Mar - Apr 2022

3 commits
Disable synchronize_seqscans in 027_stream_regress.pl.
4 Apr 2022Tom Lanemaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Tom Lane proposed disabling synchronized scans in #78 and confirmed doing so in #79; the commit adds synchronize_seqscans = off and explanatory comments without a matching attachment.

  • Thomas Munrosaw the final versionLLM analysis, 80% sureReported small-shared_buffers nondeterminism in #77, which Tom Lane connected to the synchronized-scan failure and fix in #78.
Try to stabilize vacuum test.
23 Mar 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1REL_14_STABLE14.3REL_13_STABLE13.7+1 more
LLM analysis, 100% surelanded as postedfrom patchset #701 of 1 reviewer saw the final version

The changes to sql/vacuum.sql and expected/vacuum.out exactly match #70: use a temporary table and DISABLE_PAGE_SKIPPING for the truncation checks.

  • Masahiko Sawadasaw the final versionLLM analysis, 92% sureReported the concrete vacuum truncation failure and regression diff in #69, directly prompting the fix posted in #70.
Log regression.diffs in 027_stream_regress.pl.
20 Mar 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surelanded as postedfrom patchset #661 of 1 reviewer saw the final version

The complete 027_stream_regress.pl diff matches #66, replacing system_or_bail with exit-status checking and logging regression.diffs on failure.

  • Andrew Dunstansaw the final versiontrailerExplicitly accepted the logging patch in #67, before it was committed unchanged.

Jan - Feb 2022

10 commits
Fix recovery conflict in 027_stream_regress.pl.
2 Feb 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #572 of 2 reviewers saw the final version

Matches #57 except that max_standby_streaming_delay is 600s rather than -1, implementing Tom Lane's finite-timeout recommendation in #63.

  • Andres Freundsaw the final versiontrailerAccepted the proposed delay adjustment in #58, after which only the timeout value changed.
  • Tom Lanesaw the final versiontrailerExamined the failure timings and requested the ten-minute finite timeout in #63, which was adopted.
Specify --host in 027_stream_regress.pl's pg_regress invocation.
27 Jan 2022Andres Freundmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surelanded as postedfrom patchset #511 of 1 reviewer saw the final version

The single added --host argument in 027_stream_regress.pl exactly matches Andres Freund's attachment in #51.

  • Andrew Dunstansaw the final versionLLM analysis, 91% sureReported excessive runtime in the landed replay test in #44 and supplied comparative timings in #49, leading to this one-line host fix.
Add new simple TAP test for tablespaces, attempt II.
21 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #321 of 2 reviewers saw the final version

Relands the standalone tablespace TAP-test component of #32, adding perl2host conversion and removal of internal /./ path components as explained in #37 and #38.

  • Andres Freundsaw an earlier versionLLM analysis, 98% sureRequested ordinary-tablespace coverage in #25 and tablespace movement cases in #29, before the later Windows path corrections.
  • Noah Mischsaw the final versionLLM analysis, 98% sureDiagnosed the MSYS path mismatch and prescribed perl2host in #37, which the relanding adopted.
Test replay of regression tests, attempt II.
17 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #321 of 6 reviewers saw the final version

Relands the replay-test component of #32 with quoted paths and EXTRA_REGRESS_OPTS moved before explicit arguments so the test's port wins, following #35.

  • Tsunakawa, Takayukisaw an earlier versionLLM analysis, 99% sureProposed comparing primary and standby dumps and identified sequence-value differences in #2, before substantial implementation changes.
  • Andres Freundsaw the final versionLLM analysis, 99% sureReviewed successive replay designs and diagnosed the EXTRA_REGRESS_OPTS port override in #35, prescribing the argument ordering used for relanding.
  • Andrew Dunstansaw an earlier versionLLM analysis, 98% sureReviewed Perl usage and explained PG_REGRESS handling in #19, before later dump, configuration, and argument-order changes.
  • Tom Lanesaw an earlier versionLLM analysis, 99% sureRequired expensive WAL consistency checking to be opt-in in #6, before the subsequent tablespace and dump-comparison rework.
  • Anastasia Lubennikovasaw an earlier versionLLM analysis, 97% sureReviewed and rebased the early test, removed unnecessary authentication setup, and supplied cleanup fixes in #11 and #12.
  • Michael Paquiersaw an earlier versionLLM analysis, 97% sureReviewed pg_regress integration and requested extra-option handling in #30 and #31, before its implementation and later port-precedence fix.
Revert "Add new simple TAP test for tablespaces."
14 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1reverts 1 commit here
LLM analysis, 100% surewritten from the discussion, no matching patch

Removes the 96-line tablespace TAP test introduced by d1511fe1b0; #34 explains the Windows path failure and rollback, with no attached revert patch.

Revert "Test replay of regression tests."
14 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1reverts 1 commit here
LLM analysis, 100% surewritten from the discussion, no matching patch

Reverses the replay-test additions from 123828a7fa, following the buildfarm failures and impending revert reported in #33.

Test replay of regression tests.
14 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #320 of 6 reviewers saw the final version

Takes the replay-test component of #32, adding shell quotes around --dlpath and --outputdir values and changing a documentation phrase from 'some of the tests' to 'certain tests'.

  • 綱川 貴之 (Takayuki Tsunakawa)saw an earlier versiontrailerProposed dump comparison and identified sequence-cache differences in #2, before substantial changes to the test and tablespace handling.
  • Andres Freundsaw an earlier versiontrailerReviewed invocation and replay coverage, requesting prepared transactions and pg_dumpall in #29, which materially expanded the subsequent version.
  • Andrew Dunstansaw an earlier versiontrailerReviewed Perl and executable-path handling in #5, #19, and #21, before later replay coverage and extra-option changes.
  • Tom Lanesaw an earlier versiontrailerArgued for opt-in WAL consistency checking and broader eventual test-suite support in #6 and #8, before the main rework.
  • Anastasia Lubennikovasaw an earlier versiontrailerReviewed and rebased the early implementation and supplied authentication and cleanup corrections in #11 and #12.
  • Michael Paquiersaw an earlier versiontrailerDiscussed shared pg_regress infrastructure and requested EXTRA_REGRESS_OPTS support in #30 and #31, before those invocation changes were implemented.
Add new simple TAP test for tablespaces.
14 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #320 of 2 reviewers saw the final version

Commits exactly the 96-line src/test/modules/test_misc/t/002_tablespace.pl component of the four-patch series in #32.

  • Andres Freundsaw an earlier versiontrailerRequested the separate test in #25 and reviewed its initial version in #29, before its expansion from six to twenty checks.
  • Michael Paquierno review foundtrailerHis visible contributions in #30 and #31 address the replay test rather than this separate tablespace TAP test.
Use in-place tablespaces in regression test.
14 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #321 of 2 reviewers saw the final version

Takes the regression-tablespace conversion component of #32, additionally removing the obsolete rm -rf testtablespace cleanup rule from src/test/regress/GNUmakefile.

  • Andres Freundsaw the final versiontrailerReviewed the replacement of ordinary regression tablespaces and checked negative creation coverage in #25 and #29, after which this component changed only lightly.
  • Michael Paquierno review foundtrailerThe thread shows his review of replay-test integration in #30 and #31, but no specific review of this regression-tablespace conversion.
Allow "in place" tablespaces.
14 Jan 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #321 of 2 reviewers saw the final version

Takes the in-place tablespace component of #32 with unchanged backend logic and adds a 19-line allow_in_place_tablespaces description to doc/src/sgml/config.sgml.

  • Andres Freundsaw the final versiontrailerProposed directories directly under pg_tblspc in #22 and reviewed their implementation in #29, followed by redundant-check simplification and documentation.
  • Michael Paquierno review foundtrailerHis visible reviews in #30 and #31 concern replay-test invocation, without specific review of the tablespace backend change.

Credits across 13 commits

Author
Thomas MunroLLM9
Andres FreundLLM1
Reviewer
Andres Freund5
Michael Paquier4
Andrew Dunstan2
Tom Lane2
Anastasia Lubennikova1
Tsunakawa, Takayuki1
Committer
Thomas Munro11
Andres Freund1
Tom Lane1

N LLM means N of the credits come from LLM analysis, not from commit trailers.