rdelaage
/
postgresql
mirror of https://git.postgresql.org/git/postgresql.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Release notes for 12.1, 11.6, 10.11, 9.6.16, 9.5.20, 9.4.25.

This commit is contained in:
Tom Lane 2019-11-10 18:31:13 -05:00
parent f967563045
commit 3b0d044469
1 changed files with 35 additions and 785 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

820
doc/src/sgml/release-12.sgml
View File

@ -46,56 +46,6 @@ Branch: REL_12_STABLE [3b25de620] 2019-10-16 02:38:13 -0700
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [736b84eed] 2019-09-25 10:07:23 +0900
Branch: REL_12_STABLE Release: REL_12_0 [707f38e38] 2019-09-25 10:08:26 +0900
Branch: REL_11_STABLE [d01d4f237] 2019-09-25 10:08:30 +0900
Branch: REL_10_STABLE [d48168003] 2019-09-25 10:08:36 +0900
Branch: REL9_6_STABLE [98b5c3785] 2019-09-25 10:08:43 +0900
-->
<para>
Fix failure of <command>ALTER TABLE SET</command> with a custom
relation option (Michael Paquier)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [4d4c66add] 2019-08-18 17:11:57 -0400
Branch: REL_12_STABLE Release: REL_12_0 [328c3f6f9] 2019-08-18 17:11:57 -0400
Branch: REL_11_STABLE [909efc449] 2019-08-18 17:11:58 -0400
Branch: REL_10_STABLE [451432214] 2019-08-18 17:11:58 -0400
Branch: REL9_6_STABLE [3442235f2] 2019-08-18 17:11:58 -0400
Branch: REL9_5_STABLE [c511d367a] 2019-08-18 17:11:58 -0400
-->
<para>
Disallow changing a multiply-inherited column's type if not all
parent tables were changed (Tom Lane)
</para>
<para>
Previously, this was allowed, whereupon queries on the
now-out-of-sync parent would fail.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [cabe0f298] 2019-08-10 11:30:11 -0400
Branch: REL_12_STABLE Release: REL_12_0 [6ce0366be] 2019-08-10 11:30:11 -0400
Branch: REL_11_STABLE [ceb850d4a] 2019-08-10 11:30:12 -0400
-->
<para>
Avoid failure if the same target table is specified twice in
an <command>ANALYZE</command> command inside a transaction block
(Tom Lane)
</para>
</listitem>
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [5d3500da7] 2019-10-23 15:04:48 +0900
Branch: REL_12_STABLE [7668d4847] 2019-10-23 15:05:09 +0900
Author: Michael Paquier <michael@paquier.xyz>
@ -268,27 +218,6 @@ Branch: REL_12_STABLE [a9db37a18] 2019-11-05 11:42:25 -0500
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [03c811a48] 2019-08-12 13:15:47 -0400
Branch: REL_12_STABLE Release: REL_12_0 [c0c12ce39] 2019-08-12 13:15:47 -0400
Branch: REL_11_STABLE [c914e74d2] 2019-08-12 13:15:48 -0400
Branch: REL_10_STABLE [886cf85b5] 2019-08-12 13:15:48 -0400
-->
<para>
Fix planner's test for case-foldable characters
in <literal>ILIKE</literal> with an ICU collation (Tom Lane)
</para>
<para>
This mistake caused the planner to treat too much of the pattern as
being a fixed prefix, so that indexscans derived from
an <literal>ILIKE</literal> clause might miss entries that they
should find.
</para>
</listitem>
<listitem>
<!--
Author: Andrew Gierth <rhodiumtoad@postgresql.org>
Branch: master [b7a1c5539] 2019-10-03 10:54:52 +0100
Branch: REL_12_STABLE [0b11dc019] 2019-10-03 11:12:39 +0100
@ -314,29 +243,6 @@ Branch: REL9_4_STABLE [3473f81dd] 2019-10-03 11:18:20 +0100
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [7f1f72c44] 2019-09-12 18:29:45 -0400
Branch: REL_12_STABLE Release: REL_12_0 [5e9b18c78] 2019-09-12 18:29:17 -0400
Branch: REL_11_STABLE [64d926f2b] 2019-09-12 18:29:48 -0400
Branch: REL_10_STABLE [b54cff2bf] 2019-09-12 18:29:49 -0400
Branch: REL9_6_STABLE [b00132b9a] 2019-09-12 18:29:18 -0400
Branch: REL9_5_STABLE [aee5736f1] 2019-09-12 18:29:18 -0400
Branch: REL9_4_STABLE [ca08ea52b] 2019-09-12 18:29:18 -0400
-->
<para>
Fix handling of whole-row variables in <literal>WITH CHECK
OPTION</literal> expressions and row-level-security policy expressions
(Andres Freund)
</para>
<para>
Previously, such usage might result in bogus errors about row type
mismatches.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [3887e9455] 2019-10-07 12:39:09 -0400
Branch: REL_12_STABLE [7e8d0eb63] 2019-10-07 12:39:09 -0400
Branch: REL_11_STABLE [021065aac] 2019-10-07 12:39:09 -0400
@ -366,38 +272,6 @@ Branch: REL_12_STABLE [60e97d63e] 2019-10-04 13:57:37 -0700
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [1ced082b9] 2019-08-15 20:04:19 -0400
Branch: REL_12_STABLE Release: REL_12_0 [03813a50e] 2019-08-15 20:04:19 -0400
Branch: REL_11_STABLE [aed967d69] 2019-08-15 20:04:19 -0400
Branch: REL_10_STABLE [60886965a] 2019-08-15 20:04:19 -0400
Branch: REL9_6_STABLE [1fe8d209e] 2019-08-15 20:04:19 -0400
Branch: REL9_5_STABLE [cb0c79ae6] 2019-08-15 20:04:19 -0400
Branch: REL9_4_STABLE [afa71d915] 2019-08-15 20:04:19 -0400
-->
<para>
Prevent possible double-free if a <literal>BEFORE UPDATE</literal>
trigger returns the old tuple as-is, and it is not the last such
trigger (Thomas Munro)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [0662eb621] 2019-08-09 13:20:28 -0400
Branch: REL_12_STABLE Release: REL_12_0 [59d378929] 2019-08-09 13:20:28 -0400
Branch: REL_11_STABLE [2f729d832] 2019-08-09 13:20:28 -0400
-->
<para>
Fix crash if <literal><replaceable>x</replaceable> = ANY
(<replaceable>array</replaceable>)</literal>, or related operations,
contains a constant-null array (Tom Lane)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [a30531c5c] 2019-11-05 13:40:37 -0500
Branch: REL_12_STABLE [791864193] 2019-11-05 13:40:37 -0500
Branch: REL_11_STABLE [2bfe015b5] 2019-11-05 13:40:37 -0500
@ -431,30 +305,6 @@ Branch: REL9_5_STABLE [cdbb39213] 2019-10-17 13:50:59 +1300
<listitem>
<!--
Author: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Branch: master [1169fcf12] 2019-08-07 12:40:49 +0300
Branch: REL_12_STABLE Release: REL_12_0 [f8d30182b] 2019-08-07 12:41:00 +0300
Branch: REL_11_STABLE [c5b796125] 2019-08-07 12:41:16 +0300
Branch: REL_10_STABLE [65468cc70] 2019-08-07 12:41:22 +0300
Branch: REL9_6_STABLE [75774cc22] 2019-08-07 12:41:26 +0300
Branch: REL9_5_STABLE [fd298cd63] 2019-08-07 12:41:31 +0300
Branch: REL9_4_STABLE [54c98fa71] 2019-08-07 12:41:37 +0300
-->
<para>
In serializable mode, ensure that row-level predicate locks are
acquired on the correct version of the row (Thomas Munro, Heikki
Linnakangas)
</para>
<para>
If the visible version of the row is HOT-updated, the lock might be
taken on its now-dead predecessor, resulting in subtle failures to
guarantee serialization.
</para>
</listitem>
<listitem>
<!--
Author: Andres Freund <andres@anarazel.de>
Branch: master [a586cc4b6] 2019-10-04 13:34:28 -0700
Branch: REL_12_STABLE [c025165da] 2019-10-04 13:34:39 -0700
@ -474,8 +324,8 @@ Branch: REL9_4_STABLE [59800f7ce] 2019-10-09 13:31:38 +0900
<para>
Some code paths tried to do this after opening a file read-only,
which works on many platforms, but others give <quote>bad file
descriptor</quote> or related errors.
but on some platforms that causes <quote>bad file descriptor</quote>
or similar errors.
</para>
</listitem>
@ -503,26 +353,6 @@ Branch: REL9_4_STABLE [6899be289] 2019-10-03 17:34:26 -0400
<listitem>
<!--
Author: Thomas Munro <tmunro@postgresql.org>
Branch: master [720b59b55] 2019-08-28 16:18:29 +1200
Branch: REL_12_STABLE Release: REL_12_0 [8cc6016a8] 2019-08-28 16:18:39 +1200
Branch: REL_11_STABLE [b9c4ccfef] 2019-08-28 17:59:27 +1200
Branch: REL_10_STABLE [eeb68c1cd] 2019-08-28 17:59:50 +1200
-->
<para>
Avoid an unnecessary catalog lookup during heap page pruning
(Thomas Munro)
</para>
<para>
It's no longer necessary to check for unlogged indexes here, and the
check caused significant performance problems in some workloads.
There's also at least a theoretical possibility of deadlock.
</para>
</listitem>
<listitem>
<!--
Author: Andrew Gierth <rhodiumtoad@postgresql.org>
Branch: master [a9056cc63] 2019-11-06 04:13:30 +0000
Branch: REL_12_STABLE [f57c63107] 2019-11-06 04:33:35 +0000
@ -543,19 +373,6 @@ Branch: REL9_6_STABLE [747aac88f] 2019-11-06 04:33:55 +0000
<listitem>
<!--
Author: Andres Freund <andres@anarazel.de>
Branch: master [ac88807f9] 2019-09-29 16:24:32 -0700
Branch: REL_12_STABLE Release: REL_12_0 [2ce9bca36] 2019-09-29 16:27:18 -0700
Branch: REL_11_STABLE [51ad5b9bd] 2019-09-29 16:27:08 -0700
-->
<para>
Fix failure to JIT-compile equality comparisons for grouping hash
tables, leading to performance loss (Andres Freund)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [c477f3e44] 2019-10-03 13:56:26 -0400
Branch: REL_12_STABLE [9a407209a] 2019-10-03 13:56:26 -0400
@ -636,6 +453,31 @@ Branch: REL_12_STABLE [03666dfa1] 2019-10-18 22:25:42 +0900
<listitem>
<!--
Author: Peter Eisentraut <peter@eisentraut.org>
Branch: master [1c60e40ad] 2019-11-09 08:35:44 +0100
Branch: REL_12_STABLE [d891d2c89] 2019-11-09 08:35:51 +0100
Branch: REL_11_STABLE [7961da188] 2019-11-09 08:36:11 +0100
Branch: REL_10_STABLE [2e00d5976] 2019-11-09 09:15:35 +0100
Branch: master [ef8fcbff5] 2019-11-09 13:19:27 +0100
Branch: REL_12_STABLE [f96756304] 2019-11-09 16:00:34 +0100
Branch: REL_11_STABLE [aa9884578] 2019-11-09 16:01:23 +0100
Branch: REL_10_STABLE [806f9dc02] 2019-11-09 16:03:00 +0100
-->
<para>
Fix logical replication failure when publisher and subscriber have
different ideas about a table's replica identity columns
(Jehan-Guillaume de Rorthais, Peter Eisentraut)
</para>
<para>
Declaring a column as part of the replica identity on the
subscriber, when it does not exist at all on the publisher, led
to <quote>negative bitmapset member not allowed</quote> errors.
</para>
</listitem>
<listitem>
<!--
Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
Branch: master [38ddeab13] 2019-10-17 15:06:06 +0200
Branch: REL_12_STABLE [1391c13ce] 2019-10-17 15:06:06 +0200
@ -695,28 +537,6 @@ Branch: REL9_4_STABLE [15d90a02a] 2019-11-06 16:12:56 +0900
<listitem>
<!--
Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
Branch: master [bac2fae05] 2019-09-13 16:36:28 -0300
Branch: REL_12_STABLE Release: REL_12_0 [96b5033e1] 2019-09-13 16:36:28 -0300
Branch: REL_11_STABLE [41f3d2626] 2019-09-13 16:36:28 -0300
Branch: REL_10_STABLE [4f7dbf0ef] 2019-09-13 16:36:28 -0300
Branch: REL9_6_STABLE [ae4305f6d] 2019-09-13 16:36:28 -0300
Branch: REL9_5_STABLE [7110f5c37] 2019-09-13 16:36:28 -0300
Branch: REL9_4_STABLE [e8c7f40a1] 2019-09-13 16:36:28 -0300
-->
<para>
In logical decoding, ensure that sub-transactions are correctly
accounted for when reconstructing a snapshot (Masahiko Sawada)
</para>
<para>
This error leads to assertion failures; it's unclear whether any
bad effects exist in production builds.
</para>
</listitem>
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [20345197f] 2019-11-01 22:38:32 +0900
Branch: REL_12_STABLE [7b8c2de64] 2019-11-01 22:38:45 +0900
@ -735,47 +555,6 @@ Branch: REL9_4_STABLE [f88f7206e] 2019-11-01 22:39:09 +0900
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [f1bf619ac] 2019-08-14 15:09:42 -0400
Branch: REL_12_STABLE Release: REL_12_0 [75b2f011f] 2019-08-14 15:09:20 -0400
Branch: REL_11_STABLE [32d38f54a] 2019-08-14 15:09:20 -0400
Branch: REL_10_STABLE [f8c9a0852] 2019-08-14 15:09:20 -0400
Branch: REL9_6_STABLE [4784ad7a3] 2019-08-14 15:09:20 -0400
Branch: REL9_5_STABLE [29f9b1819] 2019-08-14 15:09:20 -0400
Branch: REL9_4_STABLE [a4b0d955b] 2019-08-14 15:09:20 -0400
-->
<para>
Fix <command>ALTER SYSTEM</command> to cope with duplicate entries
in <filename>postgresql.auto.conf</filename> (Ian Barwick)
</para>
<para>
<command>ALTER SYSTEM</command> itself will not generate such a state,
but external tools that modify <filename>postgresql.auto.conf</filename>
could do so. Duplicate entries for the target variable will now be
removed, and then the new setting (if any) will be appended at the end.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [6e4213056] 2019-08-27 14:44:26 -0400
Branch: REL_12_STABLE Release: REL_12_0 [1510339dc] 2019-08-27 14:44:26 -0400
Branch: REL_11_STABLE [cf86803e8] 2019-08-27 14:44:26 -0400
Branch: REL_10_STABLE [771e12701] 2019-08-27 14:44:26 -0400
Branch: REL9_6_STABLE [465f4ddda] 2019-08-27 14:44:26 -0400
Branch: REL9_5_STABLE [ef47d284d] 2019-08-27 14:44:26 -0400
-->
<para>
Reject include directives with empty file names in configuration
files, and report include-file recursion more clearly
(Ian Barwick, Tom Lane)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [3affe76ef] 2019-11-05 14:27:37 -0500
Branch: REL_12_STABLE [f9bd3b6d9] 2019-11-05 14:27:37 -0500
Branch: REL_11_STABLE [97ddc47b9] 2019-11-05 14:27:37 -0500
@ -802,29 +581,6 @@ Branch: REL9_4_STABLE [762b25653] 2019-11-05 14:27:38 -0500
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [64579be64] 2019-08-07 18:16:31 +0900
Branch: REL_12_STABLE Release: REL_12_0 [d8652ec55] 2019-08-07 18:17:34 +0900
Branch: REL_11_STABLE [d16d241a5] 2019-08-07 18:17:39 +0900
Branch: REL_10_STABLE [1ba4d0fe4] 2019-08-07 18:17:46 +0900
Branch: REL9_6_STABLE [7c64a2cd9] 2019-08-07 18:17:52 +0900
Branch: REL9_5_STABLE [1de3e0589] 2019-08-07 18:17:57 +0900
Branch: REL9_4_STABLE [1f7943698] 2019-08-07 18:18:04 +0900
-->
<para>
Fix some cases where an incomplete date specification is not
detected in <type>time with time zone</type> input (Alexander Lakhin)
</para>
<para>
If a time zone is provided, a date must be as well (since the UTC
offset implied by the zone might be time-varying). Depending on the
syntax used, this check was not enforced in all cases.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [5ac0d9360] 2019-09-22 17:45:59 -0400
Branch: REL_12_STABLE Release: REL_12_0 [860216efa] 2019-09-22 17:46:00 -0400
@ -885,43 +641,6 @@ Branch: REL_12_STABLE [43e43771b] 2019-10-28 12:21:13 -0400
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [e136a0d8c] 2019-08-19 18:01:09 -0400
Branch: REL_12_STABLE Release: REL_12_0 [cdc8d371e] 2019-08-19 18:00:57 -0400
Branch: REL_11_STABLE [2b24cf91a] 2019-08-19 18:00:57 -0400
-->
<para>
Restore the ability to take type information from
an <literal>AS</literal> clause
in <function>json[b]_populate_record()</function>
and <function>json[b]_populate_recordset()</function>
(Tom Lane)
</para>
<para>
If the record argument is NULL and has no declared composite type,
try to use the <literal>AS</literal> clause instead. This isn't
recommended usage, but it used to work, and now does again.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [592a16321] 2019-10-25 15:22:45 -0400
Branch: REL_12_STABLE [2bbdf8e2e] 2019-10-25 15:22:40 -0400
Branch: REL_11_STABLE [9f7a664fe] 2019-10-25 15:22:40 -0400
Branch: REL_10_STABLE [9bf247686] 2019-10-25 15:22:40 -0400
note: not a bug in v12/HEAD, the change there is just a test case addition
-->
<para>
Avoid crash when selecting a namespace node
in <literal>XMLTABLE</literal> (Chapman Flack)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [a7145f6bc] 2019-11-07 11:22:58 -0500
Branch: REL_12_STABLE [f6e72dc9c] 2019-11-07 11:22:59 -0500
Branch: REL_11_STABLE [b49b7f944] 2019-11-07 11:23:00 -0500
@ -929,6 +648,9 @@ Branch: REL_10_STABLE [5f794f757] 2019-11-07 11:23:02 -0500
Branch: REL9_6_STABLE [15783d057] 2019-11-07 11:23:03 -0500
Branch: REL9_5_STABLE [84780d468] 2019-11-07 11:23:04 -0500
Branch: REL9_4_STABLE [8d380864a] 2019-11-07 11:23:06 -0500
Branch: REL9_6_STABLE [a55018760] 2019-11-09 15:50:16 -0500
Branch: REL9_5_STABLE [30f6998ff] 2019-11-09 15:50:16 -0500
Branch: REL9_4_STABLE [18622caa3] 2019-11-09 15:50:16 -0500
-->
<para>
Fix detection of edge-case integer overflow in interval
@ -938,22 +660,6 @@ Branch: REL9_4_STABLE [8d380864a] 2019-11-07 11:23:06 -0500
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [3a54eb1a3] 2019-09-03 12:30:53 +0900
Branch: REL_12_STABLE Release: REL_12_0 [5380719f9] 2019-09-03 12:31:03 +0900
Branch: REL_11_STABLE [f967a1fda] 2019-09-03 12:31:08 +0900
Branch: REL_10_STABLE [604d20679] 2019-09-03 12:31:14 +0900
-->
<para>
Fix memory leaks
in <function>lower()</function>, <function>upper()</function>,
and <function>initcap()</function> functions when using ICU
collations (Konstantin Knizhnik)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [8af1624e3] 2019-11-02 16:45:32 -0400
Branch: REL_12_STABLE [43753c2cf] 2019-11-02 16:45:32 -0400
@ -974,65 +680,6 @@ Branch: REL9_6_STABLE [d43bd9dce] 2019-11-03 16:11:05 -0500
<listitem>
<!--
Author: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Branch: master [bde7493d1] 2019-08-28 12:55:33 +0300
Branch: REL_12_STABLE Release: REL_12_0 [6b7819a0b] 2019-08-28 12:56:03 +0300
Branch: REL_11_STABLE [1c99acc6e] 2019-08-28 12:57:10 +0300
Branch: REL_10_STABLE [756178232] 2019-08-28 13:02:58 +0300
Branch: REL9_6_STABLE [e2e616579] 2019-08-28 12:58:05 +0300
Branch: REL9_5_STABLE [cf00ca522] 2019-08-28 12:58:10 +0300
Branch: REL9_4_STABLE [6292cde1c] 2019-08-28 12:58:55 +0300
Branch: master [744c848dc] 2019-08-28 12:59:47 -0400
-->
<para>
Fix incorrect compression logic for GIN posting lists
(Heikki Linnakangas)
</para>
<para>
A GIN posting list item can require 7 bytes if the distance between
adjacent indexed TIDs exceeds 16TB. One step in the logic was out
of sync with that, and might try to write the value into a 6-byte
buffer. In principle this could cause a stack overrun, but on most
architectures it's likely that the next byte would be unused
alignment padding, making the bug harmless. In any case the bug
would be very difficult to hit.
</para>
</listitem>
<listitem>
<!--
Author: Alexander Korotkov <akorotkov@postgresql.org>
Branch: master [e5d8f3596] 2019-09-08 22:08:12 +0300
Branch: REL_12_STABLE Release: REL_12_0 [bc67f4189] 2019-09-08 21:17:31 +0300
Branch: REL_11_STABLE [749b04d1d] 2019-09-08 21:41:56 +0300
Branch: REL_10_STABLE [8f724002e] 2019-09-08 21:47:34 +0300
Branch: REL9_6_STABLE [b2037afec] 2019-09-08 21:48:44 +0300
Branch: REL9_5_STABLE [986319d46] 2019-09-08 21:49:15 +0300
Branch: REL9_4_STABLE [111fb7e42] 2019-09-08 21:58:17 +0300
Author: Alexander Korotkov <akorotkov@postgresql.org>
Branch: master [02f90879e] 2019-09-08 22:08:12 +0300
Branch: REL_12_STABLE Release: REL_12_0 [e6af7b367] 2019-09-08 21:17:37 +0300
Branch: REL_11_STABLE [d807200b4] 2019-09-08 21:46:58 +0300
Branch: REL_10_STABLE [92f6b49c4] 2019-09-08 21:47:34 +0300
Branch: REL9_6_STABLE [a5431b7d5] 2019-09-08 21:48:45 +0300
Branch: REL9_5_STABLE [3c155bafa] 2019-09-08 21:49:16 +0300
Branch: REL9_4_STABLE [1df412304] 2019-09-08 22:30:12 +0300
-->
<para>
Fix handling of infinity, NaN, and NULL values in KNN-GiST
(Alexander Korotkov)
</para>
<para>
The query's output order could be wrong (different from a plain
sort's result) if some distances computed for non-null column values
are infinity or NaN.
</para>
</listitem>
<listitem>
<!--
Author: Amit Kapila <akapila@postgresql.org>
Branch: master [70a6c37d5] 2019-10-21 08:57:32 +0530
Branch: REL_12_STABLE [62f4dd379] 2019-10-21 09:04:35 +0530
@ -1044,26 +691,6 @@ Branch: REL_12_STABLE [62f4dd379] 2019-10-21 09:04:35 +0530
<listitem>
<!--
Author: Alexander Korotkov <akorotkov@postgresql.org>
Branch: master [6cae9d2c1] 2019-09-19 21:48:39 +0300
Branch: REL_12_STABLE Release: REL_12_0 [31cbd7605] 2019-09-19 21:49:07 +0300
Branch: REL_11_STABLE [d6a90aac5] 2019-09-19 21:49:32 +0300
Branch: REL_10_STABLE [2da8e56db] 2019-09-19 21:50:00 +0300
Branch: REL9_6_STABLE [53d9cf2db] 2019-09-19 21:50:44 +0300
Branch: REL9_5_STABLE [ad458d0cd] 2019-09-19 22:09:51 +0300
Branch: REL9_4_STABLE [332eda5bd] 2019-09-19 22:10:46 +0300
Branch: REL_11_STABLE [984b9ba1d] 2019-09-19 23:36:01 +0300
Branch: REL_10_STABLE [2f0434e8e] 2019-09-19 23:39:26 +0300
Branch: REL9_6_STABLE [140b7b1f9] 2019-09-19 23:39:31 +0300
Branch: REL9_5_STABLE [388939748] 2019-09-19 23:39:35 +0300
-->
<para>
Fix handling of searches for NULL in KNN-SP-GiST (Nikita Glukhov)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [db477b691] 2019-10-21 14:18:01 -0400
Branch: REL_12_STABLE [4f2ad5226] 2019-10-21 14:18:16 -0400
@ -1139,84 +766,6 @@ Branch: REL9_4_STABLE [b20233aac] 2019-11-07 14:21:52 -0500
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [aef362385] 2019-09-02 14:03:05 -0400
Branch: REL_12_STABLE Release: REL_12_0 [90433c38e] 2019-09-02 14:02:45 -0400
Branch: REL_11_STABLE [5524ef558] 2019-09-02 14:02:46 -0400
Branch: REL_10_STABLE [3080f8f61] 2019-09-02 14:02:46 -0400
Branch: REL9_6_STABLE [b0b2ef25e] 2019-09-02 14:02:46 -0400
Branch: REL9_5_STABLE [62724bd95] 2019-09-02 14:02:46 -0400
Branch: REL9_4_STABLE [89535db97] 2019-09-02 14:02:46 -0400
-->
<para>
In <application>psql</application>, resynchronize internal state
about the server after an unexpected connection loss and successful
reconnection (Peter Billen, Tom Lane)
</para>
<para>
Ordinarily this is unnecessary since the state would be the same
anyway. But it can matter in corner cases, such as where the
connection might lead to one of several servers. This change
causes <application>psql</application> to re-issue any interactive
messages that it would have issued at startup, for example about
whether SSL is in use.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [6338fa3e7] 2019-08-25 15:04:04 -0400
Branch: REL_12_STABLE Release: REL_12_0 [363382521] 2019-08-25 15:04:04 -0400
Branch: REL_11_STABLE [5fc7b1e93] 2019-08-25 15:04:04 -0400
Branch: REL_10_STABLE [fb55e9539] 2019-08-25 15:04:04 -0400
Branch: REL9_6_STABLE [28d2ce3c7] 2019-08-25 15:04:04 -0400
Branch: REL9_5_STABLE [65b1cad5a] 2019-08-25 15:04:04 -0400
Branch: REL9_4_STABLE [c693c5c49] 2019-08-25 15:04:04 -0400
-->
<para>
Avoid platform-specific null pointer dereference
in <application>psql</application> (Quentin Rameau)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: REL9_6_STABLE [404cbc562] 2019-10-26 17:37:19 -0400
Branch: REL9_5_STABLE [7fc50a8a7] 2019-10-26 17:37:19 -0400
Branch: REL9_4_STABLE [ddcc582a4] 2019-10-26 17:37:20 -0400
-->
<para>
Fix <application>pg_dump</application>'s handling of circular
dependencies in views (Tom Lane)
</para>
<para>
In some cases a view may depend on an object
that <application>pg_dump</application> needs to dump later than the
view; the most common example is that a query using <literal>GROUP
BY</literal> on a primary-key column may be semantically invalid
without the primary key. This is now handled by emitting a
dummy <command>CREATE VIEW</command> command that just establishes
the view's column names and types, and then later
emitting <command>CREATE OR REPLACE VIEW</command> with the full
view definition. Previously, the dummy definition was actually
a <command>CREATE TABLE</command> command, and this was
automagically converted to a view by a later <command>CREATE
RULE</command> command. The new approach has been used successfully
in <productname>PostgreSQL</productname> version 10 and later. We
are back-patching it into older releases now because of reports that
the previous method causes bogus error messages about the view's
replica identity status. This change also avoids problems when
trying to use the <option>--clean</option> option during a restore
involving such a view.
</para>
</listitem>
<listitem>
<!--
Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
Branch: master [1752e3516] 2019-10-17 09:58:01 +0200
Branch: REL_12_STABLE [b304b2b65] 2019-10-17 09:58:01 +0200
@ -1254,57 +803,7 @@ Branch: REL9_4_STABLE [da5cd7a68] 2019-11-04 16:25:05 -0500
<para>
Previously, if two triggers on different tables had the same names,
they would be sorted in OID-based order, which is less desirable
than sorting by name. Likewise for RLS policies.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [31d43710f] 2019-08-13 16:58:32 -0400
Branch: REL_12_STABLE Release: REL_12_0 [6844adba5] 2019-08-13 16:57:58 -0400
Branch: REL_11_STABLE [4dea8ad56] 2019-08-13 16:57:58 -0400
Branch: REL_10_STABLE [f2bdfebb9] 2019-08-13 16:57:58 -0400
Branch: REL9_6_STABLE [2b608ba31] 2019-08-13 16:57:59 -0400
Branch: REL9_5_STABLE [c56858487] 2019-08-13 16:57:59 -0400
Branch: REL9_4_STABLE [63ae888a9] 2019-08-13 16:57:59 -0400
-->
<para>
Fix <application>pg_dump</application> to again work with pre-8.3
source servers (Tom Lane)
</para>
<para>
A previous fix caused <application>pg_dump</application> to always
try to query <structname>pg_opfamily</structname>, but that catalog
doesn't exist before 8.3.
</para>
</listitem>
<listitem>
<!--
Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
Branch: REL_11_STABLE [3574c0ac0] 2019-11-05 01:23:39 -0300
Branch: REL_10_STABLE [5ee8f0fe1] 2019-11-05 10:08:55 -0300
Branch: REL9_6_STABLE [12a51e2eb] 2019-11-05 09:57:36 -0300
Branch: REL9_5_STABLE [d38635725] 2019-11-05 09:57:36 -0300
Branch: REL9_4_STABLE [9fb25fda6] 2019-11-05 09:57:35 -0300
-->
<para>
In <application>pg_restore</application>, treat
<option>-f -</option> as meaning <quote>output to stdout</quote>
(&Aacute;lvaro Herrera)
</para>
<para>
This synchronizes <application>pg_restore</application>'s behavior
with some other applications, and in particular makes pre-v12 branches
act similarly to v12's <application>pg_restore</application>,
simplifying creation of dump/restore scripts that work across
multiple <productname>PostgreSQL</productname> versions. Before this
change, <application>pg_restore</application> interpreted such a
switch as meaning <quote>output to a file
named <filename>-</filename></quote>, but few people would want that.
than sorting them by table name. Likewise for RLS policies.
</para>
</listitem>
@ -1319,7 +818,7 @@ Branch: REL_12_STABLE [702fd3b71] 2019-10-15 00:35:51 +0200
<para>
In <application>pg_upgrade</application>, reject tables with
columns of type <type>sql_identifier</type>, as that has changed
representation in v12 (Tomas Vondra)
representation in version 12 (Tomas Vondra)
</para>
</listitem>
@ -1350,42 +849,7 @@ Branch: REL_10_STABLE [e86ece221] 2019-10-16 13:26:56 +0200
<para>
The previous coding could be fooled by cases where the data type of
interest underlies a domain or composite type.
</para>
</listitem>
<listitem>
<!--
Author: Robert Haas <rhaas@postgresql.org>
Branch: master [286af0ce1] 2019-09-06 08:22:32 -0400
Branch: REL_12_STABLE Release: REL_12_0 [ce5542d40] 2019-09-06 08:49:56 -0400
Branch: REL_11_STABLE [61c65cce4] 2019-09-06 08:53:13 -0400
Branch: REL_10_STABLE [62fb12d7e] 2019-09-06 08:56:45 -0400
Branch: REL9_6_STABLE [23df88226] 2019-09-06 09:01:45 -0400
Branch: REL9_5_STABLE [f697c6396] 2019-09-06 09:05:12 -0400
Branch: REL9_4_STABLE [fbe897134] 2019-09-06 09:09:09 -0400
-->
<para>
Detect file read errors
during <application>pg_basebackup</application> (Jeevan Chalke)
</para>
</listitem>
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [522baf148] 2019-09-04 13:21:11 +0900
Branch: REL_12_STABLE Release: REL_12_0 [64eb426af] 2019-09-04 13:23:41 +0900
Branch: REL_11_STABLE [996c92b27] 2019-09-04 13:24:06 +0900
Branch: REL_10_STABLE [dbbae33bc] 2019-09-04 13:24:12 +0900
-->
<para>
In <application>pg_basebackup</application>, don't fsync output files
until the end of backup (Michael Paquier)
</para>
<para>
The previous coding could result in timeout failures if fsync was slow.
interest underlies a stored column of a domain or composite type.
</para>
</listitem>
@ -1410,23 +874,6 @@ Branch: REL_12_STABLE [c2e3b311d] 2019-10-04 09:16:03 +0900
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [be182e4f9] 2019-08-28 11:47:35 +0900
Branch: REL_12_STABLE Release: REL_12_0 [e96f52443] 2019-08-28 11:48:15 +0900
Branch: REL_11_STABLE [f51006ea9] 2019-08-28 11:48:19 +0900
Branch: REL_10_STABLE [19bfa15a8] 2019-08-28 11:48:23 +0900
Branch: REL9_6_STABLE [d64789e97] 2019-08-28 11:48:29 +0900
Branch: REL9_5_STABLE [e9dcbc9c3] 2019-08-28 11:48:33 +0900
-->
<para>
In <application>pg_rewind</application> with an online source
cluster, disable timeouts, much
as <application>pg_dump</application> does (Alexander Kukushkin)
</para>
</listitem>
<listitem>
<!--
Author: Fujii Masao <fujii@postgresql.org>
Branch: master [a0c96856e] 2019-11-07 16:31:36 +0900
Branch: REL_12_STABLE [e5cfb8cbb] 2019-11-07 16:32:37 +0900
@ -1445,23 +892,6 @@ Branch: REL9_4_STABLE [1accf9974] 2019-11-07 16:33:58 +0900
<listitem>
<!--
Author: Peter Geoghegan <pg@bowt.ie>
Branch: master [3b6b54f17] 2019-09-12 15:45:08 -0700
Branch: REL_12_STABLE Release: REL_12_0 [3097a0d6e] 2019-09-12 15:45:07 -0700
Branch: REL_11_STABLE [e87b00b5f] 2019-09-12 15:45:05 -0700
Branch: REL_10_STABLE [09b236af9] 2019-09-12 15:45:03 -0700
Branch: REL9_6_STABLE [835646503] 2019-09-12 15:45:01 -0700
Branch: REL9_5_STABLE [9064cc139] 2019-09-12 15:44:59 -0700
-->
<para>
In <application>pg_waldump</application>,
include the <literal>newitemoff</literal> field in btree page split
records (Peter Geoghegan)
</para>
</listitem>
<listitem>
<!--
Author: Andres Freund <andres@anarazel.de>
Branch: master [e4d92126f] 2019-10-29 22:53:05 -0700
Branch: REL_12_STABLE [d4b5206b2] 2019-10-29 22:53:30 -0700
@ -1495,117 +925,6 @@ Branch: REL9_5_STABLE [c3882f8b8] 2019-10-29 19:28:34 -0700
<listitem>
<!--
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [71d84efba] 2019-08-26 11:14:18 +0900
Branch: REL_12_STABLE Release: REL_12_0 [63fc3b124] 2019-08-26 11:14:24 +0900
Branch: REL_11_STABLE [5d76c8037] 2019-08-26 11:14:28 +0900
Branch: REL_10_STABLE [4fca14600] 2019-08-26 11:14:33 +0900
Branch: REL9_6_STABLE [eb91b8ee6] 2019-08-26 11:14:39 +0900
Branch: REL9_5_STABLE [a21ec1a95] 2019-08-26 11:14:44 +0900
Author: Michael Paquier <michael@paquier.xyz>
Branch: master [9acda7311] 2019-08-27 09:11:31 +0900
Branch: REL_12_STABLE Release: REL_12_0 [b783a38d4] 2019-08-27 09:11:38 +0900
Branch: REL_11_STABLE [128e9b2cc] 2019-08-27 09:11:43 +0900
Branch: REL_10_STABLE [c90096009] 2019-08-27 09:11:48 +0900
Branch: REL9_6_STABLE [c4d75313e] 2019-08-27 09:12:10 +0900
Branch: REL9_5_STABLE [4ed3bda49] 2019-08-27 09:12:14 +0900
-->
<para>
Fix <application>vacuumdb</application> with a
high <option>--jobs</option> option to handle running out of file
descriptors better (Michael Paquier)
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [fe9b7b2fe] 2019-08-15 15:21:47 -0400
Branch: REL_12_STABLE Release: REL_12_0 [a4504696e] 2019-08-15 15:21:47 -0400
Branch: REL_11_STABLE [6070ccdd1] 2019-08-15 15:21:48 -0400
-->
<para>
Fix PL/pgSQL to handle replacements of composite types better
(Tom Lane)
</para>
<para>
Cover the case where a composite type is dropped entirely, and
then a new type of the same name is created, between executions
of a PL/pgSQL function. Variables of the composite type will now
update to match the new definition.
</para>
</listitem>
<listitem>
<!--
Author: Peter Geoghegan <pg@bowt.ie>
Branch: master [6754fe65a] 2019-08-12 15:21:32 -0700
Branch: REL_12_STABLE Release: REL_12_0 [a05fa2c0e] 2019-08-12 15:21:30 -0700
Branch: REL_11_STABLE [4f393793f] 2019-08-12 15:21:28 -0700
Branch: REL_10_STABLE [f8d2cdc12] 2019-08-12 15:21:27 -0700
-->
<para>
Fix <filename>contrib/amcheck</filename> to skip unlogged indexes
during hot standby (Andrey Borodin, Peter Geoghegan)
</para>
<para>
An unlogged index won't necessarily contain valid data in this
context, so don't try to check it.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [efc77cf5f] 2019-08-06 18:04:51 -0400
Branch: REL_12_STABLE Release: REL_12_0 [2f76f4182] 2019-08-06 18:04:51 -0400
Branch: REL_11_STABLE [113b3d903] 2019-08-06 18:04:51 -0400
Branch: REL_10_STABLE [5b3e6c13f] 2019-08-06 18:04:51 -0400
Branch: REL9_6_STABLE [e519eded6] 2019-08-06 18:04:51 -0400
Branch: REL9_5_STABLE [ced272d2d] 2019-08-06 18:04:52 -0400
Branch: REL9_4_STABLE [614119d00] 2019-08-06 18:04:52 -0400
-->
<para>
Fix <filename>contrib/intarray</filename>'s GiST opclasses to not
fail for empty arrays with <literal>&lt;@</literal> (Tom Lane)
</para>
<para>
A clause like <literal><replaceable>array_column</replaceable>
&lt;@ <replaceable>constant_array</replaceable></literal> is
considered indexable, but the index search may not find index
entries for empty arrays in the array column; of course, such
entries should trivially match the search.
</para>
<para>
The only practical back-patchable fix for this requires
making <literal>&lt;@</literal> index searches scan the whole index,
which is what this patch does. This is unfortunate: it means that
the query performance is likely worse than a plain sequential scan
would be.
</para>
<para>
Applications whose performance is adversely impacted by this change
have a couple of options. They could switch to a GIN index, which
doesn't have this bug, or they could replace
<literal><replaceable>array_column</replaceable>
&lt;@ <replaceable>constant_array</replaceable></literal>
with <literal><replaceable>array_column</replaceable>
&lt;@ <replaceable>constant_array</replaceable>
AND <replaceable>array_column</replaceable>
&amp;&amp; <replaceable>constant_array</replaceable></literal>.
That will provide about the same performance as before, and it will
find all non-empty subsets of the given constant array, which is all
that could reliably be expected of the query before.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [06a367c38] 2019-10-10 14:24:56 -0400
Branch: REL_12_STABLE [7ed1bcaed] 2019-10-10 14:24:57 -0400
@ -1626,32 +945,6 @@ Branch: REL_12_STABLE [7ed1bcaed] 2019-10-10 14:24:57 -0400
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: REL9_6_STABLE [984ee0f6d] 2019-09-08 13:45:13 -0400
Branch: REL9_5_STABLE [f1a9abe61] 2019-09-08 13:45:13 -0400
Branch: REL9_4_STABLE [96db9d5e2] 2019-09-08 13:45:13 -0400
-->
<para>
Allow <literal>configure --with-python</literal> to succeed when
only <filename>python3</filename> or
only <filename>python2</filename> can be found (Peter Eisentraut,
Tom Lane)
</para>
<para>
Search for <filename>python</filename>,
then <filename>python3</filename>,
then <filename>python2</filename>, so
that <application>configure</application> will succeed in the
increasingly more common situation where no executable named
simply <filename>python</filename> exists. It's still possible to
override the choice by setting the <envar>PYTHON</envar> environment
variable.
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [d995fd667] 2019-10-21 13:52:25 -0400
Branch: REL_12_STABLE [ca658c91a] 2019-10-21 13:52:25 -0400
Branch: REL_11_STABLE [4e19bd41d] 2019-10-21 13:52:25 -0400
@ -1698,31 +991,8 @@ Branch: REL9_4_STABLE [930787c7f] 2019-10-18 20:20:33 -0700
zero for use with an assembly instruction that does not accept that
register, causing a build failure. We have seen only one long-ago
report that matches this bug, but it could cause problems for people
trying to build modified code or use atypical compiler options.
</para>
</listitem>
<listitem>
<!--
Author: Noah Misch <noah@leadboat.com>
Branch: master [dd50f1a43] 2019-09-13 19:34:06 -0700
Branch: REL_12_STABLE Release: REL_12_0 [1c6b62a7d] 2019-09-13 19:34:18 -0700
Branch: REL_11_STABLE [40ad42025] 2019-09-13 19:34:18 -0700
Branch: REL_10_STABLE [8972ac696] 2019-09-13 19:34:19 -0700
Branch: REL9_6_STABLE [a1df9a015] 2019-09-13 19:34:19 -0700
Branch: REL9_5_STABLE [75941f257] 2019-09-13 19:34:19 -0700
-->
<para>
On PowerPC, avoid depending on the xlc
compiler's <function>__fetch_and_add()</function> function
(Noah Misch)
</para>
<para>
xlc 13 and newer interpret this function in a way incompatible with
our usage, resulting in an unusable build
of <productname>PostgreSQL</productname>. Fix by using custom
assembly code instead.
trying to build modified <productname>PostgreSQL</productname> code
or use atypical compiler options.
</para>
</listitem>
@ -1765,26 +1035,6 @@ Branch: REL9_4_STABLE [8f8809091] 2019-10-04 15:38:36 -0400
</para>
</listitem>
<listitem>
<!--
Author: Tom Lane <tgl@sss.pgh.pa.us>
Branch: master [df4fbcd89] 2019-09-20 19:53:33 -0400
Branch: REL_12_STABLE Release: REL_12_0 [2966e30e5] 2019-09-20 19:53:52 -0400
Branch: REL_11_STABLE [24f64fba0] 2019-09-20 19:54:00 -0400
Branch: REL_10_STABLE [769802ef9] 2019-09-20 19:54:07 -0400
Branch: REL9_6_STABLE [0bd64398e] 2019-09-20 19:54:20 -0400
Branch: REL9_5_STABLE [128abdef7] 2019-09-20 19:54:29 -0400
Branch: REL9_4_STABLE [29e47f83c] 2019-09-20 19:54:36 -0400
-->
<para>
Update time zone data files to <application>tzdata</application>
release 2019c for DST law changes in Fiji and Norfolk Island, plus
historical corrections for Alberta, Austria, Belgium, British
Columbia, Cambodia, Hong Kong, Indiana (Perry County), Kaliningrad,
Kentucky, Michigan, Norfolk Island, South Korea, and Turkey.
</para>
</listitem>
</itemizedlist>
</sect2>