summary | shortlog | log | commit | commitdiff | tree
raw | patch | inline | side by side (parent: 3d92252)
raw | patch | inline | side by side (parent: 3d92252)
| author | Tom Lane <tgl@sss.pgh.pa.us> | |
| Thu, 10 Sep 2020 21:43:16 +0000 (17:43 -0400) | ||
| committer | Tom Lane <tgl@sss.pgh.pa.us> | |
| Thu, 10 Sep 2020 21:43:16 +0000 (17:43 -0400) |
Also set the release date ... hopefully we won't have to change that.
| doc/src/sgml/release-13.sgml | patch | blob | blame | history |
index afeec3e953e8e3b020e030b4df72271bd89e2871..0d2a928e0ed5bd16229345c1d3bca8e67f6d92e4 100644 (file)
<formalpara>
<title>Release date:</title>
- <para>2020-XX-XX, CURRENT AS OF 2020-08-09</para>
+ <para>2020-09-24</para>
</formalpara>
<sect2>
<para>
This new behavior matches the <acronym>SQL</acronym> specification.
- Previously this caused the escape to be set to the default backslash
- character. The previous behavior has been retained in old views
- by keeping the original function unchanged. This also applies to
- <command>substring(text FROM pattern ESCAPE text)</command>.
+ Previously a null <literal>ESCAPE</literal> value was taken to mean
+ using the default escape string (a backslash character). This also
+ applies to <literal>substring(<replaceable>text</replaceable>
+ FROM <replaceable>pattern</replaceable>
+ ESCAPE <replaceable>text</replaceable>)</literal>. The previous
+ behavior has been retained in old views by keeping the original
+ function unchanged.
</para>
</listitem>
-->
<para>
- Have <link
- linkend="textsearch-functions-table"><function>jsonb_to_tsvector()</function></link>
- properly check "string" parameter (Dominik Czarnota)
- </para>
- </listitem>
-
- <listitem>
-<!--
-Author: Tom Lane <tgl@sss.pgh.pa.us>
-2020-03-28 [9950c8aad] Fix lquery's behavior for consecutive '*' items.
--->
-
- <para>
- In <xref linkend="ltree"/>, when using adjacent asterisks with braces,
- e.g., ".*{2}.*{3}", properly interpret that as ".*{5}" (Nikita Glukhov)
+ Make <link
+ linkend="textsearch-functions-table"><function>json[b]_to_tsvector()</function></link>
+ fully check the spelling of its <literal>string</literal> option
+ (Dominik Czarnota)
</para>
</listitem>
</para>
<para>
- Previously, this value was adjusted before effecting the number of
- concurrent requests. This value is now used directly. Conversion of
+ Previously, this value was adjusted before setting the number of
+ concurrent requests. The value is now used directly. Conversion of
old values to new ones can be done using:
</para>
<programlisting>
-SELECT round(sum(OLD / n::float)) FROM generate_series(1, OLD) s(n);
+SELECT round(sum(<replaceable>OLDVALUE</replaceable> / n::float)) FROM generate_series(1, <replaceable>OLDVALUE</replaceable>) s(n);
</programlisting>
</listitem>
</para>
<para>
- Previously it returned <command>ALTER TABLE</command>, but now returns
+ Previously it returned <command>ALTER TABLE</command>; now it returns
<command>ALTER FOREIGN TABLE</command>.
</para>
</listitem>
</para>
<para>
- Previously it returned <command>ALTER TABLE</command>, but now returns
+ Previously it returned <command>ALTER TABLE</command>; now it returns
<command>ALTER MATERIALIZED VIEW</command>.
</para>
</listitem>
</para>
<para>
- This determines how much WAL to retain for the standby server,
- specified in megabytes rather than number of files. If you
+ This determines how much WAL to retain for standby servers.
+ It is specified in megabytes, rather than number of files as with
+ the old parameter. If you
previously used <varname>wal_keep_segments</varname>,
the following formula will give you an approximately equivalent setting:
<programlisting>
<listitem>
<!--
+Author: Tom Lane <tgl@sss.pgh.pa.us>
+2020-03-28 [9950c8aad] Fix lquery's behavior for consecutive '*' items.
+-->
+
+ <para>
+ In <xref linkend="ltree"/>, when an <type>lquery</type> pattern
+ contains adjacent asterisks with braces,
+ e.g., <literal>*{2}.*{3}</literal>, properly interpret that
+ as <literal>*{5}</literal> (Nikita Glukhov)
+ </para>
+ </listitem>
+
+ <listitem>
+<!--
Author: Peter Geoghegan <pg@bowt.ie>
2020-03-07 [691e8b2e1] pageinspect: Fix types used for bt_metap() columns.
-->
-->
<para>
- Improve cases where <link
- linkend="ddl-partition-pruning">pruning</link> of partitions can happen
+ Allow <link linkend="ddl-partition-pruning">pruning</link> of
+ partitions to happen in more cases
(Yuzuko Hosoya, Amit Langote, Álvaro Herrera)
</para>
</listitem>
-->
<para>
- Allow <literal>BEFORE</literal> row-level <link
+ Support row-level <literal>BEFORE</literal> <link
linkend="triggers">triggers</link> on partitioned tables (Álvaro
Herrera)
</para>
<para>
- These triggers cannot change which partition is the destination.
+ However, such a trigger is not allowed to change which partition
+ is the destination.
</para>
</listitem>
</para>
<para>
- Previously, partitions had to be replicated individually. Now
- partitioned tables can be published explicitly causing all partitions
- to be automatically published. Addition/removal of partitions from
- partitioned tables are automatically added/removed from publications.
+ Previously, partitions had to be replicated individually. Now a
+ partitioned table can be published explicitly, causing all its
+ partitions to be published automatically. Addition/removal of a
+ partition causes it to be likewise added to or removed from the
+ publication.
The <link linkend="sql-createpublication"><command>CREATE
PUBLICATION</command></link> option
<literal>publish_via_partition_root</literal> controls whether changes
- to partitions are published as their own or their ancestor's.
+ to partitions are published as their own changes or their parent's.
</para>
</listitem>
-->
<para>
- Allow <link linkend="rowtypes"><literal>ROW</literal> values</link>
- to be used as partitioning expressions (Amit Langote)
+ Allow whole-row variables (that
+ is, <replaceable>table</replaceable><literal>.*</literal>) to be
+ used in partitioning expressions (Amit Langote)
</para>
</listitem>
<para>
More efficiently store <link
- linkend="btree-deduplication">duplicates</link> in btree indexes
+ linkend="btree-deduplication">duplicates</link> in B-tree indexes
(Anastasia Lubennikova, Peter Geoghegan)
</para>
<para>
- This allows efficient btree indexing of low cardinality columns
+ This allows efficient B-tree indexing of low-cardinality columns
by storing duplicate keys only once. Users upgrading with <link
linkend="pgupgrade"><application>pg_upgrade</application></link>
will need to use <link
- linkend="sql-reindex"><command>REINDEX</command></link> to make use
- of this feature.
+ linkend="sql-reindex"><command>REINDEX</command></link> to make an
+ existing index use this feature.
</para>
</listitem>
<para>
Allow <link linkend="gist">GiST</link> and <link
- linkend="spgist">SP-GiST</link> indexes for <link
- linkend="datatype-line">box</link>/point distance lookups (Nikita
+ linkend="spgist">SP-GiST</link> indexes on <type>box</type>
+ columns to support <literal>ORDER BY <replaceable>box</replaceable>
+ <-> <replaceable>point</replaceable></literal> queries (Nikita
Glukhov)
</para>
</listitem>
<para>
Allow <link linkend="gin"><acronym>GIN</acronym></link> indexes to
- more efficiently handle <literal>NOT</literal> restrictions (Nikita
+ more efficiently handle <literal>!</literal> (NOT) clauses
+ in <type>tsquery</type> searches (Nikita
Glukhov, Alexander Korotkov, Tom Lane, Julien Rouhaud)
</para>
</listitem>
</para>
<para>
- The index and column collations must now match so the index's
- uniqueness matches the column's uniqueness.
+ The index's collation must match that of the underlying column,
+ but <command>ALTER TABLE</command> previously failed to check
+ this.
</para>
</listitem>
-->
<para>
- Allow setting <link linkend="sql-altertable">statistics target</link>
+ Allow setting the <link linkend="sql-altertable">statistics target</link>
for <link linkend="sql-createstatistics">extended statistics</link>
(Tomas Vondra)
</para>
-->
<para>
- Allow functions in <literal>FROM</literal> clauses to be moved to their
- reference sites if they evaluate to constants (Alexander Kuzmenkov,
+ Allow functions in <literal>FROM</literal> clauses to be pulled up
+ (inlined) if they evaluate to constants (Alexander Kuzmenkov,
Aleksandr Parfenov)
</para>
</listitem>
</para>
<para>
- If a result is already sorted by several leading keys, this
- allows for batch sorting of additional trailing keys because the
- previous keys are already equal. This is controlled by <xref
+ If an intermediate query result is known to be sorted by
+ one or more leading keys of a required sort ordering, the
+ additional sorting can be done considering only the remaining
+ keys, if the rows are sorted in batches that have equal leading
+ keys.
+ </para>
+
+ <para>
+ If necessary, this can be controlled using <xref
linkend="guc-enable-incremental-sort"/>.
</para>
</listitem>
<para>
Allow <link linkend="guc-enable-hashagg">hash aggregation</link>
- and <link linkend="queries-grouping-sets">grouping sets</link>
to use disk storage for large aggregation result sets (Jeff Davis)
</para>
</para>
<para>
- This behavior is normally preferable to the old behavior. But if
- it is inferior for a particular query, behavior similar to
- previous Postgres releases can be obtained by
+ This behavior is normally preferable to the old behavior, in which
+ once hash aggregation had been chosen, the hash table would be
+ kept in memory no matter how large it got — which could be
+ very large if the planner had misestimated. If necessary,
+ behavior similar to that can be obtained by
increasing <varname>hash_mem_multiplier</varname>.
</para>
</listitem>
-->
<para>
- Allow inserts to trigger <link linkend="autovacuum">autovacuum</link>
- activity (Laurenz Albe, Darafei Praliaskouski)
+ Allow inserts, not only updates and deletes, to
+ trigger <link linkend="autovacuum">autovacuum</link> activity
+ (Laurenz Albe, Darafei Praliaskouski)
</para>
<para>
- This new behavior reduces the work necessary when the table needs to
- be frozen and allows pages to be set as all-visible. All-visible pages
- allow index-only scans to access fewer heap rows.
+ This new behavior reduces the work that will be necessary later
+ when the table needs to be frozen, and it allows pages to be
+ marked as all-visible sooner. All-visible pages allow index-only
+ scans to access fewer heap rows.
</para>
</listitem>
<listitem>
<!--
-Author: Peter Eisentraut <peter@eisentraut.org>
-2020-03-28 [8f3ec75de] Enable Unix-domain sockets support on Windows
--->
-
- <para>
- Enable <link linkend="client-authentication">Unix-domain sockets</link>
- support on Windows (Peter Eisentraut)
- </para>
- </listitem>
-
- <listitem>
-<!--
Author: Fujii Masao <fujii@postgresql.org>
2019-11-21 [e6d806952] Make DROP DATABASE command generate less WAL records.
-->
<para>
- Improve the performance when replaying <link
+ Improve performance when replaying <link
linkend="sql-dropdatabase"><command>DROP DATABASE</command></link>
commands when many tablespaces are in use (Fujii Masao)
</para>
-->
<para>
- Improve speed of <link
- linkend="storage-toast"><acronym>TOAST</acronym></link> decompression
- and the retrieval of only the leading bytes of <acronym>TOAST</acronym>
- values (Binguo Bao, Andrey Borodin)
+ Improve retrieval of the leading bytes of <link
+ linkend="storage-toast"><acronym>TOAST</acronym></link>'ed values
+ (Binguo Bao, Andrey Borodin)
</para>
<para>
- Previously, <acronym>TOAST</acronym> values were fully fetched and
- only the requested leading bytes were decompressed and returned.
- Now, only the requested bytes are fetched.
+ Previously, compressed out-of-line <acronym>TOAST</acronym> values
+ were fully fetched even when it's known that only some leading
+ bytes are needed. Now, only enough data to produce the result is
+ fetched.
</para>
</listitem>
2019-09-10 [bca6e6435] Reduce overhead of scanning the backend[] array in LISTE
Author: Tom Lane <tgl@sss.pgh.pa.us>
2019-09-22 [51004c717] Make some efficiency improvements in LISTEN/NOTIFY.
--->
-
- <para>
- Improve performance of <link
- linkend="sql-listen"><command>LISTEN</command></link>/<command>NOTIFY</command>
- (Martijn van Oosterhout)
- </para>
- </listitem>
-
- <listitem>
-<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
2019-08-15 [bb5ae8f6c] Use a hash table to de-duplicate NOTIFY events faster.
-->
<para>
- Improve the efficiency of removing duplicate <link
- linkend="sql-notify"><command>NOTIFY</command></link> events (Tom Lane)
+ Improve performance of <link
+ linkend="sql-listen"><command>LISTEN</command></link>/<command>NOTIFY</command>
+ (Martijn van Oosterhout, Tom Lane)
</para>
</listitem>
-->
<para>
- Use lookup tables to speed up integer to text conversion (David Fetter)
+ Speed up conversions of integers to text (David Fetter)
</para>
</listitem>
-->
<para>
- Reduce memory usage for query strings that contain multiple
- <acronym>SQL</acronym> statements (Amit Langote)
+ Reduce memory usage for query strings and extension scripts that
+ contain many <acronym>SQL</acronym> statements (Amit Langote)
</para>
</listitem>
-->
<para>
- Allow a sample of statements to be logged (Adrien Nayrat)
+ Allow a sample of SQL statements, rather than all statements, to
+ be logged (Adrien Nayrat)
</para>
<para>
- A <xref linkend="guc-log-statement-sample-rate"/> ratio of statements
- taking over <xref linkend="guc-log-min-duration-sample"/> duration
- will be logged.
+ A <xref linkend="guc-log-statement-sample-rate"/> fraction of
+ those statements taking more than
+ <xref linkend="guc-log-min-duration-sample"/> duration will be
+ logged.
</para>
</listitem>
<para>
The <acronym>GUC</acronym> setting <xref
linkend="guc-log-parameter-max-length"/> controls the maximum length
- of parameter values output during statement non-error logging, and
- <xref linkend="guc-log-parameter-max-length-on-error"/> does the
- same for error statement logging. Previously, prepared statement
- parameters were not logged during errors.
- </para>
- </listitem>
-
- <listitem>
-<!--
-Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
-2020-02-05 [15d13e829] Make vacuum buffer counters 64 bits wide
--->
-
- <para>
- Make <link linkend="sql-vacuum">vacuum</link> buffer counters 64-bits
- wide to avoid overflow (Álvaro Herrera)
+ of parameter values output during logging of non-error statements,
+ while <xref linkend="guc-log-parameter-max-length-on-error"/> does
+ the same for logging of statements with errors. Previously,
+ prepared statement parameters were never logged during errors.
</para>
</listitem>
-->
<para>
- Allow function call backtraces of errors to be logged (Peter
+ Allow function call backtraces to be logged after errors (Peter
Eisentraut, Álvaro Herrera)
</para>
<para>
- Server variable <xref linkend="guc-backtrace-functions"/> specifies
+ The new parameter <xref linkend="guc-backtrace-functions"/> specifies
which C functions should generate backtraces on error.
</para>
</listitem>
<listitem>
<!--
-Author: Michael Paquier <michael@paquier.xyz>
-2019-07-05 [313f87a17] Add min() and max() aggregates for pg_lsn
+Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
+2020-02-05 [15d13e829] Make vacuum buffer counters 64 bits wide
-->
<para>
- Add <function>min()</function> and
- <function>max()</function> aggregates for <link
- linkend="datatype-pg-lsn"><type>pg_lsn</type></link> (Fabrízio
- de Royes Mello)
- </para>
-
- <para>
- This is useful for monitoring queries.
+ Make <link linkend="sql-vacuum">vacuum</link> buffer counters 64-bits
+ wide to avoid overflow (Álvaro Herrera)
</para>
</listitem>
<para>
Add system view <link
linkend="monitoring-stats-dynamic-views-table"><structname>pg_stat_progress_analyze</structname></link>
- to report <link linkend="sql-analyze">analyze</link> progress
+ to report <link linkend="sql-analyze">ANALYZE</link> progress
(Álvaro Herrera, Tatsuro Yamada, Vinayak Pokale)
</para>
</listitem>
-->
<para>
- Create <link
+ Add system view <link
linkend="monitoring-stats-views-table"><structname>pg_stat_slru</structname></link>
- system view to monitor internal <acronym>SLRU</acronym> caches
+ to monitor internal <acronym>SLRU</acronym> caches
(Tomas Vondra)
</para>
</listitem>
-->
<para>
- Allow <xref linkend="guc-track-activity-query-size"/> to be set up
- to 1MB (Vyacheslav Makarov)
+ Allow <xref linkend="guc-track-activity-query-size"/> to be set as
+ high as 1MB (Vyacheslav Makarov)
</para>
<para>
-->
<para>
- Add wait event for <function>posix_fallocate()</function> (Thomas
- Munro)
+ Report a wait event while creating a DSM segment
+ with <function>posix_fallocate()</function> (Thomas Munro)
</para>
</listitem>
-->
<para>
- Only allow superusers to view the <xref
+ Allow only superusers to view the <xref
linkend="guc-ssl-passphrase-command"/> setting (Insung Moon)
</para>
</para>
<para>
- This is controlled by <xref linkend="guc-ssl-min-protocol-version"/>.
+ This choice can be controlled by
+ <xref linkend="guc-ssl-min-protocol-version"/>.
</para>
</listitem>
-->
<para>
- Tighten rules on which utility commands are possible in <xref
- linkend="guc-default-transaction-read-only"/> mode (Robert Haas)
+ Tighten rules on which utility commands are allowed in
+ read-only transaction mode (Robert Haas)
</para>
<para>
- This also increases the number of utility commands that can run in
- parallel queries.
+ This change also increases the number of utility commands that can
+ run in parallel queries.
</para>
</listitem>
</para>
</listitem>
+ <listitem>
+<!--
+Author: Peter Eisentraut <peter@eisentraut.org>
+2020-03-28 [8f3ec75de] Enable Unix-domain sockets support on Windows
+-->
+
+ <para>
+ Enable <link linkend="client-authentication">Unix-domain sockets</link>
+ support on Windows (Peter Eisentraut)
+ </para>
+ </listitem>
+
</itemizedlist>
</sect4>
-->
<para>
- Allow <acronym>WAL</acronym> receivers use a temporary replication slot
- if a permanent one is not specified (Peter Eisentraut, Sergei Kornilov)
+ Allow <acronym>WAL</acronym> receivers to use a temporary
+ replication slot when a permanent one is not specified (Peter
+ Eisentraut, Sergei Kornilov)
</para>
<para>
-->
<para>
- Allow replication slot storage to be limited by <xref
- linkend="guc-max-slot-wal-keep-size"/> (Kyotaro Horiguchi)
+ Allow WAL storage for replication slots to be limited by
+ <xref linkend="guc-max-slot-wal-keep-size"/> (Kyotaro Horiguchi)
</para>
<para>
- Replication slots that exceed this value are invalidated.
+ Replication slots that would require exceeding this value are
+ marked invalid.
</para>
</listitem>
</para>
<para>
- Previously, the end of the <acronym>WAL</acronym> would promote the
- standby, even if the target was not reached.
+ Previously, a standby would promote itself upon reaching the end
+ of <acronym>WAL</acronym>, even if the target was not reached.
</para>
</listitem>
-->
<para>
- Have <link linkend="sql-createtable"><command>CREATE TABLE
- LIKE</command></link> propagate <literal>CHECK</literal>'s <literal>NO
- INHERIT</literal> property to created tables (Ildar Musin, Chris
- Travers)
+ Make <link linkend="sql-createtable"><command>CREATE TABLE
+ LIKE</command></link> propagate a <literal>CHECK</literal>
+ constraint's <literal>NO INHERIT</literal> property to the created
+ table (Ildar Musin, Chris Travers)
</para>
</listitem>
<para>
Add <link linkend="sql-altertable"><command>ALTER
- TABLE</command></link> clause <literal>DROP EXPRESSION</literal>
- to remove generated properties from columns (Peter Eisentraut)
+ TABLE ... DROP EXPRESSION</command></link>
+ to allow removing the <literal>GENERATED</literal>
+ property from a column (Peter Eisentraut)
</para>
</listitem>
-->
<para>
- Fix bugs in <command>ALTER TABLE</command> when later clauses overlap
- changes made by earlier clauses in the same command (Tom Lane)
+ Fix bugs in multi-step <command>ALTER TABLE</command> commands (Tom
+ Lane)
+ </para>
+
+ <para>
+ <literal>IF NOT EXISTS</literal> clauses now work as expected, in
+ that derived actions (such as index creation) do not execute if the
+ column already exists. Also, certain cases of combining related
+ actions into one <command>ALTER TABLE</command> now work when they
+ did not before.
</para>
</listitem>
</para>
<para>
- This was previously possible only using <command>ALTER TABLE RENAME
- COLUMN</command>.
+ Renaming view columns was already possible, but one had to
+ write <command>ALTER TABLE RENAME COLUMN</command>, which is
+ confusing.
</para>
</listitem>
<para>
Add <link linkend="sql-altertype"><command>ALTER TYPE</command></link>
- options useful for extensions, like <acronym>TOAST</acronym> and I/O
- functions control (Tomas Vondra, Tom Lane)
+ options to modify a base type's <acronym>TOAST</acronym> properties
+ and support functions (Tomas Vondra, Tom Lane)
</para>
</listitem>
</para>
<para>
- This combines existing options <literal>LC_COLLATE</literal> and
+ This combines the existing options <literal>LC_COLLATE</literal> and
<literal>LC_CTYPE</literal> into a single option.
</para>
</listitem>
<para>
Allow <link linkend="sql-dropdatabase"><command>DROP
- DATABASE</command></link> to disconnect users so drop succeeds (Pavel
- Stehule, Amit Kapila)
+ DATABASE</command></link> to disconnect sessions using the target
+ database, allowing the drop to succeed (Pavel Stehule, Amit Kapila)
</para>
<para>
- This is enabled by <literal>WITH FORCE</literal>.
+ This is enabled by the <literal>FORCE</literal> option.
</para>
</listitem>
-->
<para>
- Add C structure member <link
+ Add structure member <link
linkend="trigger-interface"><structfield>tg_updatedcols</structfield></link>
- to record updated columns to C triggers (Peter Eisentraut)
+ to allow C-language update triggers to know which column(s) were updated
+ (Peter Eisentraut)
</para>
</listitem>
<para>
The new data types are <link
- linkend="datatype-pseudotypes-table">anycompatible</link>,
- anycompatiblearray, anycompatiblenonarray, and anycompatiblerange.
+ linkend="datatype-pseudotypes-table"><type>anycompatible</type></link>,
+ <type>anycompatiblearray</type>, <type>anycompatiblenonarray</type>,
+ and <type>anycompatiblerange</type>.
</para>
</listitem>
<para>
Add <acronym>SQL</acronym> data type <link
- linkend="datatype-oid">xid8</link> to expose FullTransactionId
- (Thomas Munro)
+ linkend="datatype-oid"><type>xid8</type></link> to expose
+ FullTransactionId (Thomas Munro)
</para>
<para>
- The xid data type is only four bytes so does not show the transaction
- epoch.
+ The existing <type>xid</type> data type is only four bytes so it
+ does not provide the transaction epoch.
</para>
</listitem>
-->
<para>
- Add data type <link linkend="datatype-oid">regcollation</link> and
- helper functions for system collations (Julien Rouhaud)
+ Add data
+ type <link linkend="datatype-oid"><type>regcollation</type></link>
+ and associated functions, to represent OIDs of collation objects
+ (Julien Rouhaud)
</para>
</listitem>
-->
<para>
- Use the glibc version in some cases as the <link
- linkend="collation">collation</link> version (Thomas Munro)
+ Use the glibc version in some cases as a <link
+ linkend="collation">collation</link> version identifier (Thomas Munro)
</para>
<para>
- If the glibc version changes, a warning will be issued when a
- mismatching collation is used.
+ If the glibc version changes, a warning will be issued about
+ possible corruption of collation-dependent indexes.
</para>
</listitem>
-->
<para>
- Allow <link linkend="rowtypes"><literal>ROW</literal> values</link>
- values to have their members extracted with suffix notation (Tom Lane)
+ Allow <link linkend="sql-syntax-row-constructors"><literal>ROW</literal>
+ expressions</link> to have their members extracted with suffix
+ notation (Tom Lane)
</para>
<para>
- For example, <literal>(ROW(4, 5.0)).f1</literal> returns 4.
+ For example, <literal>(ROW(4, 5.0)).f1</literal> now returns 4.
</para>
</listitem>
<para>
Add alternate version of <link
- linkend="functions-json-processing-table"><function>jsonb_setI()</function></link>
- with special <literal>NULL</literal> handling (Andrew Dunstan)
+ linkend="functions-json-processing-table"><function>jsonb_set()</function></link>
+ with improved <literal>NULL</literal> handling (Andrew Dunstan)
</para>
<para>
- The new function, <function>jsonb_set_lax()</function>, allows null
- new values to either set the specified key to <acronym>JSON</acronym>
- null, delete the key, raise exception, or ignore the operation.
- IS 'return_target' CLEAR?
+ The new function, <function>jsonb_set_lax()</function>, handles a
+ <literal>NULL</literal> new value by either setting the specified
+ key to a <acronym>JSON</acronym> null, deleting the key, raising an
+ exception, or returning the <type>jsonb</type> value unmodified, as
+ requested.
</para>
</listitem>
</para>
<para>
- This allows json values to be converted to timestamps, which can
- then be processed in jsonpath expressions. This also adds jsonpath
- functions that support time zone-aware output.
+ This function allows <acronym>JSON</acronym> values to be converted
+ to timestamps, which can then be processed in <type>jsonpath</type>
+ expressions. This change also adds <type>jsonpath</type> functions
+ that support time-zone-aware output.
</para>
</listitem>
<listitem>
<!--
-Author: Tom Lane <tgl@sss.pgh.pa.us>
-2020-03-06 [a6525588b] Allow Unicode escapes in any server encoding, not only U
+Author: Michael Paquier <michael@paquier.xyz>
+2019-07-05 [313f87a17] Add min() and max() aggregates for pg_lsn
-->
<para>
- Allow <link linkend="sql-syntax-strings-uescape">Unicode
- escapes</link>, e.g., E'\u####', U&'\####', to specify any
- character available in the database encoding, even when the database
- encoding is not <acronym>UTF</acronym>-8 (Tom Lane)
+ Add <function>min()</function> and
+ <function>max()</function> aggregates for <link
+ linkend="datatype-pg-lsn"><type>pg_lsn</type></link> (Fabrízio
+ de Royes Mello)
</para>
<para>
- The Unicode characters must be available in the database encoding.
+ These are particularly useful in monitoring queries.
+ </para>
+ </listitem>
+
+ <listitem>
+<!--
+Author: Tom Lane <tgl@sss.pgh.pa.us>
+2020-03-06 [a6525588b] Allow Unicode escapes in any server encoding, not only U
+-->
+
+ <para>
+ Allow <link linkend="sql-syntax-strings-uescape">Unicode
+ escapes</link>,
+ e.g., <literal>E'\u<replaceable>nnnn</replaceable>'</literal>
+ or <literal>U&'\<replaceable>nnnn</replaceable>'</literal>, to
+ specify any character available in the database encoding, even when
+ the database encoding is not <acronym>UTF</acronym>-8 (Tom Lane)
</para>
</listitem>
<para>
The names recognized are the same as those output by <link
linkend="functions-formatting"><function>to_char()</function></link>
- with the same format codes.
+ with the same format patterns.
</para>
</listitem>
-->
<para>
- Add format specifications FF1-FF6 to control display of 1-6 subsecond
- digits (Alexander Korotkov, Nikita Glukhov, Teodor Sigaev, Oleg
- Bartunov)
+ Add datetime format patterns <literal>FF1</literal>
+ – <literal>FF6</literal> to specify input or output of
+ 1 to 6 fractional-second digits (Alexander
+ Korotkov, Nikita Glukhov, Teodor Sigaev, Oleg Bartunov)
</para>
<para>
-->
<para>
- Add <literal>SSSSS</literal> time format specification as an
- <acronym>SQL</acronym> standard alias for <literal>SSSS</literal>
+ Add <literal>SSSSS</literal> datetime format pattern as an
+ <acronym>SQL</acronym>-standard alias for <literal>SSSS</literal>
(Nikita Glukhov, Alexander Korotkov)
</para>
</listitem>
<para>
Add function <link
linkend="functions-uuid"><function>gen_random_uuid()</function></link>
- to generate version 4 UUIDs (Peter Eisentraut)
+ to generate version-4 UUIDs (Peter Eisentraut)
</para>
<para>
Previously <acronym>UUID</acronym> generation functions were only
- available via external modules <xref linkend="uuid-ossp"/> and <xref
+ available in the external modules <xref linkend="uuid-ossp"/> and <xref
linkend="pgcrypto"/>.
</para>
</listitem>
-->
<para>
- Add greatest-common-denominator (<link
- linkend="functions-math-func-table">gcd</link>) and
- least-common-multiple (lcm) functions (Vik Fearing)
+ Add greatest-common-denominator
+ (<link linkend="functions-math-func-table"><function>gcd</function></link>)
+ and least-common-multiple (<function>lcm</function>) functions (Vik
+ Fearing)
</para>
</listitem>
-->
<para>
- Improve the performance and accuracy of <link
- linkend="functions-math-func-table">square root</link> and natural log
- (ln) output (Dean Rasheed)
+ Improve the performance and accuracy of the <type>numeric</type>
+ type's <link linkend="functions-math-func-table">square root</link>
+ (<function>sqrt</function>) and natural log
+ (<function>ln</function>) functions (Dean Rasheed)
</para>
</listitem>
<para>
Add function <link
linkend="functions-math-func-table"><function>min_scale()</function></link>
- that returns the number of digits to the right the decimal point
- that is required to represent the numeric value with full precision
- (Pavel Stehule)
+ that returns the number of digits to the right of the decimal point
+ that are required to represent a <type>numeric</type> value with
+ full accuracy (Pavel Stehule)
</para>
</listitem>
<para>
Add function <link
linkend="functions-math-func-table"><function>trim_scale()</function></link>
- to reduce the scale of a number by removing trailing zeros (Pavel
- Stehule)
+ to reduce the scale of a <type>numeric</type> value by removing
+ trailing zeros (Pavel Stehule)
</para>
</listitem>
</para>
<para>
- For example, previously only point <-> line was supported,
- now line <-> point works too.
+ For example, previously only <type>point</type>
+ <literal><-></literal> <type>line</type> was supported,
+ now <type>line</type> <literal><-></literal> <type>point</type>
+ works too.
</para>
</listitem>
-->
<para>
- Create xid8 versions of all <link
- linkend="functions-pg-snapshot">transaction id functions</link>
+ Create <type>xid8</type> versions of all <link
+ linkend="functions-pg-snapshot">transaction ID functions</link>
(Thomas Munro)
</para>
<para>
- The old function names were kept for backward compatibility. DO WE
- HAVE NEW NAMES?
+ The old <type>xid</type>-based functions still exist, for backward
+ compatibility.
</para>
</listitem>
<para>
Allow <link
linkend="functions-binarystring-other"><function>get_bit()</function></link>
- and <function>set_bit()</function> to set bits beyond 256MB of bytea
- data (Movead Li)
+ and <function>set_bit()</function> to set bits beyond the first
+ 256MB of a <type>bytea</type> value (Movead Li)
</para>
</listitem>
</para>
</listitem>
-
</itemizedlist>
</sect3>
-->
<para>
- Improve the performance of PL/pgSQL functions that use immutable
+ Improve performance of PL/pgSQL functions that use immutable
expressions (Konstantin Knizhnik)
</para>
</listitem>
-->
<para>
- Add libpq connection parameters to control the min/max
- <acronym>TLS</acronym> version for encrypted connections
+ Add libpq connection parameters to control the minimum and maximum
+ <acronym>TLS</acronym> version allowed for an encrypted connection
(Daniel Gustafsson)
</para>
<listitem>
<!--
-Author: Fujii Masao <fujii@postgresql.org>
-2020-03-05 [2eb3bc588] Fix issues around .pgpass file.
--->
-
- <para>
- Tighten libpq's overlength-line handling and comment detection for <link
- linkend="libpq-pgpass">.pgpass</link> files (Fujii Masao)
- </para>
- </listitem>
-
- <listitem>
-<!--
Author: Andrew Dunstan <andrew@dunslane.net>
2019-11-30 [4dc635521] libq support for sslpassword connection param, DER form
-->
<para>
- Allow specification of passwords to unlock client certificates (Craig
+ Allow use of passwords to unlock client certificates (Craig
Ringer, Andrew Dunstan)
</para>
<para>
- This is specified by the <xref linkend="libpq-connect-sslpassword"/>
- connection option.
+ This is enabled by libpq's <xref linkend="libpq-connect-sslpassword"/>
+ connection parameter.
</para>
</listitem>
-->
<para>
- Allow <acronym>DER</acronym>-encoded client certificates (Craig Ringer,
- Andrew Dunstan)
+ Allow libpq to use <acronym>DER</acronym>-encoded client
+ certificates (Craig Ringer, Andrew Dunstan)
</para>
</listitem>
-->
<para>
- Add the transaction status (%x) to the default <link
- linkend="app-psql-prompting"><application>psql</application>
- prompts</link> (Vik Fearing)
+ Add transaction status (<literal>%x</literal>)
+ to <application>psql</application>'s default
+ <link linkend="app-psql-prompting">prompts</link> (Vik Fearing)
</para>
</listitem>
-->
<para>
- Allow the secondary <application>psql</application> prompt to be same
- number of spaces as the primary prompt (Thomas Munro)
+ Allow the secondary <application>psql</application> prompt to be
+ blank but the same width as the primary prompt (Thomas Munro)
</para>
<para>
- This is accomplished by setting PROMPT2 to %w.
+ This is accomplished by setting <literal>PROMPT2</literal>
+ to <literal>%w</literal>.
</para>
</listitem>
-->
<para>
- Allow <link linkend="app-psql-meta-commands">\g</link> and \gx to
- change any <link linkend="app-psql-meta-commands">\pset</link> output
- options for a single command (Tom Lane)
+ Allow <application>psql</application>'s
+ <link linkend="app-psql-meta-commands"><literal>\g</literal></link>
+ and <literal>\gx</literal> commands to
+ change <link linkend="app-psql-meta-commands">\pset</link> output
+ options for the duration of that single command (Tom Lane)
</para>
<para>
- This allows syntax like \g (expand=on), which is equivalent to \gx.
+ This feature allows syntax like <literal>\g (expand=on)</literal>,
+ which is equivalent to <literal>\gx</literal>.
</para>
</listitem>
-->
<para>
- Add <application>psql</application> commands to report operator
+ Add <application>psql</application> commands to display operator
classes and operator families (Sergey Cherkashin, Nikita Glukhov,
Alexander Korotkov)
</para>
<para>
The new commands are <link
- linkend="app-psql-meta-commands">\dAc</link>, \dAf, \dAo, and \dAp.
+ linkend="app-psql-meta-commands"><literal>\dAc</literal></link>,
+ <literal>\dAf</literal>, <literal>\dAo</literal>,
+ and <literal>\dAp</literal>.
</para>
</listitem>
<para>
Show table persistence in <application>psql</application>'s <link
- linkend="app-psql-meta-commands">\dt+</link> and related commands
- (David Fetter)
+ linkend="app-psql-meta-commands"><literal>\dt+</literal></link>
+ and related commands (David Fetter)
</para>
<para>
-->
<para>
- Improve output of <application>psql</application> <link
- linkend="app-psql-meta-commands">\d</link> for <acronym>TOAST</acronym>
- tables (Justin Pryzby)
+ Improve output of <application>psql</application>'s <link
+ linkend="app-psql-meta-commands"><literal>\d</literal></link>
+ for <acronym>TOAST</acronym> tables (Justin Pryzby)
</para>
</listitem>
-->
<para>
- Adjust display of <application>psql</application>'s <link
- linkend="app-psql-meta-commands">\e</link> query (Tom Lane)
+ Fix redisplay after <application>psql</application>'s <link
+ linkend="app-psql-meta-commands"><literal>\e</literal></link>
+ command (Tom Lane)
</para>
<para>
When exiting the editor, if the query doesn't end with a semicolon
- or \g, the query buffer contents will now be displayed.
+ or <literal>\g</literal>, the query buffer contents will now be
+ displayed.
</para>
</listitem>
-->
<para>
- Add <link linkend="app-psql-meta-commands">\warn</link> command to
- <application>psql</application> (David Fetter)
+ Add <link
+ linkend="app-psql-meta-commands"><literal>\warn</literal></link>
+ command to <application>psql</application> (David Fetter)
</para>
<para>
- This is like \echo except that the text is sent to stderr instead
- of stdout.
+ This is like <literal>\echo</literal> except that the text is sent
+ to stderr instead of stdout.
</para>
</listitem>
-->
<para>
- Allow pgbench to partition its 'accounts' table (Fabien Coelho)
+ Allow pgbench to partition its <quote>accounts</quote> table
+ (Fabien Coelho)
</para>
<para>
-->
<para>
- Add pgbench command \aset, which behaves like \gset, but for multiple
- queries (Fabien Coelho)
+ Add pgbench command <literal>\aset</literal>, which behaves
+ like <literal>\gset</literal>, but for multiple queries (Fabien
+ Coelho)
</para>
</listitem>
-->
<para>
- Allow pgbench to generate its data server-side, rather than client side
- (Fabien Coelho)
+ Allow pgbench to generate its initial data server-side, rather
+ than client-side (Fabien Coelho)
</para>
</listitem>
-->
<para>
- Allow pgbench to dump script contents using
+ Allow pgbench to show script contents using option
<option>--show-script</option> (Fabien Coelho)
</para>
</listitem>
<para>
This computation allows <link
linkend="monitoring-stats-dynamic-views-table"><structname>pg_stat_progress_basebackup</structname></link>
- to show progress, and can be disabled by using the
- <option>--no-estimate-size</option> option. Previously, this
- computation happened only if <option>--progress</option> was used.
+ to show progress. If that is not needed, it can be disabled by
+ using the <option>--no-estimate-size</option> option. Previously,
+ this computation happened only if the <option>--progress</option>
+ option was used.
</para>
</listitem>
-->
<para>
- Add <link
+ Add an option to <link
linkend="app-pgrewind"><application>pg_rewind</application></link>
- option to configure standbys (Paul Guo, Jimmy Yih, Ashwin Agrawal)
+ to configure standbys (Paul Guo, Jimmy Yih, Ashwin Agrawal)
</para>
<para>
-->
<para>
- Increase <link linkend="sql-prepare-transaction"><command>PREPARE
- TRANSACTION</command></link> information reported by
+ Increase the <link linkend="sql-prepare-transaction"><command>PREPARE
+ TRANSACTION</command></link>-related information reported by
<application>pg_waldump</application> (Fujii Masao)
</para>
</listitem>
-->
<para>
- Allow <link linkend="app-dropdb">dropdb</link> to force disconnections
- so the drop succeeds (Pavel Stehule)
+ Allow <link linkend="app-dropdb">dropdb</link> to disconnect
+ sessions using the target database, allowing the drop to succeed
+ (Pavel Stehule)
</para>
<para>
</para>
<para>
- The long-supported options for this are called
+ The long-supported preferred options for this are called
<option>--superuser</option> and <option>--no-superuser</option>.
</para>
</listitem>
<para>
Use the directory of the <link
linkend="pgupgrade"><application>pg_upgrade</application></link>
- binary as the default new 'bindir' location when running
- <application>pg_upgrade</application> (Daniel Gustafsson)
+ program as the default <option>--new-bindir</option> setting when
+ running <application>pg_upgrade</application> (Daniel Gustafsson)
</para>
</listitem>
<para>
Reformat tables containing <link linkend="functions">function
- information</link> for better clarity (Tom Lane)
+ and operator information</link> for better clarity (Tom Lane)
</para>
</listitem>
<!--
Author: Peter Eisentraut <peter@eisentraut.org>
2019-11-27 [4513d8b07] Move configure - -disable-float8-byval to pg_config_manua
--->
-
- <para>
- Remove <link linkend="install-short">support</link> option
- <option>--disable-float8-byval</option> (Peter Eisentraut)
- </para>
-
- <para>
- This was needed for previously supported version-zero functions.
- </para>
- </listitem>
-
- <listitem>
-<!--
Author: Peter Eisentraut <peter@eisentraut.org>
2019-11-21 [2e4db241b] Remove configure - -disable-float4-byval
-->
<para>
- Remove configure option <option>--disable-float4-byval</option>
- (Peter Eisentraut)
+ Remove <link linkend="install-short">configure</link> options
+ <option>--disable-float8-byval</option>
+ and <option>--disable-float4-byval</option> (Peter Eisentraut)
</para>
<para>
- This was needed for previously supported version-zero functions.
+ These were needed for compatibility with some version-zero C
+ functions, but those are no longer supported.
</para>
</listitem>
-->
<para>
- Add the query string to planner hook functions (Pascal Legrand,
+ Pass the query string to planner hook functions (Pascal Legrand,
Julien Rouhaud)
</para>
</listitem>
-->
<para>
- Use operating system functions to cleanly erase memory that contains
+ Use operating system functions to reliably erase memory that contains
sensitive information (Peter Eisentraut)
</para>
-->
<para>
- Add "headerscheck" script to test C header-file compatibility
- (Tom Lane)
+ Add <filename>headerscheck</filename> script to test C header-file
+ compatibility (Tom Lane)
</para>
</listitem>
-->
<para>
- Implement internal lists as arrays, rather than a chain of structures
+ Implement internal lists as arrays, rather than a chain of cells
(Tom Lane)
</para>
<para>
- This improves performance for queries that access many object.
- The internal List <acronym>API</acronym> has also been improved.
- </para>
- </listitem>
-
- <listitem>
-<!--
-Author: Michael Paquier <michael@paquier.xyz>
-2019-07-13 [39aadc984] Fix some inconsistencies in MSVC scripts
--->
-
- <para>
- Update <link linkend="install-windows">Windows build scripts</link>
- to use the modern <option>--with-uuid</option> flag for
- <acronym>UUID</acronym> libraries (Kyotaro Horiguchi)
+ This improves performance for queries that access many objects.
</para>
</listitem>
</para>
<para>
- Such extensions can be installed in a database by users with creation
- rights, even if they are not superusers. This change also removes
+ Such extensions can be installed in a database by users with
+ database-level <literal>CREATE</literal> privileges, even if they
+ are not superusers. This change also removes
the <structname>pg_pltemplate</structname> system catalog.
</para>
</listitem>
-->
<para>
- Remove support for upgrading "unpackaged" extensions (Tom Lane)
+ Remove support for upgrading unpackaged (pre-9.1) extensions (Tom Lane)
</para>
</listitem>
</para>
<para>
- Specifically, allow <link linkend="sql-alterusermapping"><command>ALTER
- USER MAPPING</command></link> to set
- <literal>password_required</literal> to false. Care must still be
- taken to avoid non-superusers from using superuser credentials to
- connect to the foreign server.
+ Specifically, allow a superuser to set
+ <literal>password_required</literal> to false for
+ a <link linkend="sql-alterusermapping">user mapping</link>. Care
+ must still be taken to prevent non-superusers from using superuser
+ credentials to connect to the foreign server.
</para>
</listitem>
<para>
Have <xref linkend="pgstatstatements"/> treat <command>SELECT
- ... FOR UPDATE</command> as distinct from those without <command>FOR
- UPDATE</command> (Andrew Gierth, Vik Fearing)
+ ... FOR UPDATE</command> commands as distinct from those
+ without <command>FOR UPDATE</command> (Andrew Gierth, Vik Fearing)
</para>
</listitem>
-->
<para>
- Add option to <xref linkend="dict-int"/> extension to ignore the sign
+ Add an option to <xref linkend="dict-int"/> to ignore the sign
of integers (Jeff Janes)
</para>
</listitem>
-->
<para>
- Add btree index de-duplication processing columns to pageinspect output
+ Add B-tree index de-duplication processing columns to pageinspect output
(Peter Geoghegan)
</para>
</listitem>