
break;
case isc_blob_filter_alloc:
. . .
break;
case isc_blob_filter_free:
. . .
break;
case isc_blob_filter_seek:
. . .
break;
default:
status = isc_uns_ext /* unsupported action value */
. . .
break;
}
return status;
}

InterBase passes one of eight possible actions to the filter function, jpeg_filter(), by way
of the action parameter, and also passes an instance of the Blob control structure,
isc_blob_ctl, by way of the parameter control.
The ellipses (…) in the previous listing represent code that performs some operations
based on each action, or event, that is listed in the case statement. Each action is a
particular event invoked by a database operation the application might perform. For
more information, see “Programming filter function actions” on page 211.
The isc_blob_ctl Blob control structure provides the fundamental data exchange between
InterBase and the filter. For more information on the Blob control structure, see
“Defining the Blob control structure” on page 208.

4 Defining the Blob control structure

The Blob control structure, isc_blob_ctl, provides the fundamental method of data
exchange between InterBase and a filter. The declaration for the isc_blob_ctl control
structure is in the InterBase include file, ibase.h.

208 INTERBASE 6
WRITING AN EXTERNAL BLOB FILTER

The isc_blob_ctl structure is used in two ways:

1. When the application performs a Blob access operation, InterBase calls the
filter function and passes it an instance of isc_blob_ctl.
2. Internal filter functions can pass an instance of isc_blob_ctl to internal
InterBase access routines.
In either case, the purpose of certain isc_blob_ctl fields depends on the action being
performed.
For example, when an application attempts a Blob INSERT, InterBase passes an
isc_blob_filter_put_segment action to the filter function. The filter function passes an
instance of the control structure to InterBase. The ctl_buffer of the structure contains the
segment data to be written, as specified by the application in its Blob INSERT statement.
Because the buffer contains information to pass into the filter function, it is called an in
field. The filter function should include instructions in the case statement under the
isc_blob_filter_put_segment case for performing the write to the database.
In a different case, for instance when an application attempts a FETCH operation, the case
of an isc_blob_filter_get_segment action should include instructions for filling ctl_buffer
with segment data from the database to return to the application. In this case, because
the buffer is used for filter function output, it is called an out field.
The following table describes each of the fields in the isc_blob_ctl Blob control structure,
and whether they are used for filter function input (in), or output (out).

Field name Description

(*ctl_source)() Pointer to the internal InterBase Blob access routine. (in)
*ctl_source_handle Pointer to an instance of isc_blob_ctl to be passed to the internal InterBase
Blob access routine. (in)
ctl_to_sub_type Target subtype. Information field. Provided to support multi-purpose filters
that can perform more than one kind of translation. This field and the next one
enable such a filter to decide which translation to perform. (in)
ctl_from_sub_type Source subtype. Information field. Provided to support multi-purpose filters
that can perform more than one kind of translation. This field and the previous
one enable such a filter to decide which translation to perform. (in)
ctl_buffer_length For isc_blob_filter_put_segment, field is an in field that contains the length of
the segment data contained in ctl_buffer.
For isc_blob_filter_get_segment, field is an in field set to the size of the buffer
pointed to by ctl_buffer, which is used to store the retrieved Blob data.
TABLE 8.3 isc_blob_ctl structure field descriptions

EMBEDDED SQL GUIDE 209

 CHAPTER 8 WORKING WITH BLOB DATA

Field name Description

ctl_segment_length Length of the current segment. This field is not used for
isc_blob_filter_put_segment.
For isc_blob_filter_get_segment, the field is an OUT field set to the size of the
retrieved segment (or partial segment, in the case when the buffer length
ctl_buffer_length is less than the actual segment length).
ctl_bpb_length Length of the Blob parameter buffer. Reserved for future enhancement.
*ctl_bpb Pointer to a Blob parameter buffer. Reserved for future enhancement.
*ctl_buffer Pointer to a segment buffer. For isc_blob_filter_put_segment, field is an in
field that contains the segment data.
For isc_blob_filter_get_segment, the field is an OUT field the filter function fills
with segment data for return to the application.
ctl_max_segment Length of longest segment in the Blob. Initial value is 0. The filter function sets
this field. This field is informational only.
ctl_number_segments Total number of segments in the Blob. Initial value is 0. The filter function sets
this field. This field is informational only.
ctl_total_length Total length of the Blob. Initial value is 0. The filter function sets this field. This
field is informational only.
*ctl_status Pointer to the InterBase status vector. (OUT )
ctl_data[8] 8-element array of application-specific data. Use this field to store resource
pointers, such as memory pointers and file handles created by the
isc_blob_filter_open handler, for example. Then, the next time the filter
function is called, the resource pointers will be available for use. ( IN/OUT )
TABLE 8.3 isc_blob_ctl structure field descriptions (continued)

SETTING CONTROL STRUCTURE INFORMATION FIELD VALUES

The isc_blob_ctl structure contains three fields that store information about the Blob
currently being accessed: ctl_max_segment, ctl_number_segments, and ctl_total_length.
You should attempt to maintain correct values for these fields in the filter function,
whenever possible. Depending on the purpose of the filter, maintaining correct values for
the fields is not always possible. For example, a filter that compresses data on a
segment-by-segment basis cannot determine the size of ctl_max_segment until it
processes all segments.

210 INTERBASE 6
WRITING AN EXTERNAL BLOB FILTER

These fields are informational only. InterBase does not use the values of these fields in
internal processing.

4 Programming filter function actions

When an application performs a Blob access operation, InterBase passes a corresponding
action message to the filter function by way of the action parameter. There are eight
possible actions, each of which results from a particular access operation. The following
list of action macro definitions are declared in the ibase.h file:
#define isc_blob_filter_open 0
#define isc_blob_filter_get_segment 1
#define isc_blob_filter_close 2
#define isc_blob_filter_create 3
#define isc_blob_filter_put_segment 4
#define isc_blob_filter_alloc 5
#define isc_blob_filter_free 6
#define isc_blob_filter_seek 7

The following table describes the Blob access operation that corresponds to each action:

Action Invoked when … Use to …

isc_blob_filter_open Application opens a Blob READ Set the information fields of the Blob control
cursor structure.
Perform initialization tasks, such as allocating
memory or opening temporary files.
Set the status variable, if necessary. The value of the
status variable becomes the filter function’s return
value.
isc_blob_filter_get_segment Application executes a Blob Set the ctl_buffer and ctl_segment_length fields of
FETCH statement the Blob control structure to contain a segment’s
worth of translated data on the return of the filter
function.
Perform the data translation if the filter processes
the Blob segment-by-segment.
Set the status variable. The value of the status
variable becomes the filter function’s return value.
isc_blob_filter_close Application closes a Blob cursor Perform exit tasks, such as freeing allocated
memory, closing, or removing temporary files.
TABLE 8.4 Blob access operations

EMBEDDED SQL GUIDE 211

 CHAPTER 8 WORKING WITH BLOB DATA

Action Invoked when … Use to …

isc_blob_filter_create Application opens a Blob INSERT Set the information fields of the Blob control
cursor structure.
Perform initialization tasks, such as allocating
memory or opening temporary files.
Set the status variable, if necessary. The value of the
status variable becomes the filter function’s return
value.
isc_blob_filter_put_segment Application executes a Blob Perform the data translation on the segment data
INSERT statement passed in through the Blob control structure.
Write the segment data to the database. If the
translation process changes the segment length,
the new value must be reflected in the values
passed to the writing function.
Set the status variable. The value of the status
variable becomes the filter function’s return value.
isc_blob_filter_alloc InterBase initializes filter Set the information fields of the Blob control
processing; not a result of a structure.
particular application action Perform initialization tasks, such as allocating
memory or opening temporary files.
Set the status variable, if necessary. The value of the
status variable becomes the filter function’s return
value.
isc_blob_filter_free InterBase ends filter Perform exit tasks, such as freeing allocated
processing; not a result of a memory, closing, or removing temporary files.
particular application action
isc_blob_filter_seek Reserved for internal filter use;
not used by external filters
TABLE 8.4 Blob access operations (continued)

Tip Store resource pointers, such as memory pointers and file handles created by the
isc_blob_filter_open handler, in the ctl_data field of the isc_blob_ctl Blob control
structure. Then, the next time the filter function is called, the resource pointers are still
available.

212 INTERBASE 6
WRITING AN EXTERNAL BLOB FILTER

4 Testing the function return value

The filter function must return an integer indicating the status of the operation it
performed. You can have the function return any InterBase status value returned by an
internal InterBase routine.
In some filter applications, a filter function has to supply status values directly. The
following table lists status values that apply particularly to Blob processing:

Macro constant Value Meaning

SUCCESS 0 Indicates the filter action has been handled successfully. On a
Blob read (isc_blob_filter_get_segment) operation, indicates that
the entire segment has been read.
FAILURE 1 Indicates an unsuccessful operation. In most cases, a status more
specific to the error is returned.
isc_uns_ext See ibase.h Indicates that the attempted action is unsupported by the filter.
For example, a read-only filter would return isc_uns_ext for an
isc_blob_filter_put_segment action.
isc_segment See ibase.h During a Blob read operation, indicates that the supplied buffer is
not large enough to contain the remaining bytes in the current
segment. In this case, only ctl_buffer_length bytes are copied,
and the remainder of the segment must be obtained through
additional isc_blob_filter_get_segment calls.
isc_segstr_eof See ibase.h During a Blob read operation, indicates that the end of the Blob
has been reached; there are no additional segments remaining to
be read.
TABLE 8.5 Blob filter status values

For more information about InterBase status values, see the Language Reference.

EMBEDDED SQL GUIDE 213

 CHAPTER 8 WORKING WITH BLOB DATA

214 INTERBASE 6
 CHAPTER

Chapter9 Using Arrays

9
InterBase supports arrays of most datatypes. Using an array enables multiple data items
to be stored in a single column. InterBase can treat an array as a single unit, or as a series
of separate units, called slices. Using an array is appropriate when:
g The data items naturally form a set of the same datatype
g The entire set of data items in a single database column must be represented and
controlled as a unit, as opposed to storing each item in a separate column
g Each item must also be identified and accessed individually
The data items in an array are called array elements. An array can contain elements of
any InterBase datatype except BLOB. It cannot be an array of arrays, although InterBase
does support multidimensional arrays. All of the elements of an array must be of the same
datatype.

EMBEDDED SQL GUIDE 215

 CHAPTER 9 USING ARRAYS

Creating arrays
Arrays are defined with the CREATE DOMAIN or CREATE TABLE statements. Defining an array
column is just like defining any other column, except that you must also specify the array
dimensions.
Array indexes range from –231 to +231–1.
The following statement defines a regular character column and a single-dimension,
character array column containing four elements:
EXEC SQL
CREATE TABLE TABLE1
(
NAME CHAR(10),
CHAR_ARR CHAR(10)[4]
);

Array dimensions are always enclosed in square brackets following a column’s datatype
specification.
For a complete discussion of CREATE TABLE and array syntax, see the Language Reference.

Multi-dimensional arrays
InterBase supports multi-dimensional arrays, arrays with 1 to 16 dimensions. For
example, the following statement defines three integer array columns with two, three,
and six dimensions, respectively:
EXEC SQL
CREATE TABLE TABLE1
(
INT_ARR2 INTEGER[4,5]
INT_ARR3 INTEGER[4,5,6]
INT_ARR6 INTEGER[4,5,6,7,8,9]
);

In this example, INT_ARR2 allocates storage for 4 rows, 5 elements in width, for a total of
20 integer elements, INT_ARR3 allocates 120 elements, and INT_ARR6 allocates 60,480
elements.

216 INTERBASE 6
CREATING ARRAYS

IMPORTANT InterBase stores multi-dimensional arrays in row-major order. Some host languages,
such as FORTRAN, expect arrays to be in column-major order. In these cases, care must
be taken to translate element ordering correctly between InterBase and the host
language.

Specifying subscript ranges

In InterBase, array dimensions have a specific range of upper and lower boundaries,
called subscripts. In many cases, the subscript range is implicit: the first element of the
array is element 1, the second element 2, and the last is element n. For example, the
following statement creates a table with a column that is an array of four integers:
EXEC SQL
CREATE TABLE TABLE1
(
INT_ARR INTEGER[4]
);

The subscripts for this array are 1, 2, 3, and 4.

A different set of upper and lower boundaries for each array dimension can be explicitly
defined when an array column is created. For example, C programmers, familiar with
arrays that start with a lower subscript boundary of zero, may want to create array
columns with a lower boundary of zero as well.
To specify array subscripts for an array dimension, both the lower and upper boundaries
of the dimension must be specified using the following syntax:
lower:upper

For example, the following statement creates a table with a single-dimension array
column of four elements where the lower boundary is 0 and the upper boundary is 3:
EXEC SQL
CREATE TABLE TABLE1
(
INT_ARR INTEGER[0:3]
);

The subscripts for this array are 0, 1, 2, and 3.

When creating multi-dimensional arrays with explicit array boundaries, separate each
dimension’s set of subscripts from the next with commas. For example, the following
statement creates a table with a two-dimensional array column where each dimension
has four elements with boundaries of 0 and 3:

EMBEDDED SQL GUIDE 217

 CHAPTER 9 USING ARRAYS

EXEC SQL
CREATE TABLE TABLE1
(
INT_ARR INTEGER[0:3, 0:3]
);

Accessing arrays
InterBase can perform operations on an entire array, effectively treating it as a single
element, or it can operate on an array slice, a subset of array elements. An array slice can
consist of a single element, or a set of many contiguous elements.
InterBase supports the following data manipulation operations on arrays:
g Selecting data from an array
g Inserting data into an array
g Updating data in an array slice
g Selecting data from an array slice
g Evaluating an array element in a search condition
A user-defined function (UDF) can only reference a single array element.
The following array operations are not supported:
g Referencing array dimensions dynamically in DSQL
g Inserting data into an array slice
g Setting individual array elements to NULL
g Using the aggregate functions, MIN(), MAX(), SUM(), AVG(), and COUNT() with arrays
g Referencing arrays in the GROUP BY clause of a SELECT
g Creating views that select from array slices

Selecting data from an array

To select data from an array, perform the following steps:
1. Declare a host-language array variable of the correct size to hold the array
data. For example, the following statements create three such variables:
EXEC SQL
BEGIN DECLARE SECTION;

218 INTERBASE 6
ACCESSING ARRAYS

BASED ON TABLE1.CHAR_ARR char_arr;

BASED ON TABLE1.INT_ARR int_arr;
BASED ON TABLE1.FLOAT_ARR float_arr;
EXEC SQL
END DECLARE SECTION;

2. Declare a cursor that specifies the array columns to select. For example,
EXEC SQL
DECLARE TC1 CURSOR FOR
SELECT NAME, CHAR_ARR[], INT_ARR[]
FROM TABLE1;

Be sure to include brackets ([]) after the array column name to select the array data.
If the brackets are left out, InterBase reads the array ID for the column, instead of the
array data.
The ability to read the array ID, which is actually a Blob ID, is included only to
support applications that access array data using InterBase API calls.
3. Open the cursor, and fetch data:
EXEC SQL
OPEN TC1;
EXEC SQL
FETCH TC1 INTO :name, :char_arr, :int_arr;

Note It is not necessary to use a cursor to select array data. For example, a singleton
SELECT might be appropriate, too.
When selecting array data, keep in mind that InterBase stores elements in row-major
order. For example, in a 2-dimensional array, with 2 rows and 3 columns, all 3 elements
in row 1 are returned, then all three elements in row two.

Inserting data into an array

INSERT can be used to insert data into an array column. The data to insert must exactly
fill the entire array, or an error can occur.
To insert data into an array, follow these steps:
1. Declare a host-language variable to hold the array data. Use the BASED ON
clause as a handy way of declaring array variables capable of holding data to
insert into the entire array. For example, the following statements create three
such variables:
EXEC SQL

EMBEDDED SQL GUIDE 219

 CHAPTER 9 USING ARRAYS

BEGIN DECLARE SECTION;

BASED ON TABLE1.CHAR_ARR char_arr;
BASED ON TABLE1.INT_ARR int_arr;
BASED ON TABLE1.FLOAT_ARR float_arr;
EXEC SQL
END DECLARE SECTION;

2. Load the host-language variables with data.

3. Use INSERT to write the arrays. For example,
EXEC SQL
INSERT INTO TABLE1 (NAME, CHAR_ARR, INT_ARR, FLOAT_ARR)
VALUES (’Sample’, :char_arr, :int_arr, :float_arr);

4. Commit the changes:

EXEC SQL
COMMIT;

IMPORTANT When inserting data into an array column, provide data to fill all array elements, or the
results will be unpredictable.

Selecting from an array slice

The SELECT statement supports syntax for retrieving contiguous ranges of elements from
arrays. These ranges are referred to as array slices. Array slices to retrieve are specified
in square brackets ([]) following a column name containing an array. The number inside
the brackets indicates the elements to retrieve. For a one-dimensional array, this is a
single number. For example, the following statement selects the second element in a
one-dimensional array:
EXEC SQL
SELECT JOB_TITLE[2]
INTO :title
FROM EMPLOYEE
WHERE LAST_NAME = :lname;

To retrieve a subset of several contiguous elements from a one-dimensional array, specify

both the first and last elements of the range to retrieve, separating the values with a colon.
The syntax is as follows:
[lower_bound:upper_bound]
For example, the following statement retrieves a subset of three elements from a
one-dimensional array:

220 INTERBASE 6
ACCESSING ARRAYS

EXEC SQL
SELECT JOB_TITLE[2:4]
INTO :title
FROM EMPLOYEE
WHERE LAST_NAME = :lname;

For multi-dimensional arrays, the lower and upper values for each dimension must be
specified, separated from one another by commas, using the following syntax:
[lower:upper, lower:upper [, lower:upper ...]]

Note In this syntax, the bold brackets must be included.

For example, the following statement retrieves two rows of three elements each:
EXEC SQL
DECLARE TC2 CURSOR FOR
SELECT INT_ARR[1:2,1:3]
FROM TABLE1

Because InterBase stores array data in row-major order, the first range of values between
the brackets specifies the subset of rows to retrieve. The second range of values specifies
which elements in each row to retrieve.
To select data from an array slice, perform the following steps:
1. Declare a host-language variable large enough to hold the array slice data
retrieved. For example,
EXEC SQL
BEGIN DECLARE SECTION;
char char_slice[11]; /* 11-byte string for CHAR(10) datatype
*/
long int_slice[2][3];
EXEC SQL
END DECLARE SECTION;

The first variable, char_slice, is intended to store a single element from the CHAR_ARR
column. The second example, int_slice, is intended to store a six-element slice from
the INT_ARR integer column.
2. Declare a cursor that specifies the array slices to read. For example,
EXEC SQL
DECLARE TC2 CURSOR FOR
SELECT CHAR_ARR[1], INT_ARR[1:2,1:3]
FROM TABLE1

3. Open the cursor, and the fetch data:

EMBEDDED SQL GUIDE 221

 CHAPTER 9 USING ARRAYS

EXEC SQL
OPEN TC2;
EXEC SQL
FETCH TC2 INTO :char_slice, :int_slice;

Updating data in an array slice

A subset of elements in an array can be updated with a cursor. To perform an update,
follow these steps:
1. Declare a host-language variable to hold the array slice data. For example,
EXEC SQL
BEGIN DECLARE SECTION;
char char_slice[11]; /* 11-byte string for CHAR(10) datatype
*/
long int_slice[2][3];
EXEC SQL
END DECLARE SECTION;

The first variable, char_slice, is intended to hold a single element of the CHAR_ARR
array column defined in the programming example in the previous section. The
second example, int_slice, is intended to hold a six-element slice of the INT_ARR
integer array column.
2. Select the row that contains the array data to modify. For example, the
following cursor declaration selects data from the INT_ARRAY and CHAR_ARRAY
columns:
EXEC SQL
DECLARE TC1 CURSOR FOR
SELECT CHAR_ARRAY[1], INT_ARRAY[1:2,1:3] FROM TABLE1;
EXEC SQL
OPEN TC1;
EXEC SQL
FETCH TC1 INTO :char_slice, :int_slice;

This example fetches the data currently stored in the specified slices of CHAR_ARRAY
and INT_ARRAY, and stores it into the char_slice and int_slice host-language variables,
respectively.
3. Load the host-language variables with new or updated data.

222 INTERBASE 6
ACCESSING ARRAYS

4. Execute an UPDATE statement to insert data into the array slices. For example,
the following statements put data into parts of CHAR_ARRAY and INT_ARRAY,
assuming char_slice and int_slice contain information to insert into the table:
EXEC SQL
UPDATE TABLE1
SET
CHAR_ARR[1] = :char_slice,
INT_ARR[1:2,1:3] = :int_slice
WHERE CURRENT OF TC1;

5. Commit the changes:

EXEC SQL
COMMIT;

The following fragment of the output from this example illustrates the contents of the
columns, CHAR_ARR and INT_ARR after this operation.

char_arr values:
[0]:string0 [1]:NewString [2]:string2 [3]:string3

int_arr values:
updated values
[0][0]:0 [0][1]:1 [0][2]:2 [0][3]:3
[1][0]:10 [1][1]:999 [1][2]:999 [1][3]:999
[2][0]:20 [2][1]:999 [2][2]:999 [2][3]:999
[3][0]:30 [3][1]:31 [3][2]:32 [3][3]:33

Testing a value in a search condition

A single array element’s value can be evaluated in the search condition of a WHERE
clause. For example,
EXEC SQL
DECLARE TC2 CURSOR FOR
SELECT CHAR_ARR[1], INT_ARR[1:2,1:3]
FROM TABLE1
WHERE SMALLINT_ARR[1,1,1] = 111;

IMPORTANT You cannot evaluate multi-element array slices.

EMBEDDED SQL GUIDE 223

 CHAPTER 9 USING ARRAYS

Using host variables in array subscripts

Integer host variables can be used as array subscripts. For example, the following cursor
declaration uses host variables, getval, and testval, in array subscripts:
EXEC SQL
DECLARE TC2 CURSOR FOR
SELECT CHAR_ARR[1], INT_ARR[:getval:1,1:3]
FROM TABLE1
WHERE FLOAT_ARR[:testval,1,1] = 111.0;

Using arithmetic expressions with arrays

Arithmetic expressions involving arrays can be used only in search conditions. For
example, the following code fetches a row of array data at a time that meets the search
criterion:
for (i = 1; i < 100 && SQLCODE == 0; i++)
{
EXEC SQL
SELECT ARR[:i] INTO :array_var
FROM TABLE1
WHERE ARR1[:j + 1] = 5;
process_array(array_var);
}

224 INTERBASE 6
 CHAPTER

Working with
Chapter10
10
Stored Procedures

A stored procedure is a self-contained set of extended SQL statements stored in a database

as part of its metadata.
Applications can interact with stored procedures in the following ways:
g They can pass parameters to and receive return values from stored procedures.
g They can invoke stored procedures directly to perform a task.
g They can substitute an appropriate stored procedure for a table or view in a SELECT
statement.
The advantages of using stored procedures are:
g Applications can share code. A common piece of SQL code written once and stored in the
database can be used in any application that accesses the database, including the new
InterBase interactive SQL tool, isql.
g Modular design. Stored procedures can be shared among applications, eliminating
duplicate code, and reducing the size of applications.
g Streamlined maintenance. When a procedure is updated, the changes are automatically
reflected in all applications that use it without the need to recompile and relink them.

EMBEDDED SQL GUIDE 225

 CHAPTER 10 WORKING WITH STORED PROCEDURES

g Improved performance, especially for remote client access. Stored procedures are
executed by the server, not the client.
This chapter describes how to call and execute stored procedures in applications once
they are written. For information on how to create a stored procedure, see the Data
Definition Guide.

Using stored procedures

There are two types of procedures that can be called from an application:
g Select procedures that an application can use in place of a table or view in a SELECT
statement. A select procedure must return one or more values, or an error results.
g Executable procedures that an application can call directly, with the EXECUTE PROCEDURE
statement. An executable procedure may or may not return values to the calling program.
Both kinds of procedures are defined with CREATE PROCEDURE and have the same syntax.
The difference is in how the procedure is written and how it is intended to be used. Select
procedures always return zero or more rows, so that to the calling program they appear
as a table or view. Executable procedures are simply routines invoked by the calling
program that can return only a single set of values.
In fact, a single procedure conceivably can be used as a select procedure or an executable
procedure, but this is not recommended. In general a procedure is written specifically to
be used in a SELECT statement (a select procedure) or to be used in an EXECUTE
PROCEDURE statement (an executable procedure). For more information on creating
stored procedures, see the Data Definition Guide.

Procedures and transactions

Procedures operate within the context of a transaction in the program that uses them. If
procedures are used in a transaction, and the transaction is rolled back, then any actions
performed by the procedures are also rolled back. Similarly, a procedure’s actions are not
final until its controlling transaction is committed.

226 INTERBASE 6
USING SELECT PROCEDURES

Security for procedures

When an application calls a stored procedure, the person running the application must
have EXECUTE privilege on the stored procedure. An extension to the GRANT statement
enables assignment of EXECUTE privilege, and an extension to the REVOKE statement
enables removal of the privilege. For more information about granting privileges to users,
see the Data Definition Guide.
In addition, if the stored procedure accesses objects in the database, one of two things
must be true: either the user running the application or the called stored procedure must
have the appropriate permissions on the accessed objects. The GRANT statement assigns
privileges to procedures, and REVOKE eliminates privileges.

Using select procedures

A select procedure is used in place of a table or view in a SELECT statement and can return
zero or more rows. A select procedure must return one or more output parameters, or an
error results. If returned values are not specified, NULL values are returned by default.
The advantages of select procedures over tables or views are:
g They can take input parameters that can affect the output produced.
g They can contain control statements, local variables, and data manipulation statements,
offering great flexibility to the user.
Input parameters are passed to a select procedure in a comma-delimited list in
parentheses following the procedure name.
The following isql script defines the procedure, GET_EMP_PROJ, which returns EMP_PROJ,
the project numbers assigned to an employee, when passed the employee number,
EMP_NO, as the input parameter:
SET TERM !! ;
CREATE PROCEDURE GET_EMP_PROJ (emp_no SMALLINT)
RETURNS (emp_proj SMALLINT) AS
BEGIN
FOR SELECT PROJ_ID
FROM EMPLOYEE_PROJECT
WHERE EMP_NO = :emp_no
INTO :emp_proj
DO

EMBEDDED SQL GUIDE 227

 CHAPTER 10 WORKING WITH STORED PROCEDURES

SUSPEND;
END !!

The following statement retrieves PROJ_ID from the above procedure, passing the host
variable, number, as input:
SELECT PROJ_ID FROM GET_EMP_PROJ (:number);

Calling a select procedure

To use a select procedure in place of a table or view name in an application, use the
procedure name anywhere a table or view name is appropriate. Supply any input
parameters required in a comma-delimited list in parentheses following the procedure
name.
EXEC SQL
SELECT PROJ_ID FROM GET_EMP_PROJ (:emp_no)
ORDER BY PROJ_ID;

IMPORTANT InterBase does not support creating a view by calling a select procedure.

Using a select procedure with cursors

A select procedure can also be used in a cursor declaration. For example, the following
code declares a cursor named PROJECTS, using the GET_EMP_PROJ procedure in place of a
table:
EXEC SQL
DECLARE PROJECTS CURSOR FOR
SELECT PROJ_ID FROM GET_EMP_PROJ (:emp_no)
ORDER BY PROJ_ID;

The following application C code with embedded SQL then uses the PROJECTS cursor to
print project numbers to standard output:
EXEC SQL
OPEN PROJECTS

/* Print employee projects. */

while (SQLCODE == 0)
{
EXEC SQL
FETCH PROJECTS INTO :proj_id :nullind;

228 INTERBASE 6
USING EXECUTABLE PROCEDURES

if (SQLCODE == 100)
break;
if (nullind == 0)
printf("\t%s\n", proj_id);
}

Using executable procedures

An executable procedure is called directly by an application, and often performs a task
common to applications using the same database. Executable procedures can receive
input parameters from the calling program, and can optionally return a single row to the
calling program.
Input parameters pass to an executable procedure in a comma-delimited list following
the procedure name.
Note Executable procedures cannot return multiple rows.

Executing a procedure
To execute a procedure in an application, use the following syntax:
EXEC SQL
EXECUTE PROCEDURE name [:param [[INDICATOR]:indicator]]
[, :param [[INDICATOR]:indicator] ...]
[RETURNING_VALUES :param [[INDICATOR]:indicator]
[, :param [[INDICATOR]:indicator]...]];

When an executable procedure uses input parameters, the parameters can be literal
values (such as 7 or “Fred”), or host variables. If a procedure returns output parameters,
host variables must be supplied in the RETURNING_VALUES clause to hold the values
returned.
For example, the following statement demonstrates how the executable procedure,
DEPT_BUDGET, is called with literal parameters:
EXEC SQL
EXECUTE PROCEDURE DEPT_BUDGET 100 RETURNING_VALUES :sumb;

The following statement also calls the same procedure using a host variable instead of a
literal as the input parameter:
EXEC SQL
EXECUTE PROCEDURE DEPT_BUDGET :rdno RETURNING_VALUES :sumb;

EMBEDDED SQL GUIDE 229

 CHAPTER 10 WORKING WITH STORED PROCEDURES

4 Indicator variables
Both input parameters and return values can have associated indicator variables for
tracking NULL values. You must use indicator variables to indicate unknown or NULL
values of return parameters. The INDICATOR keyword is optional. An indicator variable
that is less than zero indicates that the parameter is unknown or NULL. An indicator
variable that is 0 indicates that the associated parameter contains a non-NULL value. For
more information about indicator variables, see Chapter 6, “Working with Data.”

Executing a procedure in a DSQL application

To execute a stored procedure in a dynamic SQL (DSQL) application follow these steps:
1. Use a PREPARE statement to parse and prepare the procedure call for
execution using the following syntax:
EXEC SQL
PREPARE sql_statement_name FROM :var | ’<statement>’;

2. Set up an input XSQLDA using the following syntax:

EXEC SQL
DESCRIBE INPUT sql_statement_name INTO SQL DESCRIPTOR
input_xsqlda;

3. Use DESCRIBE OUTPUT to set up an output XSQLDA using the following syntax:
EXEC SQL
DESCRIBE OUTPUT sql_statement_name INTO SQL DESCRIPTOR
output_xsqlda;
Setting up an output XSQLDA is only necessary for procedures that return values.
4. Execute the statement using the following syntax:
EXEC SQL
EXECUTE statement USING SQL DESCRIPTOR input_xsqlda
INTO DESCRIPTOR output_xsqlda;

Input parameters to stored procedures can be passed as run-time values by substituting

a question mark (?) for each value. For example, the following DSQL statements prepare
and execute the ADD_EMP_PROJ procedure:
. . .
strcpy(uquery, "EXECUTE PROCEDURE ADD_EMP_PROJ ?, ?");
. . .
EXEC SQL
PREPARE QUERY FROM :uquery;

230 INTERBASE 6
USING EXECUTABLE PROCEDURES

EXEC SQL
DESCRIBE INPUT QUERY INTO SQL DESCRIPTOR input_xsqlda;
EXEC SQL
DESCRIBE OUTPUT QUERY INTO SQL DESCRIPTOR output_xsqlda;
EXEC SQL
EXECUTE QUERY USING SQL DESCRIPTOR input_xsqlda INTO SQL DESCRIPTOR
output_xsqlda;
. . .

EMBEDDED SQL GUIDE 231

 CHAPTER 10 WORKING WITH STORED PROCEDURES

232 INTERBASE 6
 CHAPTER

Working with Events

Chapter11
11
This chapter describes the InterBase event mechanism and how to write applications that
register interest in and respond to events. The event mechanism enables applications to
respond to actions and database changes made by other, concurrently running
applications without the need for those applications to communicate directly with one
another, and without incurring the expense of CPU time required for periodic polling to
determine if an event has occurred.

Understanding the event mechanism

In InterBase, an event is a message passed by a trigger or a stored procedure to the
InterBase event manager to announce the occurrence of a specified condition or action,
usually a database change such as an INSERT, UPDATE, or DELETE. Events are passed by
triggers or stored procedures only when the transaction under which they occur is
committed.
The event manager maintains a list of events posted to it by triggers and stored
procedures. It also maintains a list of applications that have registered an interest in
events. Each time a new event is posted to it, the event manager notifies interested
applications that the event has occurred.
Applications can respond to specific events that might be posted by a trigger or stored
procedure by:

EMBEDDED SQL GUIDE 233

 CHAPTER 11 WORKING WITH EVENTS

1. Indicating an interest in the events to the event manager.

2. Waiting for event notification.
3. Determining which event occurred (if an application is waiting for more than
one event to occur).
The InterBase event mechanism, then, consists of three parts:
g A trigger or stored procedure that posts an event to the event manager.
g The event manager that maintains an event queue and notifies applications when an
event occurs.
g An application that registers interest in the event and waits for it to occur.
A second application that uses the event-posting stored procedure (or that fires the
trigger) causes the event manager to notify the waiting application so that it can resume
processing.

Signaling event occurrences

A trigger or stored procedure must signal the occurrence of an event, usually a database
change such as an INSERT, UPDATE, or DELETE, by using the POST_EVENT statement.
POST_EVENT alerts the event manager to the occurrence of an event after a transaction is
committed. At that time, the event manager passes the information to registered
applications.
A trigger or stored procedure that posts an event is sometimes called an event alerter. For
example, the following isql script creates a trigger that posts an event to the event
manager whenever any application inserts data in a table:
SET TERM !! ;
CREATE TRIGGER POST_NEW_ORDER FOR SALES
ACTIVE
AFTER INSERT
POSITION 0
AS
BEGIN
POST_EVENT ’new_order’;
END
!!
SET TERM ; !!
Event names are restricted to 15 characters in size.

234 INTERBASE 6
REGISTERING INTEREST IN EVENTS

Note POST_EVENT is a stored procedure and trigger language extension, available only
within stored procedures and triggers.
For a complete discussion of writing a trigger or stored procedure as an event alerter, see
the Data Definition Guide.

Registering interest in events

An application must register a request to be notified about a particular event with the
InterBase event manager before waiting for the event to occur. To register interest in an
event, use the EVENT INIT statement. EVENT INIT requires two arguments:
g An application-specific request handle to pass to the event manager.
g A list of events to be notified about, enclosed in parentheses.
The application-specific request handle is used by the application in a subsequent EVENT
WAIT statement to indicate a readiness to receive event notification. The request handle
is used by the event manager to determine where to send notification about particular
events to wake up a sleeping application so that it can respond to them.
The list of event names in parentheses must match event names posted by triggers or
stored procedures, or notification cannot occur.
To register interest in a single event, use the following EVENT INIT syntax:
EXEC SQL
EVENT INIT request_name (event_name);

event_name can be up to 15 characters in size, and can be passed as a constant string in

quotes, or as a host-language variable.
For example, the following application code creates a request named RESPOND_NEW that
registers interest in the “new_order” event:
EXEC SQL
EVENT INIT RESPOND_NEW (’new_order’);

The next example illustrates how RESPOND_NEW might be initialized using a

host-language variable, myevent, to specify the name of an event:
EXEC SQL
EVENT INIT RESPOND_NEW (:myevent);

After an application registers interest in an event, it is not notified about an event until it
first pauses execution with EVENT WAIT. For more information about waiting for events,
see “Waiting for events with EVENT WAIT” on page 236.

EMBEDDED SQL GUIDE 235

 CHAPTER 11 WORKING WITH EVENTS

Note As an alternative to registering interest in an event and waiting for the event to
occur, applications can use an InterBase API call to register interest in an event, and
identify an asynchronous trap (AST) function to receive event notification. This method
enables an application to continue other processing instead of waiting for an event to
occur. For more information about programming events with the InterBase API, see the
API Guide.

Registering interest in multiple events

Often, an application may be interested in several different events even though it can only
wait on a single request handle at a time. EVENT INIT enables an application to specify a
list of event names in parentheses, using the following syntax:
EXEC SQL
EVENT INIT request_name (event_name [event_name ...]);

Each event_name can be up to 15 characters in size, and should correspond to event

names posted by triggers or stored procedures, or notification may never occur. For
example, the following application code creates a request named RESPOND_MANY that
registers interest in three events, “new_order,” “change_order,” and “cancel_order”:
EXEC SQL
EVENT INIT RESPOND_MANY (’new_order’, ’change_order’,
’cancel_order’);

Note An application can also register interest in multiple events by using a separate
EVENT INIT statement with a unique request handle for a single event or groups of events,
but it can only wait on one request handle at a time.

Waiting for events with EVENT WAIT

Even after an application registers interest in an event, it does not receive notification
about that event. Before it can receive notification, it must use the EVENT WAIT statement
to indicate its readiness to the event manager, and to suspend its processing until
notification occurs.
To signal the event manager and suspend an application’s processing, use the following
EVENT WAIT statement syntax:
EXEC SQL
EVENT WAIT request_name;

236 INTERBASE 6
RESPONDING TO EVENTS

request_name must be the name of a request handle declared in a previous EVENT INIT
statement.
The following statements register interest in an event, and wait for event notification:
EXEC SQL
EVENT INIT RESPOND_NEW (’new_order’);
EXEC SQL
EVENT WAIT RESPOND_NEW;

Once EVENT WAIT is executed, application processing stops until the event manager sends
a notification message to the application.
Note An application can contain more than one EVENT WAIT statement, but all processing
stops when the first statement is encountered. Each time processing restarts, it stops
when it encounters the next EVENT WAIT statement.
If one event occurs while an application is processing another, the event manager sends
notification the next time the application returns to a wait state.

Responding to events
When event notification occurs, a suspended application resumes normal processing at
the next statement following EVENT WAIT.
If an application has registered interest in more than one event with a single EVENT INIT
call, then the application must determine which event occurred by examining the event
array, isc_event[]. The event array is automatically created for an application during
preprocessing. Each element in the array corresponds to an event name passed as an
argument to EVENT INIT. The value of each element is the number of times that event
occurred since execution of the last EVENT WAIT statement with the same request handle.
In the following code, an application registers interest in three events, then suspends
operation pending event notification:
EXEC SQL
EVENT INIT RESPOND_MANY (’new_order’, ’change_order’,
’cancel_order’);
EXEC SQL
EVENT WAIT RESPOND_MANY;

When any of the “new_order,” “change_order,” or “cancel_order” events are posted and
their controlling transactions commit, the event manager notifies the application and
processing resumes. The following code illustrates how an application might test which
event occurred:

EMBEDDED SQL GUIDE 237

 CHAPTER 11 WORKING WITH EVENTS

for (i = 0; i < 3; i++)

{
if (isc_$event[i] > 0)
{
/* this event occurred, so process it */
. . .
}
}

238 INTERBASE 6
 CHAPTER

Error Handling and

Chapter12
12
Recovery

All SQL applications should include mechanisms for trapping, responding to, and
recovering from run-time errors, the errors that can occur when someone uses an
application. This chapter describes both standard, portable SQL methods for handling
errors, and additional error handling specific to InterBase.

EMBEDDED SQL GUIDE 239

 CHAPTER 12 ERROR HANDLING AND RECOVERY

Standard error handling

Every time an SQL statement is executed, it returns a status indicator in the SQLCODE
variable, which is declared automatically for SQL programs during preprocessing with
gpre. The following table summarizes possible SQLCODE values and their meanings:

Value Meaning
0 Success
1–99 Warning or informational message
100 End of file (no more data)
<0 Error. Statement failed to complete
TABLE 12.1 Possible SQLCODE values

To trap and respond to run-time errors, SQLCODE should be checked after each SQL
operation. There are three ways to examine SQLCODE and respond to errors:
g Use WHENEVER statements to automate checking SQLCODE and handle errors when they
occur.
g Test SQLCODE directly after individual SQL statements.
g Use a judicious combination of WHENEVER statements and direct testing.
Each method has advantages and disadvantages, described fully in the remainder of this
chapter.

WHENEVER statements
The WHENEVER statement enables all SQL errors to be handled with a minimum of coding.
WHENEVER statements specify error-handling code that a program should execute when
SQLCODE indicates errors, warnings, or end-of-file. The syntax of WHENEVER is:
EXEC SQL
WHENEVER {SQLERROR | SQLWARNING | NOT FOUND}
{GOTO label | CONTINUE};

After WHENEVER appears in a program, all subsequent SQL statements automatically jump
to the specified code location identified by label when the appropriate error or warning
occurs.

240 INTERBASE 6
STANDARD ERROR HANDLING

Because they affect all subsequent statements, WHENEVER statements are usually
embedded near the start of a program. For example, the first statement in the following
C code’s main() function is a WHENEVER that traps SQL errors:
main()
{
EXEC SQL
WHENEVER SQLERROR GOTO ErrorExit;
. . .
Error Exit:
if (SQLCODE)
{
print_error();
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT;
exit(1);
}
}
. . .
print_error()
{
printf("Database error, SQLCODE = %d\n", SQLCODE);
}

Up to three WHENEVER statements can be active at any time:

g WHENEVER SQLERROR is activated when SQLCODE is less than zero, indicating that a
statement failed.
g WHENEVER SQLWARNING is activated when SQLCODE contains a value from 1 to 99,
inclusive, indicating that while a statement executed, there is some question about the
way it succeeded.
g WHENEVER NOT FOUND is activated when SQLCODE is 100, indicating that end-of-file was
reached during a FETCH or SELECT.
Omitting a statement for a particular condition means it is not trapped, even if it occurs.
For example, if WHENEVER NOT FOUND is left out of a program, then when a FETCH or
SELECT encounters the end-of-file, SQLCODE is set to 100, but program execution
continues as if no error condition has occurred.
Error conditions also can be overlooked by using the CONTINUE statement inside a
WHENEVER statement:

EMBEDDED SQL GUIDE 241

 CHAPTER 12 ERROR HANDLING AND RECOVERY

. . .
EXEC SQL
WHENEVER SQLWARNING
CONTINUE;
. . .
This code traps SQLCODE warning values, but ignores them. Ordinarily, warnings should
be investigated, not ignored.

IMPORTANT Use WHENEVER SQLERROR CONTINUE at the start of error-handling routines to disable
error handling temporarily. Otherwise, there is a possibility of an infinite loop; should
another error occur in the handler itself, the routine will call itself again.

SCOPE OF WHENEVER STATEMENTS

WHENEVER only affects all subsequent SQL statements in the module, or source code file,
where it is defined. In programs with multiple source code files, each module must define
its own WHENEVER statements.

CHANGING ERROR-HANDLING ROUTINES

To switch to another error-handling routine for a particular error condition, embed
another WHENEVER statement in the program at the point where error handling should be
changed. The new assignment overrides any previous assignment, and remains in effect
until overridden itself. For example, the following program fragment sets an initial jump
point for SQLERROR conditions to ErrorExit1, then resets it to ErrorExit2:
EXEC SQL
WHENEVER SQLERROR
GOTO ErrorExit1;
. . .
EXEC SQL
WHENEVER SQLERROR
GOTO ErrorExit2;
. . .

LIMITATIONS OF WHENEVER STATEMENTS

There are two limitations to WHENEVER. It:
g Traps errors indiscriminately. For example, WHENEVER SQLERROR traps both missing
databases and data entry that violates a CHECK constraint, and jumps to a single
error-handling routine. While a missing database is a severe error that may require action
outside the context of the current program, invalid data entry may be the result of a
typing mistake that could be fixed by reentering the data.

242 INTERBASE 6
STANDARD ERROR HANDLING

g Does not easily enable a program to resume processing at the point where the error
occurred. For example, a single WHENEVER SQLERROR can trap data entry that violates a
CHECK constraint at several points in a program, but jumps to a single error-handling
routine. It might be helpful to allow the user to reenter data in these cases, but the error
routine cannot determine where to jump to resume program processing.
Error-handling routines can be very sophisticated. For example, in C or C++, a routine
might use a large CASE statement to examine SQLCODE directly and respond differently to
different values. Even so, creating a sophisticated routine that can resume processing at
the point where an error occurred is difficult. To resume processing after error recovery,
consider testing SQLCODE directly after each SQL statement, or consider using a
combination of error-handling methods.

Testing SQLCODE directly

A program can test SQLCODE directly after each SQL statement instead of relying on
WHENEVER to trap and handle all errors. The main advantage to testing SQLCODE directly
is that custom error-handling routines can be designed for particular situations.
For example, the following C code fragment checks if SQLCODE is not zero after a SELECT
statement completes. If so, an error has occurred, so the statements inside the if clause
are executed. These statements test SQLCODE for two specific values,
–1, and 100, handling each differently. If SQLCODE is set to any other error value, a generic
error message is displayed and the program is ended gracefully.
EXEC SQL
SELECT CITY INTO :city FROM STATES
WHERE STATE = :stat:statind;

if (SQLCODE)
{
if (SQLCODE == –1)
printf("too many records found\n");
else if (SQLCODE == 100)
printf("no records found\n");
else
{
printf("Database error, SQLCODE = %d\n", SQLCODE);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT;

EMBEDDED SQL GUIDE 243

 CHAPTER 12 ERROR HANDLING AND RECOVERY

exit(1);
}
}
printf("found city named %s\n", city);
EXEC SQL
COMMIT;
EXEC SQL
DISCONNECT;

The disadvantage to checking SQLCODE directly is that it requires many lines of extra code
just to see if an error occurred. On the other hand, it enables errors to be handled with
function calls, as the following C code illustrates:
EXEC SQL
SELECT CITY INTO :city FROM STATES
WHERE STATE = :stat:statind;

switch (SQLCODE)
{
case 0:
break; /* NO ERROR */
case –1
ErrorTooMany();
break;
case 100:
ErrorNotFound();
break;
default:
ErrorExit(); /* Handle all other errors */
break;
}
. . .

Using function calls for error handling enables programs to resume execution if errors
can be corrected.

Combining error-handling techniques

Error handling in many programs can benefit from combining WHENEVER with direct
checking of SQLCODE. A program might include generic WHENEVER statements for
handling most SQL error conditions, but for critical operations, WHENEVER statements
might be temporarily overridden to enable direct checking of SQLCODE.

244 INTERBASE 6
STANDARD ERROR HANDLING

For example, the following C code:

g Sets up generic error handling with three WHENEVER statements.
g Overrides the WHENEVER SQLERROR statement to force program continuation using the
CONTINUE clause.
g Checks SQLCODE directly.
g Overrides WHENEVER SQLERROR again to reset it.
main()
{
EXEC SQL
WHENEVER SQLERROR GOTO ErrorExit; /* trap all errors */
EXEC SQL
WHENEVER SQLWARNING GOTO WarningExit; /* trap warnings */
EXEC SQL
WHENEVER NOT FOUND GOTO AllDone; /* trap end of file */
. . .
EXEC SQL
WHENEVER SQLERROR CONTINUE; /* prevent trapping of errors */
EXEC SQL
SELECT CITY INTO :city FROM STATES
WHERE STATE = :stat:statind;
switch (SQLCODE)
{
case 0:
break; /* NO ERROR */
case –1
ErrorTooMany();
break;
case 100:
ErrorNotFound();
break;
default:
ErrorExitFunction(); /* Handle all other errors */
break;
}
EXEC SQL
WHENEVER SQLERROR GOTO ErrorExit; /* reset to trap all errors
*/
. . .
}

EMBEDDED SQL GUIDE 245

 CHAPTER 12 ERROR HANDLING AND RECOVERY

Guidelines for error handling

The following guidelines apply to all error-handling routines in a program.

USING SQL AND HOST-LANGUAGE STATEMENTS

All SQL statements and InterBase functions can be used in error-handling routines, except
for CONNECT.
Any host-language statements and functions can appear in an error-handling routine
without restriction.

IMPORTANT Use WHENEVER SQLERROR CONTINUE at the start of error-handling routines to disable
error-handling temporarily. Otherwise, there is a possibility of an infinite loop; should
another error occur in the handler itself, the routine will call itself again.

NESTING ERROR-HANDLING ROUTINES

Although error-handling routines can be nested or called recursively, this practice is not
recommended unless the program preserves the original contents of SQLCODE and the
InterBase error status array.

HANDLING UNEXPECTED AND UNRECOVERABLE ERRORS

Even if an error-handling routine catches and handles recoverable errors, it should always
contain statements to handle unexpected or unrecoverable errors.
The following code handles unrecoverable errors:
. . .
isc_print_sqlerr(SQLCODE, isc_status);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT;
exit(1);

PORTABILITY
For portability among different SQL implementations, SQL programs should limit error
handling to WHENEVER statements or direct examination of SQLCODE values.
InterBase internal error recognition occurs at a finer level of granularity than SQLCODE
representation permits. A single SQLCODE value can represent many different internal
InterBase errors. Where portability is not an issue, it may be desirable to perform
additional InterBase error handling. The remainder of this chapter explains how to use
these additional features.

246 INTERBASE 6
ADDITIONAL INTERBASE ERROR HANDLING

Additional InterBase error handling

The same SQLCODE value can be returned by multiple InterBase errors. For example, the
SQLCODE value, –901, is generated in response to many different InterBase errors. When
portability to other vendors’ SQL implementations is not required, SQL programs can
sometimes examine the InterBase error status array, isc_status, for more specific error
information.
isc_status is an array of twenty elements of type ISC_STATUS. It is declared automatically
for programs when they are preprocessed with gpre. Two kinds of InterBase error
information are reported in the status array:
g InterBase error message components.
g InterBase error numbers.
As long as the current SQLCODE value is not –1, 0, or 100, error-handling routines that
examine the error status array can do any of the following:
g Display SQL and InterBase error messages.
g Capture SQL and InterBase error messages to a storage buffer for further manipulation.
g Trap for and respond to particular InterBase error codes.
The InterBase error status array is usually examined only after trapping errors with
WHENEVER or testing SQLCODE directly.

Displaying error messages

If SQLCODE is less than –1, additional InterBase error information can be displayed using
the InterBase isc_print_sqlerror() function inside an error-handling routine. During
preprocessing with gpre, this function is automatically declared for InterBase applications.
isc_print_sqlerror() displays the SQLCODE value, a related SQL error message, and any
InterBase error messages in the status array. It requires two parameters: SQLCODE, and a
pointer to the error status array, isc_status.
For example, when an error occurs, the following code displays the value of SQLCODE,
displays a corresponding SQL error message, then displays additional InterBase error
message information if possible:
. . .
EXEC SQL
SELECT CITY INTO :city FROM STATES
WHERE STATE = :stat:statind;
if(SQLCODE)

EMBEDDED SQL GUIDE 247

 CHAPTER 12 ERROR HANDLING AND RECOVERY

{
isc_print_sqlerror(SQLCODE, isc_status);
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT ALL;
exit(1);
}
. . .

IMPORTANT Some windowing systems do not encourage or permit direct screen writes. Do not use
isc_print_sqlerror() when developing applications for these environments. Instead, use
isc_sql_interprete() and isc_interprete() to capture messages to a buffer for display.

Capturing SQL error messages

Instead of displaying SQL error messages, an application can capture the text of those
messages in a buffer by using isc_sql_interprete(). Capture messages in a buffer when
applications:
g Run under windowing systems that do not permit direct writing to the screen.
g Store a record of all error messages in a log file.
g Manipulate or format error messages for display.
Given SQLCODE, a pointer to a storage buffer, and the maximum size of the buffer in bytes,
isc_sql_interprete() builds an SQL error message string, and puts the formatted string in
the buffer where it can be manipulated. A buffer size of 256 bytes is large enough to hold
any SQL error message.
For example, the following code stores an SQL error message in err_buf, then writes
err_buf to a log file:
. . .
char err_buf[256]; /* error message buffer for isc_sql_interprete() */
FILE *efile=NULL; /* code fragment assumes pointer to an open file */
. . .
EXEC SQL
SELECT CITY INTO :city FROM STATES
WHERE STATE = :stat:statind;
if (SQLCODE)
{
isc_sql_interprete(SQLCODE, err_buf, sizeof(err_buf));

248 INTERBASE 6
ADDITIONAL INTERBASE ERROR HANDLING

if(efile==NULL) efile=fopen("errors", "w");

fprintf(efile, "%s\n", err_buf); /* write buffer to log file */
EXEC SQL
ROLLBACK; /* undo database changes */
EXEC SQL
DISCONNECT ALL; /* close open databases */
exit(1); /* exit with error flag set */
}
. . .

isc_sql_interprete() retrieves and formats a single message corresponding to a given

SQLCODE. When SQLCODE is less than –1, more specific InterBase error information is
available. It, too, can be retrieved, formatted, and stored in a buffer by using the
isc_interprete() function.

Capturing InterBase error messages

The text of InterBase error messages can be captured in a buffer by using isc_interprete().
Capture messages in a buffer when applications:
g Run under windowing systems that do not permit direct writing to the screen.
g Store a record of all error messages in a log file.
g Manipulate or format error messages for display.

IMPORTANT isc_interprete() should not be used unless SQLCODE is less than –1 because the contents
of isc_status may not contain reliable error information in these cases.
Given both the location of a storage buffer previously allocated by the program, and a
pointer to the start of the status array, isc_interprete() builds an error message string from
the information in the status array, and puts the formatted string in the buffer where it
can be manipulated. It also advances the status array pointer to the start of the next
cluster of available error information.
isc_interprete() retrieves and formats a single error message each time it is called. When
an error occurs in an InterBase program, however, the status array may contain more
than one error message. To retrieve all relevant error messages, error-handling routines
should repeatedly call isc_interprete() until it returns no more messages.
Because isc_interprete() modifies the pointer to the status array that it receives, do not
pass isc_status directly to it. Instead, declare a pointer to isc_status, then pass the pointer
to isc_interprete().

EMBEDDED SQL GUIDE 249

 CHAPTER 12 ERROR HANDLING AND RECOVERY

The following C code fragment illustrates how InterBase error messages can be captured
to a log file, and demonstrates the proper declaration of a string buffer and pointer to
isc_status. It assumes the log file is properly declared and opened before control is passed
to the error-handling routine. It also demonstrates how to set the pointer to the start of
the status array in the error-handling routine before isc_interprete() is first called.
. . .
#include "ibase.h";
. . .
main()
{
char msg[512];
ISC_STATUS *vector;
FILE *efile; /* code fragment assumes pointer to an open file */
. . .
if (SQLCODE < –1)
ErrorExit();
}
. . .

ErrorExit()
{
vector = isc_status; /* (re)set to start of status vector */
isc_interprete(msg, &vector); /* retrieve first mesage */
fprintf(efile, "%s\n", msg); /* write buffer to log file */
msg[0] = '-'; /* append leading hyphen to secondary messages */
while (isc_interprete(msg + 1, &vector)) /* more?*/
fprintf(efile, "%s\n", msg); /* if so, write it to log */
fclose(efile); /* close log prior to quitting program */
EXEC SQL
ROLLBACK;
EXEC SQL
DISCONNECT ALL;
exit(1); /* quit program with an 'abnormal termination' code */
}
. . .

In this example, the error-handling routine performs the following tasks:

g Sets the error array pointer to the starting address of the status vector, isc_status.
g Calls isc_interprete() a single time to retrieve the first error message from the status
vector.

250 INTERBASE 6
ADDITIONAL INTERBASE ERROR HANDLING

g Writes the first message to a log file.

g Makes repeated calls to isc_interprete() within a WHILE loop to retrieve any additional
messages. If additional messages are retrieved, they are also written to the log file.
g Rolls back the transaction.
g Closes the database and releases system resources.

Handling InterBase error codes

Whenever SQLCODE is less than –1, the error status array, isc_status, may contain detailed
error information specific to InterBase, including error codes, numbers that uniquely
identify each error. With care, error-handling routines can trap for and respond to specific
codes.
To trap and handle InterBase error codes in an error-handling routine, follow these steps:
1. Check SQLCODE to be sure it is less than –1.
2. Check that the first element of the status array is set to isc_arg_gds,
indicating that an InterBase error code is available. In C programs, the first
element of the status array is isc_status[0].
Do not attempt to handle errors reported in the status array if the first status array
element contains a value other than 1.
3. If SQLCODE is less than –1 and the first element in isc_status is set to
isc_arg_gds, use the actual InterBase error code in the second element of
isc_status to branch to an appropriate routine for that error.

Tip InterBase error codes are mapped to mnemonic definitions (for example, isc_arg_gds)
that can be used in code to make it easier to read, understand, and maintain. Definitions
for all InterBase error codes can be found in the ibase.h file.
The following C code fragment illustrates an error-handling routine that:
g Displays error messages with isc_print_sqlerror().
g Illustrates how to parse for and handle six specific InterBase errors which might be
corrected upon roll back, data entry, and retry.
g Uses mnemonic definitions for InterBase error numbers.
. . .
int c, jval, retry_flag = 0;
jmp_buf jumper;
. . .

EMBEDDED SQL GUIDE 251

 CHAPTER 12 ERROR HANDLING AND RECOVERY

main()
{
. . .
jval = setjmp(jumper);
if (retry_flag)
ROLLBACK;
. . .
}
int ErrorHandler(void)
{
retry_flag = 0; /* reset to 0, no retry */
isc_print_sqlerror(SQLCODE, isc_status); /* display errors */
if (SQLCODE < –1)
{
if (isc_status[0] == isc_arg_gds)
{
switch (isc_status[1])
{
case isc_convert_error:
case isc_deadlock:
case isc_integ_fail:
case isc_lock_conflict:
case isc_no_dup:
case isc_not_valid:
printf("\n Do you want to try again? (Y/N)");
c = getchar();
if (c == 'Y' || c == 'y')
{
retry_flag = 1; /* set flag to retry */
longjmp(jumper, 1);
}
break;
case isc_end_arg: /* there really isn’t an error */
retry_flag = 1; /* set flag to retry */
longjump(jumper, 1);
break;
default: /* we can’t handle everything, so abort */
break;
}
}
}
EXEC SQL

252 INTERBASE 6
ADDITIONAL INTERBASE ERROR HANDLING

ROLLBACK;
EXEC SQL
DISCONNECT ALL;
exit(1);
}

EMBEDDED SQL GUIDE 253

 CHAPTER 12 ERROR HANDLING AND RECOVERY

254 INTERBASE 6
 CHAPTER

Using Dynamic SQL

Chapter13
13
This chapter describes how to write dynamic SQL applications, applications that elicit or
build SQL statements for execution at run time.
In many database applications, the programmer specifies exactly which SQL statements
to execute against a particular database. When the application is compiled, these
statements become fixed. In some database applications, it is useful to build and execute
statements from text string fragments or from strings elicited from the user at run time.
These applications require the capability to create and execute SQL statements
dynamically at run time. Dynamic SQL (DSQL) provides this capability. For example, the
InterBase isql utility is a DSQL application.

Overview of the DSQL programming process

Building and executing DSQL statements involves the following general steps:
g Embedding SQL statements that support DSQL processing in an application.
g Using host-language facilities, such as datatypes and macros, to provide input and output
areas for passing statements and parameters at run time.
g Programming methods that use these statements and facilities to process SQL statements
at run time.
These steps are described in detail throughout this chapter.

EMBEDDED SQL GUIDE 255

 CHAPTER 13 USING DYNAMIC SQL

DSQL limitations
Although DSQL offers many advantages, it also has the following limitations:
g Access to one database at a time.
g Dynamic transaction processing is not permitted; all named transactions must be
declared at compile time.
g Dynamic access to Blob and array data is not supported; Blob and array data can be
accessed, but only through standard, statically processed SQL statements, or through
low-level API calls.
g Database creation is restricted to CREATE DATABASE statements executed within the context
of EXECUTE IMMEDIATE.
For more information about handling transactions in DSQL applications, see “Handling
transactions” on page 257. For more information about working with Blob data in
DSQL, see “Processing Blob data” on page 259. For more information about handling
array data in DSQL, see “Processing array data” on page 259. For more information
about dynamic creation of databases, see “Creating a database” on page 258.

Accessing databases
Using standard SQL syntax, a DSQL application can only use one database handle per
source file module, and can, therefore, only be connected to a single database at a time.
Database handles must be declared and initialized when an application is preprocessed
with gpre. For example, the following code creates a single handle, db1, and initializes it
to zero:
#include "ibase.h"
isc_db_handle db1;
. . .
db1 = 0L;

After a database handle is declared and initialized, it can be assigned dynamically to a

database at run time as follows:
char dbname[129];
. . .
prompt_user("Name of database to open: ");
gets(dbname);
EXEC SQL
SET DATABASE db1 = :dbname;
EXEC SQL

256 INTERBASE 6
DSQL LIMITATIONS

CONNECT db1;
. . .

The database accessed by DSQL statements is always the last database handle mentioned
in a SET DATABASE command. A database handle can be used to connect to different
databases as long as a previously connected database is first disconnected with
DISCONNECT. DISCONNECT automatically sets database handles to NULL. The following
statements disconnect from a database, zero the database handle, and connect to a new
database:
EXEC SQL
DISCONNECT db1;
EXEC SQL
SET DATABASE db1 = ’employee.gdb’;
EXEC SQL
CONNECT db1;

To access more than one database using DSQL, create a separate source file module for
each database, and use low-level API calls to attach to the databases and access data. For
more information about accessing databases with API calls, see the API Guide. For more
information about SQL database statements, see Chapter 3, “Working with Databases.”

Handling transactions
InterBase requires that all transaction names be declared when an application is
preprocessed with gpre. Once fixed at precompile time, transaction handles cannot be
changed at run time, nor can new handles be declared dynamically at run time.
SQL statements such as PREPARE, DESCRIBE, EXECUTE, and EXECUTE IMMEDIATE, can be
coded at precompile time to include an optional TRANSACTION clause specifying which
transaction controls statement execution. The following code declares, initializes, and
uses a transaction handle in a statement that processes a run-time DSQL statement:
#include "ibase.h"
isc_tr_handle t1;
. . .
t1 = 0L;
EXEC SQL
SET TRANSACTION NAME t1;
EXEC SQL
PREPARE TRANSACTION t1 Q FROM :sql_buf;

EMBEDDED SQL GUIDE 257

 CHAPTER 13 USING DYNAMIC SQL

DSQL statements that are processed with PREPARE, DESCRIBE, EXECUTE, and EXECUTE
IMMEDIATE cannot use a TRANSACTION clause, even if it is permitted in standard,
embedded SQL.
The SET TRANSACTION statement cannot be prepared, but it can be processed with
EXECUTE IMMEDIATE if:
1. Previous transactions are first committed or rolled back.
2. The transaction handle is set to NULL.
For example, the following statements commit the previous default transaction, then start
a new one with EXECUTE IMMEDIATE:
EXEC SQL
COMMIT;
/* set default transaction name to NULL */
gds__trans = NULL;
EXEC SQL
EXECUTE IMMEDIATE ’SET TRANSACTION READ ONLY’;

Creating a database
To create a new database in a DSQL application:
1. Disconnect from any currently attached databases. Disconnecting from a
database automatically sets its database handle to NULL.
2. Build the CREATE DATABASE statement to process.
3. Execute the statement with EXECUTE IMMEDIATE.
For example, the following statements disconnect from any currently connected
databases, and create a new database. Any existing database handles are set to NULL, so
that they can be used to connect to the new database in future DSQL statements.
char *str = "CREATE DATABASE \"new_emp.gdb\"";
. . .
EXEC SQL
DISCONNECT ALL;
EXEC SQL
EXECUTE IMMEDIATE :str;

258 INTERBASE 6
WRITING A DSQL APPLICATION

Processing Blob data

DSQL does not directly support Blob processing. Blob cursors are not supported in DSQL.
DSQL applications can use API calls to process Blob data. For more information about
Blob API calls, see the API Guide.

Processing array data

DSQL does not directly support array processing. DSQL applications can use API calls to
process array data. For more information about array API calls, see the API Guide.

Writing a DSQL application

Write a DSQL application when any of the following are not known until run time:
g The text of the SQL statement
g The number of host variables
g The datatypes of host variables
g References to database objects
Writing a DSQL application is usually more complex than programming with regular SQL
because for most DSQL operations, the application needs explicitly to allocate and
process an extended SQL descriptor area (XSQLDA) data structure to pass data to and from
the database.
To use DSQL to process an SQL statement, follow these basic steps:
1. Determine if DSQL can process the SQL statement.
2. Represent the SQL statement as a character string in the application.
3. If necessary, allocate one or more XSQLDAs for input parameters and return
values.
4. Use an appropriate DSQL programming method to process the SQL
statement.

EMBEDDED SQL GUIDE 259

 CHAPTER 13 USING DYNAMIC SQL

SQL statements that DSQL can process

DSQL can process most but not all SQL statements. The following table lists SQL
statement that are available to DSQL:

ALTER DATABASE DECLARE FILTER SET GENERATOR

ALTER DOMAIN DELETE UPDATE
ALTER EXCEPTION DROP DATABASE
ALTER INDEX DROP DOMAIN
ALTER PROCEDURE DROP EXCEPTION
ALTER TABLE DROP EXTERNAL FUNCTION
ALTER TRIGGER DROP FILTER
COMMIT DROP INDEX
CONNECT DROP PROCEDURE
CREATE DATABASE DROP ROLE
CREATE DOMAIN DROP SHADOW
CREATE EXCEPTION DROP TABLE
CREATE GENERATOR DROP TRIGGER
CREATE INDEX DROP VIEW
CREATE PROCEDURE EXECUTE PROCEDURE
CREATE ROLE GRANT
CREATE SHADOW INSERT
CREATE TABLE INSERT CURSOR (BLOB)
CREATE TRIGGER REVOKE
CREATE VIEW ROLLBACK
DECLARE EXTERNAL FUNCTION SELECT

The following ESQL statements cannot be processed by DSQL: CLOSE, DECLARE, CURSOR,
DESCRIBE, EXECUTE, EXECUTE IMMEDIATE, FETCH, OPEN, PREPARE.
The following ISQL commands cannot be processed by DSQL: BLOBDUMP, EDIT, EXIT, HELP,
INPUT, OUTPUT, QUIT, SET, SET AUTODDL, SET BLOBDISPLAY, SET COUNT, SET ECHO, SET LIST,
SET NAMES, SET PLAN, SET STATS, SET TERM, SET TIME, SHELL, SHOW CHECK, SHOW DATABASE,
SHOW DOMAINS, SHOW EXCEPTIONS, SHOW FILTERS, SHOW FUNCTIONS, SHOW GENERATORS,
SHOW GRANT, SHOW INDEX, SHOW PROCEDURES, SHOW SYSTEM, SHOW TABLES, SHOW
TRIGGERS, SHOW VERSION, SHOW VIEWS.

260 INTERBASE 6
WRITING A DSQL APPLICATION

SQL character strings

Within a DSQL application, an SQL statement can come from different sources. It can
come directly from a user who enters a statement at a prompt, as does isql. Or it can be
generated by the application in response to user interaction. Whatever the source of the
SQL statement it must be represented as an SQL statement string, a character string that
is passed to DSQL for processing.
Because SQL statement strings are C character strings that are processed directly by
DSQL, they cannot begin with the EXEC SQL prefix or end with a semicolon (;). The
semicolon is, of course, the appropriate terminator for the C string declaration itself. For
example, the following host-language variable declaration is a valid SQL statement string:
char *str = "DELETE FROM CUSTOMER WHERE CUST_NO = 256";

Value parameters in statement strings

SQL statement strings often include value parameters, expressions that evaluate to a
single numeric or character value. Parameters can be used anywhere in statement strings
where SQL expects a value that is not the name of a database object.
A value parameter in a statement string can be passed as a constant, or passed as a
placeholder at run time. For example, the following statement string passes 256 as a
constant:
char *str = "DELETE FROM CUSTOMER WHERE CUST_NO = 256";

It is also possible to build strings at run time from a combination of constants. This
method is useful for statements where the variable is not a true constant, or it is a table
or column name, and where the statement is executed only once in the application.
To pass a parameter as a placeholder, the value is passed as a question mark (?)
embedded within the statement string:
char *str = "DELETE FROM CUSTOMER WHERE CUST_NO = ?";

When DSQL processes a statement containing a placeholder, it replaces the question

mark with a value supplied in the XSQLDA. Use placeholders in statements that are
prepared once, but executed many times with different parameter values.
Replaceable value parameters are often used to supply values in WHERE clause
comparisons and in the UPDATE statement SET clause.

EMBEDDED SQL GUIDE 261

 CHAPTER 13 USING DYNAMIC SQL

Understanding the XSQLDA

All DSQL applications must declare one or more extended SQL descriptor areas
(XSQLDAs). The XSQLDA structure definition can be found in the ibase.h header file in the
InterBase include directory. Applications declare instances of the XSQLDA for use.
The XSQLDA is a host-language data structure that DSQL uses to transport data to or from
a database when processing an SQL statement string. There are two types of XSQLDAs:
input descriptors and output descriptors. Both input and output descriptors are
implemented using the XSQLDA structure.
One field in the XSQLDA, the XSQLVAR, is especially important, because one XSQLVAR must
be defined for each input parameter or column returned. Like the XSQLDA, the XSQLVAR is
a structure defined in ibase.h in the InterBase include directory.
Applications do not declare instances of the XSQLVAR ahead of time, but must, instead,
dynamically allocate storage for the proper number of XSQLVAR structures required for
each DSQL statement before it is executed, then deallocate it, as appropriate, after
statement execution.
The following figure illustrates the relationship between the XSQLDA and the XSQLVAR:

262 INTERBASE 6
UNDERSTANDING THE XSQLDA

FIGURE 13.1 XSQLDA and XSQLVAR relationship

Single instance of XSQLDA

short version
char sqldaid[8]
ISC_LONG sqldabc
short sqln
short sqld
XSQLVAR sqlvar[n]

Array of n instances of XSQLVAR

1st instance nth instance

short sqltype short sqltype
short sqlscale short sqlscale
short sqlsubtype short sqlsubtype
short sqllen short sqllen
char *sqldata char *sqldata
short *sqlind short *sqlind
short sqlname_length short sqlname_length
char sqlname[32] char sqlname[32]
short relname_length short relname_length
char relname[32] char relname[32]
short ownname_length short ownname_length
char ownname[32] char ownname[32]
short aliasname_length short aliasname_length
char aliasname[32] char aliasname[32]

An input XSQLDA consists of a single XSQLDA structure, and one XSQLVAR structure for
each input parameter. An output XSQLDA also consists of one XSQLDA structure and one
XSQLVAR structure for each data item returned by the statement. An XSQLDA and its
associated XSQLVAR structures are allocated as a single block of contiguous memory.

EMBEDDED SQL GUIDE 263

 CHAPTER 13 USING DYNAMIC SQL

The PREPARE and DESCRIBE statements can be used to determine the proper number of
XSQLVAR structures to allocate, and the XSQLDA_LENGTH macro can be used to allocate the
proper amount of space. For more information about the XSQLDA_LENGTH macro, see
“Using the XSQLDA_LENGTH macro” on page 267.

XSQLDA field descriptions

The following table describes the fields that comprise the XSQLDA structure:

Field definition Description

short version Indicates the version of the XSQLDA structure. Set by an application. The current
version is defined in ibase.h as SQLDA_VERSION1.
char sqldaid[8] Reserved for future use.
ISC_LONG Reserved for future use.
sqldabc
short sqln Indicates the number of elements in the sqlvar array. Set by the application.
Whenever the application allocates storage for a descriptor, it should set this field.
short sqld Indicates the number of parameters (for an input XSQLDA), or the number of
select-list items (for an output XSQLDA). Set by InterBase during a DESCRIBE or
PREPARE.
For an input descriptor, an sqld of 0 indicates that the SQL statement has no
parameters. For an output descriptor, an sqld of 0 indicates that the SQL statement is
not a SELECT statement.
XSQLVAR sqlvar The array of XSQLVAR structures. The number of elements in the array is specified in
the sqln field.
TABLE 13.1 XSQLDA field descriptions

264 INTERBASE 6
UNDERSTANDING THE XSQLDA

XSQLVAR field descriptions

The following table describes the fields that comprise the XSQLVAR structure:

Field definition Description

short sqltype Indicates the SQL datatype of parameters or select-list items. Set by InterBase
during PREPARE or DESCRIBE.
short sqlscale Provides scale, specified as a negative number, for exact numeric datatypes
(DECIMAL, NUMERIC). Set by InterBase during PREPARE or DESCRIBE.
short sqlsubtype Specifies the subtype for Blob data. Set by InterBase during PREPARE or
DESCRIBE.

short sqllen Indicates the maximum size, in bytes, of data in the sqldata field. Set by
InterBase during PREPARE or DESCRIBE.
char *sqldata For input descriptors, specifies either the address of a select-list item or a
parameter. Set by the application.
For output descriptors, contains a value for a select-list item. Set by InterBase.
short *sqlind On input, specifies the address of an indicator variable. Set by an application.
On output, specifies the address of column indicator value for a select-list
item following a FETCH. A value of 0 indicates that the column is not NULL; a
value of –1 indicates the column is NULL. Set by InterBase.
short sqlname_length Specifies the length, in bytes, of the data in field, sqlname. Set by InterBase
during DESCRIBE OUTPUT.
char sqlname[32] Contains the name of the column. Not null (\0) terminated. Set by InterBase
during DESCRIBE OUTPUT.
short relname_length Specifies the length, in bytes, of the data in field, relname. Set by InterBase
during DESCRIBE OUTPUT.
TABLE 13.2 XSQLVAR field descriptions

EMBEDDED SQL GUIDE 265

 CHAPTER 13 USING DYNAMIC SQL

Field definition Description

char relname[32] Contains the name of the table. Not null (\0) terminated. Set by InterBase
during DESCRIBE OUTPUT.
short Specifies the length, in bytes, of the data in field, ownname. Set by InterBase
ownname_length during DESCRIBE OUTPUT.
char ownname[32] Contains the owner name of the table. Not null (\0) terminated. Set by
InterBase during DESCRIBE OUTPUT.
short Specifies the length, in bytes, of the data in field, aliasname. Set by InterBase
aliasname_length during DESCRIBE OUTPUT.
char aliasname[32] Contains the alias name of the column. If no alias exists, contains the column
name. Not null (\0) terminated. Set by InterBase during DESCRIBE OUTPUT.
TABLE 13.2 XSQLVAR field descriptions (continued)

Input descriptors
Input descriptors process SQL statement strings that contain parameters. Before an
application can execute a statement with parameters, it must supply values for them. The
application indicates the number of parameters passed in the XSQLDA sqld field, then
describes each parameter in a separate XSQLVAR structure. For example, the following
statement string contains two parameters, so an application must set sqld to 2, and
describe each parameter:
char *str = "UPDATE DEPARTMENT SET BUDGET = ? WHERE LOCATION = ?";

When the statement is executed, the first XSQLVAR supplies information about the BUDGET
value, and the second XSQLVAR supplies the LOCATION value.
For more information about using input descriptors, see “DSQL programming
methods” on page 273.

Output descriptors
Output descriptors return values from an executed query to an application. The sqld field
of the XSQLDA indicates how many values were returned. Each value is stored in a
separate XSQLVAR structure. The XSQLDA sqlvar field points to the first of these XSQLVAR
structures. The following statement string requires an output descriptor:
char *str = "SELECT * FROM CUSTOMER WHERE CUST_NO > 100";

266 INTERBASE 6
UNDERSTANDING THE XSQLDA

For information about retrieving information from an output descriptor, see “DSQL
programming methods” on page 273.

Using the XSQLDA_LENGTH macro

The ibase.h header file defines a macro, XSQLDA_LENGTH, to calculate the number of bytes
that must be allocated for an input or output XSQLDA. XSQLDA_LENGTH is defined as
follows:
#define XSQLDA_LENGTH (n) (sizeof (XSQLDA) + (n - 1) * sizeof(XSQLVAR))

n is the number of parameters in a statement string, or the number of select-list items

returned from a query. For example, the following C statement uses the XSQLDA_LENGTH
macro to specify how much memory to allocate for an XSQLDA with 5 parameters or
return items:
XSQLDA *my_xsqlda;
. . .
my_xsqlda = (XSQLDA *) malloc(XSQLDA_LENGTH(5));
. . .

For more information about using the XSQLDA_LENGTH macro, see “DSQL programming
methods” on page 273.

EMBEDDED SQL GUIDE 267

 CHAPTER 13 USING DYNAMIC SQL

SQL datatype macro constants

InterBase defines a set of macro constants to represent SQL datatypes and NULL status
information in an XSQLVAR. An application should use these macro constants to specify
the datatype of parameters and to determine the datatypes of select-list items in an SQL
statement. The following table lists each SQL datatype, its corresponding macro constant
expression, C datatype or InterBase typedef, and whether or not the sqlind field is used
to indicate a parameter or variable that contains NULL or unknown data:

sqlind
SQL datatype Macro expression C datatype or typedef used?
Array SQL_ARRAY ISC_QUAD No
Array SQL_ARRAY + 1 ISC_QUAD Yes
Blob SQL_BLOB ISC_QUAD No
Blob SQL_BLOB + 1 ISC_QUAD Yes
CHAR SQL_TEXT char[ ] No
CHAR SQL_TEXT + 1 char[ ] Yes
DATE SQL_DATE ISC_DATE No
DATE SQL_DATE + 1 ISC_DATE Yes
DECIMAL SQL_SHORT, SQL_LONG, int, long, double, or ISC_INT64 No
SQL_DOUBLE, or SQL_INT64

DECIMAL SQL_SHORT+ 1, SQL_LONG + 1, int, long, double, or ISC_INT64 Yes

SQL_DOUBLE+ 1, or SQL_INT64 + 1

DOUBLE PRECISON SQL_DOUBLE double No

DOUBLE PRECISION SQL_DOUBLE + 1 double Yes
INTEGER SQL_LONG long No
INTEGER SQL_LONG + 1 long Yes
FLOAT SQL_FLOAT float No
FLOAT SQL_FLOAT + 1 float Yes
TABLE 13.3 SQL datatypes, macro expressions, and C datatypes

268 INTERBASE 6
UNDERSTANDING THE XSQLDA

sqlind
SQL datatype Macro expression C datatype or typedef used?
NUMERIC SQL_SHORT, SQL_LONG, int, long, double, or ISC_INT64 No
SQL_DOUBLE, or SQL_INT64

NUMERIC SQL_SHORT + 1, SQL_LONG + 1, int, long, double, or ISC_INT64 Yes

SQL_DOUBLE + 1, or SQL_INT64 + 1

SMALLINT SQL_SHORT short No

SMALLINT SQL_SHORT + 1 short Yes
TIME SQL_TIME ISC_TIME No
TIME SQL_TIME + 1 ISC_TIME Yes
TIMESTAMP SQL_TIMESTAMP ISC_TIMESTAMP No
TIMESTAMP SQL_TIMESTAMP +1 ISC_TIMESTAMP Yes
VARCHAR SQL_VARYING First 2 bytes: short containing No
the length of the character string.
Remaining bytes: char[ ]
VARCHAR SQL_VARYING + 1 First 2 bytes: short containing Yes
the length of the character string.
Remaining bytes: char[ ]
TABLE 13.3 SQL datatypes, macro expressions, and C datatypes (continued)

Note DECIMAL and NUMERIC datatypes are stored internally as SMALLINT, INTEGER, DOUBLE
PRECISION, or 64-bit integer datatypes. To specify the correct macro expression to provide
for a DECIMAL or NUMERIC column, use isql to examine the column definition in the table
to see how InterBase is storing column data, then choose a corresponding macro
expression.
The datatype information for a parameter or select-list item is contained in the sqltype
field of the XSQLVAR structure. The value contained in the sqltype field provides two pieces
of information:
g The datatype of the parameter or select-list item.
g Whether sqlind is used to indicate NULL values. If sqlind is used, its value specifies
whether the parameter or select-list item is NULL (–1), or not NULL (0).

EMBEDDED SQL GUIDE 269

 CHAPTER 13 USING DYNAMIC SQL

For example, if the sqltype field equals SQL_TEXT, the parameter or select-list item is a
CHAR that does not use sqlind to check for a NULL value (because, in theory, NULL values
are not allowed for it). If sqltype equals SQL_TEXT + 1, then sqlind can be checked to see
if the parameter or select-list item is NULL.

Tip The C language expression, sqltype & 1, provides a useful test of whether a
parameter or select-list item can contain a NULL. The expression evaluates to 0 if the
parameter or select-list item cannot contain a NULL, and 1 if the parameter or select-list
item can contain a NULL. The following code fragment demonstrates how to use the
expression:
if (sqltype & 1 == 0)
{
/* parameter or select-list item that CANNOT contain a NULL */
}
else
{
/* parameter or select-list item CAN contain a NULL */
}

By default, both PREPARE INTO and DESCRIBE return a macro expression of type + 1, so
the sqlind should always be examined for NULL values with these statements.

Handling varying string datatypes

VARCHAR, CHARACTER VARYING, and NCHAR VARYING datatypes require careful handling in
DSQL. The first two bytes of these datatypes contain string length information, while the
remainder of the data contains the actual bytes of string data to process.
To avoid having to write code to extract and process variable-length strings in an
application, it is possible to force these datatypes to fixed length using SQL macro
expressions. For more information about forcing variable-length data to fixed length for
processing, see “Coercing datatypes” on page 271.
Applications can, instead, detect and process variable-length data directly. To do so, they
must extract the first two bytes from the string to determine the byte-length of the string
itself, then read the string, byte-by-byte, into a null-terminated buffer.

270 INTERBASE 6
UNDERSTANDING THE XSQLDA

NUMERIC and DECIMAL datatypes

DECIMAL and NUMERIC datatypes are stored internally as SMALLINT, INTEGER, DOUBLE
PRECISION, or 64-bit integer datatypes, depending on the precision and scale defined for
a column definition that uses these types. To determine how a DECIMAL or NUMERIC value
is actually stored in the database, use isql to examine the column definition in the table.
If NUMERIC is reported, then data is actually being stored as a 64-bit integer.
When a DECIMAL or NUMERIC value is stored as a SMALLINT or INTEGER, the value is stored
as a whole number. During retrieval in DSQL, the sqlscale field of the XSQLVAR is set to a
negative number that indicates the factor of ten by which the whole number (returned in
sqldata), must be divided in order to produce the correct NUMERIC or DECIMAL value with
its fractional part. If sqlcale is –1, then the number must be divided by 10, if it is –2, then
the number must be divided by 100, –3 by 1,000, and so forth.

Coercing datatypes
Sometimes when processing DSQL input parameters and select-list items, it is desirable
or necessary to translate one datatype to another. This process is referred to as datatype
coercion. For example, datatype coercion is often used when parameters or select-list
items are of type VARCHAR. The first two bytes of VARCHAR data contain string length
information, while the remainder of the data is the string to process. By coercing the data
from SQL_VARYING to SQL_TEXT, data processing can be simplified.
Coercion can only be from one compatible datatype to another. For example,
SQL_VARYING to SQL_TEXT, or SQL_SHORT to SQL_LONG.

4 Coercing character datatypes

To coerce SQL_VARYING datatypes to SQL_TEXT datatypes, change the sqltype field in the
parameter’s or select-list item’s XSQLVAR structure to the desired SQL macro datatype
constant. For example, the following statement assumes that var is a pointer to an
XSQLVAR structure, and that it contains an SQL_VARYING datatype to convert to SQL_TEXT:
var->sqltype = SQL_TEXT;

After coercing a character datatype, provide proper storage space for it. The XSQLVAR field,
sqllen, contains information about the size of the uncoerced data. Set the XSQLVAR sqldata
field to the address of the data.

EMBEDDED SQL GUIDE 271

 CHAPTER 13 USING DYNAMIC SQL

4 Coercing numeric datatypes

To coerce one numeric datatype to another, change the sqltype field in the parameter’s or
select-list item’s XSQLVAR structure to the desired SQL macro datatype constant. For
example, the following statement assumes that var is a pointer to an XSQLVAR structure,
and that it contains an SQL_SHORT datatype to convert to SQL_LONG:
var->sqltype = SQL_LONG;

IMPORTANT Do not coerce a larger datatype to a smaller one. Data can be lost in such a translation.

4 Setting a NULL indicator

If a parameter or select-list item can contain a NULL value, the sqlind field is used to
indicate its NULL status. Appropriate storage space must be allocated for sqlind before
values can be stored there.
On insertion, set sqlind to –1 to indicate that NULL values are legal. Otherwise set sqlind
to 0.
On selection, an sqlind of –1 indicates a field contains a NULL value. Other values indicate
a field contains non-NULL data.

Aligning numerical data

Ordinarily, when a variable with a numeric datatype is created, the compiler will ensure
that the variable is stored at a properly aligned address, but when numeric data is stored
in a dynamically allocated buffer space, such as can be pointed to by the XSQLDA and
XSQLVAR structures, the programmer must take precautions to ensure that the storage
space is properly aligned.
Certain platforms, in particular those with RISC processors, require that numeric data in
dynamically allocated storage structures be aligned properly in memory. Alignment is
dependent both on datatype and platform.
For example, a short integer on a Sun SPARCstation must be located at an address divisible
by 2, while a long on the same platform must be located at an address divisible by 4. In
most cases, a data item is properly aligned if the address of its starting byte is divisible by
the correct alignment number. Consult specific system and compiler documentation for
alignment requirements.
A useful rule of thumb is that the size of a datatype is always a valid alignment number
for the datatype. For a given type T, if size of (T) equals n, then addresses divisible by n
are correctly aligned for T. The following macro expression can be used to align data:
#define ALIGN(ptr, n) ((ptr + n - 1) & ~(n - 1))

272 INTERBASE 6
DSQL PROGRAMMING METHODS

where ptr is a pointer to char.

The following code illustrates how the ALIGN macro might be used:
char *buffer_pointer, *next_aligned;
next_aligned = ALIGN(buffer_pointer, sizeof(T));

DSQL programming methods

There are four possible DSQL programming methods for handling an SQL statement
string. The best method for processing a string depends on the type of SQL statement in
the string, and whether or not it contains placeholders for parameters. The following
decision table explains how to determine the appropriate processing method for a given
string.

Is it a query? Does it have placeholders? Processing method to use

No No Method 1
No Yes Method 2
Yes No Method 3
Yes Yes Method 4
TABLE 13.4 SQL statement strings and recommended processing methods

Method 1: Non-query statements without parameters

There are two ways to process an SQL statement string containing a non-query statement
without placeholder parameters:
g Use EXECUTE IMMEDIATE to prepare and execute the string a single time.
g Use PREPARE to parse the statement for execution and assign it a name, then use EXECUTE
to carry out the statement’s actions as many times as required in an application.

4 Using EXECUTE IMMEDIATE

1. To execute a statement string a single time, use EXECUTE IMMEDIATE:
2. Elicit a statement string from the user or create one that contains the SQL
statement to be processed. For example, the following statement creates an
SQL statement string:
char *str = "UPDATE DEPARTMENT SET BUDGET = BUDGET * 1.05";

EMBEDDED SQL GUIDE 273

 CHAPTER 13 USING DYNAMIC SQL

3. Parse and execute the statement string using EXECUTE IMMEDIATE:

EXEC SQL
EXECUTE IMMEDIATE :str;

Note EXECUTE IMMEDIATE also accepts string literals. For example,

EXEC SQL
EXECUTE IMMEDIATE
’UPDATE DEPARTMENT SET BUDGET = BUDGET * 1.05’;

4 Using PREPARE and EXECUTE

To execute a statement string several times, use PREPARE and EXECUTE:
1. Elicit a statement string from the user or create one that contains the SQL
statement to be processed. For example, the following statement creates an
SQL statement string:
char *str = "UPDATE DEPARTMENT SET BUDGET = BUDGET * 1.05";

2. Parse and name the statement string with PREPARE. The name is used in
subsequent calls to EXECUTE:
EXEC SQL
PREPARE SQL_STMT FROM :str;

SQL_STMT is the name assigned to the parsed statement string.

3. Execute the named statement string using EXECUTE. For example, the
following statement executes a statement string named SQL_STMT:
EXEC SQL
EXECUTE SQL_STMT;

Note PREPARE also accepts string literals. For example,

EXEC SQL
PREPARE SQL_STMT FROM
’UPDATE DEPARTMENT SET BUDGET = BUDGET * 1.05’;

Once a statement string is prepared, it can be executed as many times as required in

an application.

274 INTERBASE 6
DSQL PROGRAMMING METHODS

Method 2: Non-query statements with parameters

There are two steps to processing an SQL statement string containing a non-query
statement with placeholder parameters:
1. Creating an input XSQLDA to process a statement string’s parameters.
2. Preparing and executing the statement string with its parameters.

4 Creating the input XSQLDA

Placeholder parameters are replaced with actual data before a prepared SQL statement
string is executed. Because those parameters are unknown when the statement string is
created, an input XSQLDA must be created to supply parameter values at execute time. To
prepare the XSQLDA, follow these steps:
1. Declare a variable to hold the XSQLDA needed to process parameters. For
example, the following declaration creates an XSQLDA called in_sqlda:
XSQLDA *in_sqlda;
2. Optionally declare a variable for accessing the XSQLVAR structure of the
XSQLDA:
XSQLVAR *var;

Declaring a pointer to the XSQLVAR structure is not necessary, but can simplify
referencing the structure in subsequent statements.
3. Allocate memory for the XSQLDA using the XSQLDA_LENGTH macro. The
following statement allocates storage for in_sqlda:
in_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(10));
In this statement space for 10 XSQLVAR structures is allocated, allowing the XSQLDA
to accommodate up to 10 parameters.
4. Set the version field of the XSQLDA to SQLDA_VERSION1, and set the sqln field
to indicate the number of XSQLVAR structures allocated:
in_sqlda_version = SQLDA_VERSION1;
in_sqlda->sqln = 10;

4 Preparing and executing a statement string with parameters

After an XSQLDA is created for holding a statement string’s parameters, the statement
string can be created and prepared. Local variables corresponding to the placeholder
parameters in the string must be assigned to their corresponding sqldata fields in the
XSQLVAR structures.
To prepare and execute a non-query statement string with parameters, follow these steps:

EMBEDDED SQL GUIDE 275

 CHAPTER 13 USING DYNAMIC SQL

1. Elicit a statement string from the user or create one that contains the SQL
statement to be processed. For example, the following statement creates an
SQL statement string with placeholder parameters:
char *str = "UPDATE DEPARTMENT SET BUDGET = ?, LOCATION = ?";
This statement string contains two parameters: a value to be assigned to the BUDGET
field and a value to be assigned to the LOCATION field.
2. Parse and name the statement string with PREPARE. The name is used in
subsequent calls to DESCRIBE and EXECUTE:
EXEC SQL
PREPARE SQL_STMT FROM :str;

SQL_STMT is the name assigned to the prepared statement string.

3. Use DESCRIBE INPUT to fill the input XSQLDA with information about the
parameters contained in the SQL statement:
EXEC SQL
DESCRIBE INPUT SQL_STMT USING SQL DESCRIPTOR in_sqlda;
4. Compare the value of the sqln field of the XSQLDA to the value of the sqld field
to make sure enough XSQLVARs are allocated to hold information about each
parameter. sqln should be at least as large as sqln. If not, free the storage
previously allocated to the input descriptor, reallocate storage to reflect the
number of parameters specified by sqld, reset sqln and version, then execute
DESCRIBE INPUT again:
if (in_sqlda->sqld > in_sqlda->sqln)
{
n = in_sqlda->sqld;
free(in_sqlda);
in_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(n));
in_sqlda->sqln = n;
in_sqlda->version = SQLDA_VERSION1;
EXEC SQL
DESCRIBE INPUT SQL_STMT USING SQL DESCRIPTOR in_sqlda;
}

5. Process each XSQLVAR parameter structure in the XSQLDA. Processing a

parameter structure involves up to four steps:
- Coercing a parameter’s datatype (optional).
- Allocating local storage for the data pointed to by the sqldata field of the XSQLVAR.
This step is only required if space for local variables is not allocated until run time.
The following example illustrates dynamic allocation of local variable storage space.

276 INTERBASE 6
DSQL PROGRAMMING METHODS

- Providing a value for the parameter consistent with its datatype (required).
- Providing a NULL value indicator for the parameter.
The following code example illustrates these steps, looping through each XSQLVAR
structure in the in_sqlda XSQLDA:
for (i=0, var = in_sqlda->sqlvar; i < in_sqlda->sqld; i++, var++)
{
/* Process each XSQLVAR parameter structure here.
The parameter structure is pointed to by var.*/
dtype = (var->sqltype & ~1) /* drop NULL flag for now */
switch(dtype)
{
case SQL_VARYING: /* coerce to SQL_TEXT */
var->sqltype = SQL_TEXT;
/* Allocate local variable storage. */
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen);
. . .
break;
case SQL_TEXT:
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen);
/* Provide a value for the parameter. */
. . .
break;
case SQL_LONG:
var->sqldata = (char *)malloc(sizeof(long));
/* Provide a value for the parameter. */
*(long *)(var->sqldata) = 17;
break;
. . .
} /* End of switch statement. */
if (sqltype & 1)
{
/* Allocate variable to hold NULL status. */
var->sqlind = (short *)malloc(sizeof(short));
}
} /* End of for loop. */

For more information about datatype coercion and NULL indicators, see “Coercing
datatypes” on page 271.

EMBEDDED SQL GUIDE 277

 CHAPTER 13 USING DYNAMIC SQL

6. Execute the named statement string with EXECUTE. Reference the parameters
in the input XSQLDA with the USING SQL DESCRIPTOR clause. For example, the
following statement executes a statement string named SQL_STMT:
EXEC SQL
EXECUTE SQL_STMT USING SQL DESCRIPTOR in_sqlda;

4 Re-executing the statement string

Once a non-query statement string with parameters is prepared, it can be executed as
often as required in an application. Before each subsequent execution, the input XSQLDA
can be supplied with new parameter and NULL indicator data.
To supply new parameter and NULL indicator data for a prepared statement, repeat steps
3–5 of “Preparing and Executing a Statement String with Parameters,” in this chapter.

Method 3: Query statements without parameters

There are three steps to processing an SQL query statement string without parameters:
1. Preparing an output XSQLDA to process the select-list items returned when the
query is executed.
2. Preparing the statement string.
3. Using a cursor to execute the statement and retrieve select-list items from the
output XSQLDA.

4 Preparing the output XSQLDA

Most queries return one or more rows of data, referred to as a select-list. Because the
number and kind of items returned are unknown when a statement string is created, an
output XSQLDA must be created to store select-list items that are returned at run time. To
prepare the XSQLDA, follow these steps:
1. Declare a variable to hold the XSQLDA needed to store the column data for
each row that will be fetched. For example, the following declaration creates
an XSQLDA called out_sqlda:
XSQLDA *out_sqlda;

2. Optionally declare a variable for accessing the XSQLVAR structure of the

XSQLDA:
XSQLVAR *var;

278 INTERBASE 6
DSQL PROGRAMMING METHODS

Declaring a pointer to the XSQLVAR structure is not necessary, but can simplify
referencing the structure in subsequent statements.
3. Allocate memory for the XSQLDA using the XSQLDA_LENGTH macro. The
following statement allocates storage for out_sqlda:
out_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(10));

Space for 10 XSQLVAR structures is allocated in this statement, enabling the XSQLDA to
accommodate up to 10 select-list items.
4. Set the version field of the XSQLDA to SQLDA_VERSION1, and set the sqln field
of the XSQLDA to indicate the number of XSQLVAR structures allocated:
out_sqlda->version = SQLDA_VERSION1;
out_sqlda->sqln = 10;

4 Preparing a query statement string

After an XSQLDA is created for holding the items returned by a query statement string, the
statement string can be created, prepared, and described. When a statement string is
executed, InterBase creates the select-list of selected rows.
To prepare a query statement string, follow these steps:
1. Elicit a statement string from the user or create one that contains the SQL
statement to be processed. For example, the following statement creates an
SQL statement string that performs a query:
char *str = "SELECT * FROM CUSTOMER";

The statement appears to have only one select-list item (*). The asterisk is a wildcard
symbol that stands for all of the columns in the table, so the actual number of items
returned equals the number of columns in the table.
2. Parse and name the statement string with PREPARE. The name is used in
subsequent calls to statements such as DESCRIBE and EXECUTE:
EXEC SQL
PREPARE SQL_STMT FROM :str;

SQL_STMT is the name assigned to the prepared statement string.

3. Use DESCRIBE OUTPUT to fill the output XSQLDA with information about the
select-list items returned by the statement:
EXEC SQL
DESCRIBE OUTPUT SQL_STMT INTO SQL DESCRIPTOR out_sqlda;

EMBEDDED SQL GUIDE 279

 CHAPTER 13 USING DYNAMIC SQL

4. Compare the sqln field of the XSQLDA to the sqld field to determine if the
output descriptor can accommodate the number of select-list items specified
in the statement. If not, free the storage previously allocated to the output
descriptor, reallocate storage to reflect the number of select-list items
specified by sqld, reset sqln and version, then execute DESCRIBE OUTPUT
again:
if (out_sqlda->sqld > out_sqlda->sqln)
{
n = out_sqlda->sqld;
free(out_sqlda);
out_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(n));
out_sqlda->sqln = n;
out_sqlda->version = SQLDA_VERSION1;
EXEC SQL
DESCRIBE OUTPUT SQL_STMT INTO SQL DESCRIPTOR out_sqlda;
}

5. Set up an XSQLVAR structure for each item returned. Setting up an item

structure involves the following steps:
- Coercing an item’s datatype (optional).
- Allocating local storage for the data pointed to by the sqldata field of the XSQLVAR.
This step is only required if space for local variables is not allocated until run time.
The following example illustrates dynamic allocation of local variable storage space.
- Providing a NULL value indicator for the parameter.
The following code example illustrates these steps, looping through each XSQLVAR
structure in the out_sqlda XSQLDA:
for (i=0, var = out_sqlda->sqlvar; i < out_sqlda->sqld; i++, var++)
{
dtype = (var->sqltype & ~1) /* drop flag bit for now */
switch (dtype)
{
case SQL_VARYING:
var->sqltype = SQL_TEXT;
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen +
2);
break;
case SQL_TEXT:
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen);
break;
case SQL_LONG:

280 INTERBASE 6
DSQL PROGRAMMING METHODS

var->sqldata = (char *)malloc(sizeof(long));

break;
. . .
/* process remaining types */
} /* end of switch statements */
if (sqltype & 1)
{
/* allocate variable to hold NULL status */
var->sqlind = (short *)malloc(sizeof(short));
}
} /* end of for loop */

For more information about datatype coercion and NULL indicators, see “Coercing
datatypes” on page 271.

4 Executing a statement string within the context of a cursor

To retrieve select-list items from a prepared statement string, the string must be executed
within the context of a cursor. All cursor declarations in InterBase are fixed, embedded
statements inserted into the application before it is compiled. DSQL application
developers must anticipate the need for cursors when writing the application and declare
them ahead of time.
A looping construct is used to fetch a single row at a time from the cursor and to process
each select-list item (column) in that row before the next row is fetched.
To execute a statement string within the context of a cursor and retrieve rows of select-list
items, follow these steps:
1. Declare a cursor for the statement string. For example, the following
statement declares a cursor, DYN_CURSOR, for the SQL statement string,
SQL_STMT:
EXEC SQL
DECLARE DYN_CURSOR CURSOR FOR SQL_STMT;

2. Open the cursor:

EXEC SQL
OPEN DYN_CURSOR;

Opening the cursor causes the statement string to be executed, and an active set of
rows to be retrieved. For more information about cursors and active sets, see Chapter
6, “Working with Data.”

EMBEDDED SQL GUIDE 281

 CHAPTER 13 USING DYNAMIC SQL

3. Fetch one row at a time and process the select-list items (columns) it
contains. For example, the following loops retrieve one row at a time from
DYN_CURSOR and process each item in the retrieved row with an
application-specific function (here called process_column()):
while (SQLCODE == 0)
{
EXEC SQL
FETCH DYN_CURSOR USING SQL DESCRIPTOR out_sqlda;
if (SQLCODE == 100)
break;
for (i = 0; i < out_sqlda->sqld; i++)
{
process_column(out_sqlda->sqlvar[i]);
}
}

The process_column() function mentioned in this example processes each returned

select-list item. The following skeleton code illustrates how such a function can be set
up:
void process_column(XSQLVAR *var)
{
/* test for NULL value */
if ((var->sqltype & 1) && (*(var->sqlind) = -1))
{
/* process the NULL value here */
}
else
{
/* process the data instead */
}
. . .
}

4. When all the rows are fetched, close the cursor:

EXEC SQL
CLOSE DYN_CURSOR;

4 Re-executing a query statement string

Once a query statement string without parameters is prepared, it can be executed as often
as required in an application by closing and reopening its cursor.

282 INTERBASE 6
DSQL PROGRAMMING METHODS

To reopen a cursor and process select-list items, repeat steps 2–4 of “Executing a
Statement String Within the Context of a Cursor,” in this chapter.

Method 4: Query statements with parameters

There are four steps to processing an SQL query statement string with placeholder
parameters:
1. Preparing an input XSQLDA to process a statement string’s parameters.
2. Preparing an output XSQLDA to process the select-list items returned when the
query is executed.
3. Preparing the statement string and its parameters.
4. Using a cursor to execute the statement using input parameter values from
an input XSQLDA, and to retrieve select-list items from the output XSQLDA.

4 Preparing the input XSQLDA

Placeholder parameters are replaced with actual data before a prepared SQL statement
string is executed. Because those parameters are unknown when the statement string is
created, an input XSQLDA must be created to supply parameter values at run time. To
prepare the XSQLDA, follow these steps:
1. Declare a variable to hold the XSQLDA needed to process parameters. For
example, the following declaration creates an XSQLDA called in_sqlda:
XSQLDA *in_sqlda;

2. Optionally declare a variable for accessing the XSQLVAR structure of the

XSQLDA:
XSQLVAR *var;

Declaring a pointer to the XSQLVAR structure is not necessary, but can simplify
referencing the structure in subsequent statements.
3. Allocate memory for the XSQLDA using the XSQLDA_LENGTH macro. The
following statement allocates storage for in_slqda:
in_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(10));

In this statement, space for 10 XSQLVAR structures is allocated, allowing the XSQLDA to
accommodate up to 10 input parameters. Once structures are allocated, assign values
to the sqldata field in each XSQLVAR.

EMBEDDED SQL GUIDE 283

 CHAPTER 13 USING DYNAMIC SQL

4. Set the version field of the XSQLDA to SQLDA_VERSION1, and set the sqln field
of the XSQLDA to indicate the number of XSQLVAR structures allocated:
in_sqlda->version = SQLDA_VERSION1;
in_sqlda->sqln = 10;

4 Preparing the output XSQLDA

Because the number and kind of items returned are unknown when a statement string is
executed, an output XSQLDA must be created to store select-list items that are returned at
run time. To prepare the XSQLDA, follow these steps:
1. Declare a variable to hold the XSQLDA needed to process parameters. For
example, the following declaration creates an XSQLDA called out_sqlda:
XSQLDA *out_sqlda;

2. Optionally declare a variable for accessing the XSQLVAR structure of the

XSQLDA:
XSQLVAR *var;

Declaring a pointer to the XSQLVAR structure is not necessary, but can simplify
referencing the structure in subsequent statements.
3. Allocate memory for the XSQLDA using the XSQLDA_LENGTH macro. The
following statement allocates storage for out_sqlda:
out_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(10));

Space for 10 XSQLVAR structures is allocated in this statement, enabling the XSQLDA to
accommodate up to 10 select-list items.
4. Set the version field of the XSQLDA to SQLDA_VERSION1, and set the sqln field
of the XSQLDA to indicate the number of XSQLVAR structures allocated:
out_sqlda->version = SQLDA_VERSION1;
out_sqlda->sqln = 10;

4 Preparing a query statement string with parameters

After an input and an output XSQLDA are created for holding a statement string’s
parameters, and the select-list items returned when the statement is executed, the
statement string can be created and prepared. When a statement string is prepared,
InterBase replaces the placeholder parameters in the string with information about the
actual parameters used. The information about the parameters must be assigned to the
input XSQLDA (and perhaps adjusted) before the statement can be executed. When the
statement string is executed, InterBase stores select-list items in the output XSQLDA.
To prepare a query statement string with parameters, follow these steps:

284 INTERBASE 6
DSQL PROGRAMMING METHODS

1. Elicit a statement string from the user or create one that contains the SQL
statement to be processed. For example, the following statement creates an
SQL statement string with placeholder parameters:
char *str = "SELECT * FROM DEPARTMENT WHERE BUDGET = ?, LOCATION = ?";

This statement string contains two parameters: a value to be assigned to the BUDGET
field and a value to be assigned to the LOCATION field.
2. Prepare and name the statement string with PREPARE. The name is used in
subsequent calls to DESCRIBE and EXECUTE:
EXEC SQL
PREPARE SQL_STMT FROM :str;

SQL_STMT is the name assigned to the prepared statement string.

3. Use DESCRIBE INPUT to fill the input XSQLDA with information about the
parameters contained in the SQL statement:
EXEC SQL
DESCRIBE INPUT SQL_STMT USING SQL DESCRIPTOR in_sqlda;

4. Compare the sqln field of the XSQLDA to the sqld field to determine if the
input descriptor can accommodate the number of parameters contained in
the statement. If not, free the storage previously allocated to the input
descriptor, reallocate storage to reflect the number of parameters specified by
sqld, reset sqln and version, then execute DESCRIBE INPUT again:
if (in_sqlda->sqld > in_sqlda->sqln)
{
n = in_sqlda->sqld;
free(in_sqlda);
in_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(n));
in_sqlda->sqln = n;
in_sqlda->version = SQLDA_VERSION1;
EXEC SQL
DESCRIBE INPUT SQL_STMT USING SQL DESCRIPTOR in_sqlda;
}

5. Process each XSQLVAR parameter structure in the input XSQLDA. Processing a

parameter structure involves up to four steps:
- Coercing a parameter’s datatype (optional).
- Allocating local storage for the data pointed to by the sqldata field of the XSQLVAR.
This step is only required if space for local variables is not allocated until run time.
The following example illustrates dynamic allocation of local variable storage space.

EMBEDDED SQL GUIDE 285

 CHAPTER 13 USING DYNAMIC SQL

- Providing a value for the parameter consistent with its datatype (required).
- Providing a NULL value indicator for the parameter.
These steps must be followed in the order presented. The following code example
illustrates these steps, looping through each XSQLVAR structure in the in_sqlda
XSQLDA:
for (i=0, var = in_sqlda->sqlvar; i < in_sqlda->sqld; i++, var++)
{
/* Process each XSQLVAR parameter structure here.
The parameter structure is pointed to by var.*/
dtype = (var->sqltype & ~1) /* drop flag bit for now */
switch (dtype)
{
case SQL_VARYING: /* coerce to SQL_TEXT */
var->sqltype = SQL_TEXT;
/* allocate proper storage */
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen);
/* provide a value for the parameter. See case SQL_LONG */
. . .
break;
case SQL_TEXT:
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen);
/* provide a value for the parameter. See case SQL_LONG */
. . .
break;
case SQL_LONG:
var->sqldata = (char *)malloc(sizeof(long));
/* provide a value for the parameter */
*(long *)(var->sqldata) = 17;
break;
. . .
} /* end of switch statement */
if (sqltype & 1)
{
/* allocate variable to hold NULL status */
var->sqlind = (short *)malloc(sizeof(short));
}
} /* end of for loop */

For more information about datatype coercion and NULL indicators, see “Coercing
datatypes” on page 271.

286 INTERBASE 6
DSQL PROGRAMMING METHODS

6. Use DESCRIBE OUTPUT to fill the output XSQLDA with information about the
select-list items returned by the statement:
EXEC SQL
DESCRIBE OUTPUT SQL_STMT INTO SQL DESCRIPTOR out_sqlda;

7. Compare the sqln field of the XSQLDA to the sqld field to determine if the
output descriptor can accommodate the number of select-list items specified
in the statement. If not, free the storage previously allocated to the output
descriptor, reallocate storage to reflect the number of select-list items
specified by sqld, reset sqln and version, and execute DESCRIBE OUTPUT again:
if (out_sqlda->sqld > out_sqlda->sqln)
{
n = out_sqlda->sqld;
free(out_sqlda);
out_sqlda = (XSQLDA *)malloc(XSQLDA_LENGTH(n));
out_sqlda->sqln = n;
out_sqlda->version = SQLDA_VERSION1;
EXEC SQL
DESCRIBE OUTPUT SQL_STMT INTO SQL DESCRIPTOR out_sqlda;
}
8. Set up an XSQLVAR structure for each item returned. Setting up an item
structure involves the following steps:
- Coercing an item’s datatype (optional).
- Allocating local storage for the data pointed to by the sqldata field of the XSQLVAR.
This step is only required if space for local variables is not allocated until run time.
The following example illustrates dynamic allocation of local variable storage space.
- Providing a NULL value indicator for the parameter (optional).
The following code example illustrates these steps, looping through each XSQLVAR
structure in the out_sqlda XSQLDA:
for (i=0, var = out_sqlda->sqlvar; i < out_sqlda->sqld; i++, var++)
{
dtype = (var->sqltype & ~1) /* drop flag bit for now */
switch (dtype)
{
case SQL_VARYING:
var->sqltype = SQL_TEXT;
break;
case SQL_TEXT:
var->sqldata = (char *)malloc(sizeof(char)*var->sqllen);

EMBEDDED SQL GUIDE 287

 CHAPTER 13 USING DYNAMIC SQL

break;
case SQL_LONG:
var->sqldata = (char *)malloc(sizeof(long));
break;
/* process remaining types */
} /* end of switch statements */
if (sqltype & 1)
{
/* allocate variable to hold NULL status */
var->sqlind = (short *)malloc(sizeof(short));
}
} /* end of for loop */

For more information about datatype coercion and NULL indicators, see “Coercing
datatypes” on page 271.

4 Executing a query statement string within the context of a cursor

To retrieve select-list items from a statement string, the string must be executed within
the context of a cursor. All cursor declarations in InterBase are fixed, embedded
statements inserted into the application before it is compiled. DSQL application
developers must anticipate the need for cursors when writing the application and declare
them ahead of time.
A looping construct is used to fetch a single row at a time from the cursor and to process
each select-list item (column) in that row before the next row is fetched.
To execute a statement string within the context of a cursor and retrieve rows of select-list
items, follow these steps:
1. Declare a cursor for the statement string. For example, the following
statement declares a cursor, DYN_CURSOR, for the prepared SQL statement
string, SQL_STMT:
EXEC SQL
DECLARE DYN_CURSOR CURSOR FOR SQL_STMT;

2. Open the cursor, specifying the input descriptor:

EXEC SQL
OPEN DYN_CURSOR USING SQL DESCRIPTOR in_sqlda;

Opening the cursor causes the statement string to be executed, and an active set of
rows to be retrieved. For more information about cursors and active sets, see Chapter
6, “Working with Data.”

288 INTERBASE 6
DSQL PROGRAMMING METHODS

3. Fetch one row at a time and process the select-list items (columns) it
contains. For example, the following loops retrieve one row at a time from
DYN_CURSOR and process each item in the retrieved row with an
application-specific function (here called process_column()):
while (SQLCODE == 0)
{
EXEC SQL
FETCH DYN_CURSOR USING SQL DESCRIPTOR out_sqlda;
if (SQLCODE == 100)
break;
for (i = 0; i < out_sqlda->sqld; i++)
{
process_column(out_sqlda->sqlvar[i]);
}
}

4. When all the rows are fetched, close the cursor:

EXEC SQL
CLOSE DYN_CURSOR;

4 Re-executing a query statement string with parameters

Once a query statement string with parameters is prepared, it can be used as often as
required in an application. Before each subsequent use, the input XSQLDA can be supplied
with new parameter and NULL indicator data. The cursor must be closed and reopened
before processing can occur.
To provide new parameters to the input XSQLDA, follow steps 3–5 of “Preparing a Query
Statement String with Parameters,” in this chapter.
To provide new information to the output XSQLDA, follow steps 6–8 of “Preparing a Query
Statement String with Parameters,” in this chapter.
To reopen a cursor and process select-list items, repeat steps 2–4 of “Executing a Query
Statement String Within the Context of a Cursor,” in this chapter.

EMBEDDED SQL GUIDE 289

 CHAPTER 13 USING DYNAMIC SQL

290 INTERBASE 6
 CHAPTER

Preprocessing, Compiling,
Chapter14
14
and Linking

This chapter describes how to preprocess a program by using gpre, and how to compile
and link it for execution.

Preprocessing
After coding an SQL or dynamic SQL (DSQL) program, the program must be
preprocessed with gpre before it can be compiled. gpre translates SQL and DSQL
commands into statements the host-language compiler accepts by generating InterBase
library function calls. gpre translates SQL and DSQL database variables into ones the
host-language compiler accepts and declares these variables in host-language format.
gpre also declares certain variables and data structures required by SQL, such as the
SQLCODE variable and the extended SQL descriptor area (XSQLDA) used by DSQL.

Using gpre
The syntax for gpre is:
gpre [-language] [-options] infile [outfile]

EMBEDDED SQL GUIDE 291

 CHAPTER 14 PREPROCESSING, COMPILING, AND LINKING

The infile argument specifies the name of the input file.

The optional outfile argument specifies the name of the output file. If no file is specified,
gpre sends its output to a file with the same name as the input file, with an extension
depending on the language of the input file.
gpre has switches that allow you to specify the language of the source program and a
number of other options. You can place the switches either before or after the input and
output file specification. Each switch must include at least a hyphen preceded by a space
and a unique character specifying the switch.

4 Language switches
The language switch specifies the language of the source program. C and C++ are
languages available on all platforms. The switches are shown in the following table:

Switch Language
-c C
-cxx C++
TABLE 14.1 gpre language switches available on all platforms

In addition, some platforms support other languages if an additional InterBase license

for the language is purchased. The following table lists the available languages and the
corresponding switches:

Switch Language
-al[sys] Ada (Alsys)
-a[da] Ada (VERDIX, VMS, Telesoft)
-ansi ANSI-85 COBOL
-co[bol] COBOL
-f[ortran] FORTRAN
-pa[scal] Pascal
TABLE 14.2 Additional gpre language switches

For example, to preprocess a C program called census.e, type:

gpre -c census.e

292 INTERBASE 6
PREPROCESSING

4 Option switches
The option switches specify preprocessing options. The following table describes the
available switches:

Switch Description
-charset name Determines the active character set at compile time, where name is the
character set name.
-d[atabase] filename Declares a database for programs. filename is the file name of the database
to access. Use this option if a program contains SQL statements and does
not attach to the database itself. Do not use this option if the program
includes a database declaration.
-d_float VAX/VMS only. Specifies that double-precision data will be passed from the
application in D_FLOAT format and stored in the database in G_FLOAT
format. Data comparisons within the database will be performed in
G_FLOAT format. Data returned to the application from the database will be
in D_FLOAT format.
-e[ither_case] Enables gpre to recognize both uppercase and lowercase. Use the
-either_case switch whenever SQL keywords appear in code in lowercase
letters. If case is mixed, and this switch is not used, gpre cannot process the
input file. This switch is not
necessary with languages other than C, since they are case-insensitive.
-m[anual] Suppresses the automatic generation of transactions. Use the
-m switch for SQL programs that perform their own transaction handling,
and for all DSQL programs that must, by definition, explicitly control their
own transactions.
-n[o_lines] Suppresses line numbers for C programs.
-o[utput] Directs gpre’s output to standard output, rather than to a file.
-password password Specifies password, the database password, if the program connects to a
database that requires one.
TABLE 14.3 gpre option switches

EMBEDDED SQL GUIDE 293

 CHAPTER 14 PREPROCESSING, COMPILING, AND LINKING

Switch Description
-r[aw] Prints BLR as raw numbers, rather than as their mnemonic equivalents. This
option cam be useful for making the gpre output file smaller; however, it
will be unreadable.
-sqldialect Sets the SQL dialect. Valid values are 1, 2, and 3.
-user username Specifies username, the database user name, if the program connects to a
database that requires one.
-x handle Gives the database handle identified with the -database option an external
declaration. This option directs the program to pick up a global declaration
from another linked module. Use only if the -d switch is also used.
-z Prints the version number of gpre and the version number of all declared
databases. These databases can be declared either in the program or with
the -database switch.
TABLE 14.3 gpre option switches (continued)

For sites with the appropriate license and are using a language other than C, additional
gpre options can be specified, as described in the following table:

Switch Description
-h[andles] pkg Specifies, pkg, an Ada handles package.
TABLE 14.4 Language-specific gpre option switches

4 Examples
The following command preprocesses a C program in a file named appl1.e. The output
file will be appl1.c. Since no database is specified, the source code must connect to the
database.
gpre -c appl1

The following command is the same as the previous, except that it does not assume the
source code opens a database, instead, explicitly declaring the database, mydb.gdb:
gpre -c appl1 -d mydb.gdb

294 INTERBASE 6
PREPROCESSING

Using a file extension to specify language

In addition to using a language switch to specify the host language, it is also possible to
indicate the host language with the file-name extension of the source file. The following
table lists the file-name extensions for each language that gpre supports and the default
extension of the output file:

Input file Default output file

Language extension extension
Ada ( VERDIX) ea a
Ada (Alsys, Telesoft) eada ada
C e c
C++ exx cxx
COBOL ecob cob
FORTRAN ef f
Pascal epas pas
TABLE 14.5 File extensions for language specification

For example, to preprocess a COBOL program called census.ecob, type:

gpre census_report.ecob

This generates an output file called census.cob.

When specifying a file-name extension, it is possible to specify a language switch as well:
gpre -cob census.ecob

Specifying the source file

Because both the language switch and the filename extension are optional, gpre can
encounter three different situations:
g A language switch and input file with no extension
g No language switch, but an input file with extension
g Neither a language switch, nor a file extension
This section describes gpre’s behavior in each of these cases.

EMBEDDED SQL GUIDE 295

 CHAPTER 14 PREPROCESSING, COMPILING, AND LINKING

Language switch and no input file extension

If gpre encounters a language switch, but the specified input file has no extension, it does
the following:
1. It looks for the input file without an extension. If gpre finds the file, it
processes it and generates an output file with the appropriate extension.
If gpre does not find the input file, it looks for the file with the extension that
corresponds to the indicated language. If it finds such a file, it generates an output
file with the appropriate extension.
2. If gpre cannot find either the named file or the named file with the appropriate
extension, it returns the following error:
gpre: can’t open filename or filename.extension

filename is the file specified in the gpre command. extension is the language-specific
file extension for the specified program.
For example, suppose the following command is issued:
gpre -c census

gpre performs the following sequence of actions:

1. It looks for a file called census without an extension. If it finds the file, it
processes it and generates census.c.
2. If it cannot find census, it looks for a file called census.e. If it finds census.e, it
processes the file and generates census.c.
3. If it cannot find census or census.e, it returns this error:
gpre: can’t open census or census.e

No language switch and an input file with extension

If a language switch is not specified, but the input file includes a file-name extension,
gpre looks for the specified file and assumes the language is indicated by the extension.
For example, suppose the following command is processed:
gpre census.e

gpre looks for a file called census.e. If gpre finds this file, it processes it as a C program and
generates an output file called census.c. If gpre does not find this file, it returns the
following error:
gpre: can’t open census.e

296 INTERBASE 6
COMPILING AND LINKING

Neither a language switch nor a file extension

If gpre finds neither a language extension nor a filename extension, it looks for a file in
the following order:
1. filename.e (C)
2. filename.epas (Pascal)
3. filename.ef (FORTRAN)
4. filename.ecob (COBOL)
5. filename.ea (VERDIX Ada)
6. filename.eada (Alsys, and Telesoft Ada)
If gpre finds such a file, it generates an output file with the appropriate extension. If gpre
does not find the file, it returns the following error:
gpre: can’t find filename with any known extension. Giving up.

Compiling and linking

After preprocessing a program, it must be compiled and linked. Compiling creates an
object module from the preprocessed source file. Use a host-language compiler to
compile the program.
The linking process resolves external references and creates an executable object. Use the
tools available on a given platform to link a program’s object module to other object
modules and libraries, based on the platform, operating system and host language used.

Compiling an Ada program

Before compiling an Ada program, be sure the Ada library contains the package
interbase.ada (or interbase.a for VERDIX Ada). This package is in the InterBase include
directory.
To use the programs in the InterBase examples directory, use the package basic_io.ada (or
basic_io.a for VERDIX Ada), also located in the examples directory.

EMBEDDED SQL GUIDE 297

 CHAPTER 14 PREPROCESSING, COMPILING, AND LINKING

Linking
On Unix platforms, programs can be linked to the following libraries:
g A library that uses pipes, obtained with the -lgds option. This library yields faster links and
smaller images. It also lets your application work with new versions of InterBase
automatically when they are installed.
g A library that does not use pipes, obtained with the -lgds_b option. This library has faster
execution, but binds an application to a specific version of InterBase. When installing a
new version of InterBase, programs must be relinked to use the new features or databases
created with that version.
Under SunOS-4, programs can be linked to a shareable library by using the
-lgdslib option. This creates a dynamic link at run time and yields smaller images with the
execution speed of the full library. This option also provides the ability to upgrade
InterBase versions automatically.
For specific information about linking options for InterBase on a particular platform,
consult the online readme in the interbase directory.

298 INTERBASE 6
 APPENDIX

InterBase Document
AppendixA
A
Conventions

This appendix covers the following topics:

g The InterBase 6 documentation set
g The printing conventions used to display information in text
g The printing conventions used to display information in syntax, code, and examples

EMBEDDED SQL GUIDE 299

 APPENDIX A INTERBASE DOCUMENT CONVENTIONS

The InterBase documentation set

The InterBase documentation set is an integrated package designed for all levels of users.
It consists of six full-length printed books plus the Installation Guide. Each of these
books is also provided in Adobe Acrobat PDF format and is accessible on line. If Adobe
Acrobat is not already installed on your system, you can find it on the InterBase
distribution CD-ROM or at http//www.adobe.com/prodindex/acrobat/readstep.html. Acrobat is
available for Windows NT, Windows 95, and most flavors of UNIX.
Book Description

Operations Guide Provides an introduction to InterBase and an explanation of tools and

procedures for performing administrative tasks on databases and
database servers; also includes full reference on InterBase utilities,
including isql, gbak, gfix, and others
Data Definition Guide Explains how to create, alter, and delete database objects using the SQL
language
Developer’s Guide Provides both reference and task-oriented material for users of the
Borland RAD tools (Delphi, C++ Builder, and JBuilder); includes chapters
on writing UDFs, driver configuration, developing embedded installation
applications, and using the new InterBase Data Access Components
Language Reference Describes the SQL language syntax and usage; includes references for
procedure and trigger language, InterBase keywords, functions in the
InterBase UDF library, error codes, character sets, and the system tables
Embedded SQL Guide (formerly called the Programmer’s Guide) Describes how to write
embedded SQL database applications in a host language, precompiled
through gpre
API Guide Explains how to write database applications using the InterBase API
TABLE A.1 Books in the InterBase 6 documentation set

300 INTERBASE 6
PRINTING CONVENTIONS

Printing conventions
The InterBase documentation set uses various typographic conventions to identify objects
and syntactic elements.
The following table lists typographic conventions used in text, and provides examples of
their use:
Convention Purpose Example
UPPERCASE SQL keywords, SQL functions, and names of • the SELECT statement retrieves data from the CITY column
all database objects such as tables, columns, in the CITIES table
indexes, and stored procedures • can be used in CHAR, VARCHAR, and BLOB text columns
• the CAST() function
italic New terms, emphasized words, all elements • isc_decode_date()
from host languages, and all user-supplied • the host variable, segment_length
items • contains six variables, or data members
bold File names, menu picks, and all commands • gbak, isql, gsec. gfix
that are entered at a system prompt, • specify the gpre -sqlda old switch
including their switches, arguments, and • a script, ib_udf.sql, in the examples subdirectory
parameters
• the employee.gdb database; the employee database
• the Session | Advanced Settings command
TABLE A.2 Text conventions

EMBEDDED SQL GUIDE 301

 APPENDIX A INTERBASE DOCUMENT CONVENTIONS

Syntax conventions
The following table lists the conventions used in syntax statements and sample code, and
provides examples of their use:
Convention Purpose Example
UPPERCASE Keywords that must be typed exactly as •SET TERM !!;
they appear when used •ADD [CONSTRAINT] CHECK

italic User-supplied parameters that cannot be •CREATE TRIGGER name FOR table;
broken into smaller units •ALTER EXCEPTION name 'message'

<italic> Parameters in angle brackets can be WHILE (<condition>) DO <compound_statement>

broken into smaller syntactic units; the
expansion syntax for these parameters
follows the current syntax statement
[ ] Optional syntax: you do not need to •CREATE [UNIQUE][ASCENDING | DESCENDING]
include anything that is enclosed in •[FILTER [FROM subtype] TO subtype]
square brackets; when elements within
these brackets are separated by the pipe
symbol (|), you can choose only one
{ } You must include one and only one of the {INTO | USING}
enclosed options, which are separated by
the pipe symbol (|)
| You can choose only one of a group whose SELECT [DISTINCT | ALL]
elements are separated by this pipe
symbol
... You can repeat the clause enclosed in (<col> [,<col>…])
brackets with the “…” symbol as many
times as necessary
TABLE A.3 Syntax conventions

302 INTERBASE 6
Index

* (asterisk), in code 128 altering

* operator 113 column definitions 103
+ operator 113 metadata 100–106
/ operator 113 views 100, 105
[ ] (brackets), arrays 216, 219–220 AND operator 113
|| operator 112 ANY operator 115, 120
– operator 113 API calls for Blob data 202
appending tables 141
applications 21
A
See also DSQL applications
access mode parameter 62 building 90
default transactions 56 event handling 233, 235–237
access privileges See security porting 22, 246
accessing preprocessing See gpre
arrays 218–224 arithmetic expressions 224
Blob data 202 arithmetic functions See aggregate functions
data 25, 45, 48 arithmetic operators 113
actions See events precedence 113, 123
active database 33 array elements 220
Ada programs 297 defined 215
adding evaluating 223
See also inserting porting 217
columns 101 retrieving 220
addition operator (+) 113 array IDs 219
aggregate functions 129–130 array slices 220–221
arrays and 218 adding data 218
NULL values 130 defined 218
alerter (events) 234 updating data 222
aliases arrays 109
database 36 See also error status array
tables 134 accessing 218–224
ALIGN macro 272–273 aggregate functions 218
ALL keyword 48 creating 216–217
ALL operator 115, 120 cursors 219, 222
allocating memory 47 DSQL applications and 218
ALTER INDEX 105–106 inserting data 219
ALTER TABLE 101–104 multi-dimensional 216, 221
ADD option 101 referencing 218
DROP option 102 search conditions 223–224

i
 selecting data 218–221 brackets ([ ]), arrays 216, 219–220
storing data 215 buffers
subscripts 217, 224 database cache 47–48
UDFs and 218 byte-matching rules 119
updating 222
views and 218
C
ASC keyword 138
ascending sort order 97, 138 C language
asterisk (*), in code 128 character variables 24
attaching to databases 27, 41 host terminator 28
multiple 38, 43–46 host-language variables 23–25
averages 129 cache buffers 47–48
CACHE keyword 47
calculations 113, 129
B case-insensitive comparisons 116
BASED ON 24 case-sensitive comparisons 117, 119
arrays and 219 CAST() 125, 186
basic_io.a 297 CHAR datatype
basic_io.ada 297 converting to DATE 186
BEGIN DECLARE SECTION 23 description 108
BETWEEN operator 115 CHAR VARYING keyword 109
NOT operator and 115 CHARACTER keyword 108
binary large objects See Blob character sets
Blob API functions 202 converting 119
Blob data 196–213 default 90
deleting 201–202 NONE 90
filtering 203–213 specifying 40, 90
inserting 199–200 character strings
selecting 196–199 comparing 116, 117, 119
storing 192, 194 literal file names 42–44
updating 200–201 CHARACTER VARYING keyword 109
Blob filter function 206 closing
action macro definitions 211–212 databases 30, 38, 49–51
return values 213 multiple 38
Blob filters 203–213 transactions 28–29
external 203 coercing datatypes 271–272
declaring 204 COLLATE clause 136
writing 205 collation orders
invoking 205 GROUP BY clause 140
text 203 ORDER BY clause 139
types 206 WHERE clause 137
Blob segments 194–196 column names
Blob subtypes 193 qualifying 130
Boolean expressions 135 views 94
evaluating 113 column-major order 217
Borland C/C++ See C language columns

ii INTERBASE 6
 adding 101 specifying character sets 90
altering 103 CREATE DOMAIN 90–91
computed 93, 102 arrays 216
creating 91 CREATE GENERATOR 98
defining CREATE INDEX 96–97
views 94 DESCENDING option 97
dropping 102 UNIQUE option 97
selecting 128–130 CREATE PROCEDURE 226
eliminating duplicates 128 CREATE TABLE 91–94
sorting by 138 arrays 216
using domainst 90 multiple tables 92
values, returning 129 CREATE VIEW 94–96
COMMIT 29, 50, 54, 74–78 WITH CHECK OPTION 96
multiple databases 38 creating
comparison operators 114–122 arrays 216–217
NULL values and 114, 121 columns 91
precedence 123 computed columns 93, 102
subqueries 115, 117–122 integrity constraints 91
COMPILETIME keyword 38 metadata 88–98
compiling cursors 144
programs 297–298 arrays 219, 222
computed columns multiple transaction programs 80
creating 93, 102 select procedures 228
defined 93
concatenation operator (||) 112
D
CONNECT 27, 35, 41–48
ALL option 48 data 107
CACHE option 47 accessing 25, 45, 48
error handling 46 DSQL applications 25, 33
multiple databases 43–46 host-language variables and 23
omitting 27 changes
SET DATABASE and 42 committing See COMMIT
constraints 91, 99, 100, 101, 102 rolling back See ROLLBACK
See also specific constraints defining 87
optional 92 protecting See security
CONTAINING operator 116 retrieving
NOT operator and 116 optimizing 142, 225
conversion functions 125–126, 186 selecting 96, 110, 126
converting multiple tables 130, 133
datatypes 125 storing 215
date and time datatypes 187 data structures
dates 183–189 host-language 25
international character sets 119 database cache buffers 47–48
CREATE DATABASE 89–90 database handles 26, 36, 42
in DSQL 258 DSQL applications 30, 32
global 39

iii
 multiple databases 37–38, 44 default transactions
naming 36 access mode parameter 56
scope 39 default behavior 56
transactions and 36, 49 DSQL applications 57
database specification parameter 54, 61 isolation level parameter 56
databases lock resolution parameter 56
attaching to 27, 41 rolling back 29
multiple 38, 43–46 starting 55–57
closing 30, 38, 49–51 deleting See dropping
creating 89–90 DESC keyword 138
declaring multiple 25–27, 37–39 DESCENDING keyword 97
DSQL and attaching 256 descending sort order 97, 138
initializing 25–28 detaching from databases 38, 50
naming 41 directories
opening 35, 41, 43 specifying 36
remote 89 dirty reads 64
datatypes 108–109 DISCONNECT 30, 49
coercing 271–272 multiple databases 38, 50
compatible 125 DISTINCT keyword 128
converting 125 division operator (/) 113
DSQL applications 270–273 domains
macro constants 268–270 creating 90–91
DATE datatype DOUBLE PRECISION datatype 108
converting 186, 187 DROP INDEX 99
description 108, 183 DROP TABLE 100
date literals 111, 189 DROP VIEW 99
dates dropping
converting 183–189 columns 102
inserting 185, 185–186 metadata 98–100
selecting 184–185 DSQL
selecting from tables 183 CREATE DATABASE 258
updating 186 limitations 256
DECIMAL datatype 108 macro constants 268–270
declarations, changing scope 39 programming methods 273–289
DECLARE CURSOR 80 requirements 30–32
DECLARE TABLE 93 DSQL applications 21, 255
declaring accessing data 25, 33
Blob filters 204 arrays and 218
host-language variables 22–25 attaching to databases 256
multiple databases 25–27, 37–39 creating databases 258
one database only 27, 35 data definition statements 87
SQLCODE variable 28 database handles 30, 32
transaction names 59 datatypes 270–273
XSQLDAs 31–32 default transactions 57
default character set 90 executing stored procedures 230

iv INTERBASE 6
 multiple transactions 83 posting 234
porting 22 responding to 237
preprocessing 34, 55, 291 executable objects 297
programming requirements 30–34 executable procedures 226, 229–231
SQL statements DSQL 230
embedded 33 input parameters 229–230
transaction names 30, 32–34 EXECUTE 30, 33
transactions 32 EXECUTE IMMEDIATE 31, 33, 84
writing 259 EXECUTE PROCEDURE 229
XSQLDAs 262–273 EXISTS operator 115, 121
DSQL limitations 32–34 NOT operator and 121
DSQL statements 255 expression-based columns See computed columns
dynamic link libraries See DLLs expressions 135
dynamic SQL See DSQL evaluating 113
extended SQL descriptior areas See XSQLDAs
EXTERN keyword 40
E
END DECLARE SECTION 23
error codes and messages 28, 251 F
capturing 248–251 file names
displaying 247 specifying 42–44
error status array 247, 251 files
error-handling routines 46, 239, 246 See also specific files
changing 242 source, specifying 295
disabling 246 FLOAT datatype 108
guidelines 246 FROM keyword 132–134
nesting 246 functions
testing SQLCODE directly 243, 244 aggregate 129–130
WHENEVER and 240–243, 244 conversion 125–126, 186
errors 28 error-handling 246
run-time numeric 98
recovering from 239 user-defined See UDFs
trapping 240, 242, 251
unexpected 246
G
user-defined See exceptions
ESCAPE keyword 118 GEN_ID() 98
EVENT INIT 235 generators
multiple events 236 creating 98
EVENT WAIT 236–237 defined 98
events 233–238 global column definitions 90
See also triggers global database handles 39
alerter 234 gpre 34, 83, 291–297
defined 233 command-line options 293–294
manager 233 databases, specifying 38
multiple 236–237 DSQL applications 55
notifying applications 235–236 handling transactions 257

v
 language options 292 INDICATOR keyword 230
file names vs. 295–297 indicator variables 230
-m switch 55, 89 NULL values 230
programming requirements 21 initializing
specifying source files 295 databases 25–28
-sqlda old switch 31 transaction names 59
syntax 291 input parameters 227, 229–230
group aggregates 139 See also stored procedures
grouping rows 139 INSERT
restrictions 140 arrays 219
inserting
See also adding
H
Blob data 199–200
hard-coded strings dates 185–186
file names 42–44 INTEGER datatype 108
HAVING keyword 140 integrity constraints 91
header files See ibase.h See also specific type
host languages 28 optional 92
data structures 25 Interactive SQL See isql
host-language variables 42 interbase.a 297
arrays 224 interbase.ada 297
declaring 22–25 international character sets 119
specifying 131 INTO keyword 131, 143
hosts, specifying 36 IS NULL operator 118
NOT operator and 119
I isc_blob_ctl 208
I/O See input, output field descriptions 209
ibase.h 31, 251 isc_blob_default_desc() 202
identifiers 36 isc_blob_gen_bpb() 202
database handles 36 isc_blob_info() 202
databases 41 isc_blob_lookup_desc() 202
views 94 isc_blob_set_desc() 202
IN operator 116 isc_cancel_blob() 202
NOT operator and 117 isc_close_blob() 202
INDEX keyword 143 isc_create_blob2() 202
indexes isc_decode_date() 185
altering 100, 105–106 isc_decode_sql_date() 183
creating 96–97 isc_decode_sql_time() 183
dropping 99 isc_decode_timestamp() 183
preventing duplicate entries 97 isc_encode_date() 186
primary keys 98 isc_encode_sql_date() 183
sort order 97 isc_encode_timestamp() 183
changing 106 isc_get_segment() 202
system-defined 97 isc_interprete() 248, 249–251
unique 97 isc_open_blob2() 202

vi INTERBASE 6
isc_put_segment() 202 maximum values 129
ISC_QUAD structure 184–185 memory
isc_sql_interprete() 248–249 allocating 47
isc_status 247, 251 metadata 87
ISC_TIMESTAMP 184 altering 100–106
isolation level parameter 54, 61, 63 creating 88–98
default transactions 56 dropping 98–100
failing 100
Microsoft C/C++ See C language
J
minimum values 129
JOIN keyword 143 modifying See altering;updating
joins 134 modules
object 297
K multi-column sorts 138
key constraints See FOREIGN KEY constraints; multi-dimensional arrays
PRIMARY KEY constraints creating 216
keys selecting data 221
primary 98 multi-module programs 40
multiple databases
attaching to 38, 43–46
L closing 38
language options (gpre) 292 database handles 37–38, 44
file names vs. 295–297 declaring 25–27, 37–39
libraries detaching 38, 50
dynamic link See DLLs opening 43
Unix platforms 298 transactions 49
LIKE operator 117 multiple tables
NOT operator and 118 creating 92
limbo transactions 28 selecting data 130, 133
linking multiple transactions 130
programs 297–298 DSQL applications 83
literal strings, file names 42–44 running 79–85
literal symbols 118 multiplication operator (*) 113
lock resolution parameter 54, 61, 69 multi-row selects 132, 144–152
default transactions 56
logical operators 113–114
precedence 114, 124 N
loops See repetitive statements named transactions 54, 72
lost updates 64 starting 57–58
names
column 94, 130
M qualifying 36, 37, 49
-m switch 55 in SELECT statements 130
macro constants 268–270 specifying at run time 42
mathematical operators 113 naming
precedence 113, 123 database handles 36

vii
 databases 41 data retrieval 142, 225
transactions 58–60 OR operator 113, 114
views 94 ORDER keyword 143
NATURAL keyword 143 order of evaluation (operators) 122–124
NO RECORD_VERSION 61 changing 124
NO WAIT 61, 69 output parameters
NONE character set option 90 See also stored procedures
non-reproducible reads 64
NOT operator 113
P
BETWEEN operator and 115
CONTAINING operator and 116 parameters
EXISTS operator and 121 access mode 56, 62
IN operator and 117 database specification 54, 61, 72
IS NULL operator and 119 isolation level 54, 56, 61, 63
LIKE operator and 118 lock resolution 54, 56, 61, 69
SINGULAR operator and 122 table reservation 54, 61, 71
STARTING WITH operator and 119 unknown 230
NOW 111 phantom rows 64
NOW date literal 189 PLAN keyword 142
NULL values porting
aggregate functions 130 applications 22, 246
arrays and 218 arrays 217
comparisons 114, 121 POST_EVENT 234
indicator variables 230 precedence of operators 122–124
numbers changing 124
generating 98 PREPARE 30, 83
NUMERIC datatype 109 preprocessor See gpre
converting to DATE 186 PRIMARY KEY constraints 97
numeric function 98 primary keys 98
numeric values See values privileges See security
procedures See stored procedures
programming
O DSQL applications 30–34
object modules 297 gpre 21
opening programs
databases 35, 41, 43 compiling and linking 297–298
multiple 43 projection (defined) 126
operators PROTECTED READ 71
arithmetic 113 PROTECTED WRITE 71
comparison 114–122 protecting data See security
concatenation 112
logical 113–114
Q
precedence 122–124
changing 124 qualify (defined) 36, 49
string 112 queries 96, 126
optimizing See also SQL

viii INTERBASE 6
 eliminating duplicate columns 128 changing 39
grouping rows 139 database handles 39
multi-column sorts 138 WHENEVER 242
restricting row selection 134, 140 search conditions (queries) 110–122, 135–137
search conditions 110–122, 135–137 arrays and 223–224
arrays and 223–224 combining simple 113
combining simple 113 reversing 113
reversing 113 SELECT 110–122, 126–144, 227
selecting multiple rows 132, 144–152 arrays 218–221
selecting single rows 143 CREATE VIEW and 94, 95
sorting rows 138 DISTINCT option 128
specific tables 132–134 FROM clause 132–134
with joins 134, 143 GROUP BY clause 139–140
query optimizer 142 collation order 140
HAVING clause 140
INTO option 131, 143
R
ORDER BY clause 138
READ COMMITTED 61, 63, 65 collation order 139
read-only views 95 PLAN clause 142
RECORD_VERSION 61 TRANSACTION option 130
remote databases 89 WHERE clause 110–125, 134–137, 144
RESERVING clause 61, 70 ALL operator 120
table reservation options 71 ANY operator 120
result tables 144 BETWEEN operator 115
See also joins CAST option 125, 186
ROLLBACK 29, 50, 54, 74, 78–79 collation order 137
multiple databases 38 CONTAINING operator 116
rollbacks 29 EXISTS operator 121
routines 226 IN operator 117
See also error-handling routines IS NULL operator 119
row-major order 217 LIKE operator 118
rows SINGULAR operator 121
counting 129 SOME operator 120
grouping 139 STARTING WITH operator 119
restrictions 140 select procedures 226, 227–229
selecting 134 calling 228
multiple 132, 144–152 cursors 228
single 143 input parameters 227
sorting 138 selecting 132
run-time errors tables vs. 227
recovering from 239 views vs. 227
RUNTIME keyword 39 SELECT statements
singleton SELECTs 126, 131, 143
S selecting
scope Blob data 196–199

ix
 columns 128–130 file names 42–44
data 96, 110, 126 host-language variables 131
See also SELECT hosts 36
dates 184–185 SQL statements
multiple rows 132, 144–152 DSQL applications 33
single rows 143 strings 261
views 132 sql_encode_sql_time() 183
SET DATABASE 26, 35 SQLCODE variable
COMPILETIME option 38 declaring 28
CONNECT and 42 examining 240
DSQL applications 32 return values 240, 247, 251
EXTERN option 40 displaying 247
multiple databases and 37, 44 testing 243, 244
omitting 27, 44 SQLDAs 31
RUNTIME option 39 porting applications and 22
STATIC option 40 starting default transactions 55–57
SET NAMES 35 STARTING WITH operator 119
SET TRANSACTION 54, 56, 60–72 NOT operator and 119
access mode parameter 54 statements
parameters 61 See also DSQL statements; SQL statements
syntax 62 data definition 87
SHARED READ 71 data structures and 25
SHARED WRITE 71 embedded 28, 107
singleton SELECTs 126, 131 error-handling 246
defined 143 transaction management 53, 54
SINGULAR operator 115, 121 STATIC keyword 40
NOT operator and 122 status array See error status array
SMALLINT datatype 109 sticky sort order 138
SNAPSHOT 61, 63, 65 stored procedures 225–231, 233
SNAPSHOT TABLE STABILITY 61, 63, 68 defined 225
SOME operator 115, 120 return values 226, 230
SORT MERGE keywords 143 values 226, 230
sort order XSQLDAs and 230
ascending 97, 138 string operator (||) 112
descending 97, 138 subqueries
indexes 97, 106 comparison operators 115, 117–122
queries 138 defined 161
sticky 138 subscripts (arrays) 217, 224
sorting subtraction operator (-) 113
multiple columns 138 SunOS-4 platforms 298
rows 138 system tables 89
source files 295 system-defined indexes 97
specifying
character sets 40, 90
T
directories 36
table names

x INTERBASE 6
 aliases 134 named 54, 72
duplicating 92 starting 57–58
identical 36, 37, 49 naming 58–60
table reservation parameter 54, 61 rolling back 29
tables running multiple 79–85, 130
altering 101–104 unnamed 29
appending with UNION 141 trapping
creating 91–94 errors 240, 242, 251
multiple 92 triggers 233
declaring 93
dropping 100
U
qualifying 36, 37, 49
querying specific 132–134 UDFs
select procedures vs. 227 arrays and 218
TIME datatype unexpected errors 246
converting 186, 187 UNION
time structures 184 appending tables 141
time.h 184 in SELECT statements 127
times unique indexes 97
inserting 185 UNIQUE keyword 97
selecting from tables 183 unique values 98
updating 186 Unix platforms 298
TIMESTAMP datatype unknown values, testing for 118
converting 186, 187 unrecoverable errors 246
TODAY 111 updatable views 95
TODAY date literal 189 UPDATE
TOMORROW 111 arrays 223
totals, calculating 129 dates and times 186
TRANSACTION keyword 130 update side effects 64
transaction management statements 53, 54 updating
transaction names 57, 257 See also altering
declaring 59 updating Blob data 200–201
DSQL applications 30, 32–34 user-defined functions See UDFs
initializing 59 USING clause 61, 72
multi-table SELECTs 130
transactions 226 V
accessing data 48 values
closing 28–29 See also NULL values
committing 29 comparing 114
database handles and 36, 49 manipulating 113
default 55–57 matching 116, 120
rolling back 29 maximum 129
DSQL applications 32 minimum 129
ending 74 selecting 129
multiple databases 49 stored procedures 226, 230

xi
 unique 98 embedding 242
unknown, testing for 118 limitations 242
VARCHAR datatype 109 scope 242
variables WHERE clause See SELECT
host-language 42 WHERE keyword 134
arrays 224 wildcards in string comparisons 117
declaring 22–25 writing external Blob filters 205
specifying 131
indicator 230
X
views 94
altering 100, 105 XSQLDA_LENGTH macro 267
arrays and 218 XSQLDAs 262–273
creating 94–96 declaring 31–32
defining columns 94 fields 264
dropping 99 input descriptors 266
naming 94 output descriptors 266
read-only 95 porting applications and 22
select procedures vs. 227 stored procedures and 230
selecting 132 structures 31
updatable 95 XSQLVAR structure 262
virtual tables 94 fields 265

W Y
WAIT 61, 69 YESTERDAY 111
WHENEVER 240–243, 244

xii INTERBASE 6