May 2022

2 commits
Improve and fix some issues in the TAP tests of pg_upgrade
21 May 2022Michael Paquiermaster15beta2REL_19_STABLE15beta2REL_18_STABLE15beta2REL_17_STABLE15beta2REL_16_STABLE15beta2REL_15_STABLE15beta2
LLM analysis, 100% surelanded as postedfrom patchset #711 of 1 reviewer saw the final version

The commit and #71 have identical stable patch IDs: database-name restoration, EXTRA_REGRESS_OPTS spacing, and the pg_regress status assertion all match.

  • Noah Mischsaw the final versiontrailerNoah identified naming and command-status defects in #70 and confirmed the revised Windows result in #72 before commitment.
Fix several issues with the TAP tests of pg_upgrade
10 May 2022Michael Paquiermaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #641 of 1 reviewer saw the final version

Matches the command checks, connection-argument cleanup, and shell-fragment handling in #64; the committed database-name suffix uses additional backslashes, as discussed in #65.

  • Noah Mischsaw the final versionLLM analysis, 98% sureNoah's detailed review of the landed form of #45 in #62 identified every issue addressed by this focused test-fix patch.

Apr 2022

2 commits
Make upgradecheck a no-op in MSVC's vcregress.pl
2 Apr 2022Michael Paquiermaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #591 of 1 reviewer saw the final version

Implements #59, additionally removing the obsolete Windows CI upgradecheck step and unused generate_db() routine, and refining usage text and the no-op message.

  • Justin Pryzbysaw the final versiontrailerJustin requested removal of the obsolete CI invocation in #60, which Michael incorporated before committing as confirmed in #61.
Switch the regression tests of pg_upgrade to use TAP tests
1 Apr 2022Michael Paquiermaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 99% suremodified before commitlight changesfrom patchset #452 of 4 reviewers saw the final version

Derived from #45, with expanded TESTING instructions and comment cleanup plus small test-invocation corrections: restore dumps through postgres and pass an empty --bindir to pg_regress.

  • Andrew Dunstansaw an earlier versiontrailerAndrew proposed using historical dumps in #2 and examined buildfarm compatibility in #10 and #15, before substantial cross-version and Windows revisions.
  • Andres Freundsaw the final versiontrailerAndres reviewed test behavior and portability in #31, corrected option precedence in #42, and assessed the final version's buildfarm behavior in #50.
  • Rachel Heatonsaw an earlier versiontrailerRachel ran the tests and reported the removed get_new_node API in #7, before later cross-version and MSVC redesigns.
  • Tom Lanesaw the final versiontrailerTom reviewed buildfarm compatibility and output placement, then explicitly supported proceeding with the final patch in #52.

Mar 2022

1 commit
Fix check for PGHOST[ADDR] in pg_upgrade with Windows and temporary paths
1 Mar 2022Michael Paquiermaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #401 of 2 reviewers saw the final version

Takes only the server.c component of the two-patch series in #40; that component has the same stable patch ID as the commit, while the TAP conversion remains for #41.

  • Andres Freundsaw the final versionLLM analysis, 97% sureAndres identified abstract-socket paths and proposed the exact is_unixsock_path() solution in #30, which was incorporated into #40.
  • Julien Rouhaudsaw an earlier versionLLM analysis, 90% sureJulien reported the concrete Windows PGHOST rejection in #28, before the temporary environment workaround was replaced by the server.c fix.

Feb 2022

2 commits
Remove REGRESS_OUTPUTDIR environment variable.
13 Feb 2022Thomas Munromaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 99% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Implements the tmp_check substitution proposed in #34 and Thomas Munro's follow-up in #35, removing REGRESS_OUTPUTDIR from recovery tests and MSVC; the latter message's attachment has no supplied patchset entry.

  • Andres Freundsaw the final versionLLM analysis, 100% sureAndres proposed reusing PostgreSQL::Test::Utils::tmp_check in #34 and explicitly reviewed Thomas's unindexed attachment as sensible in #36 before commitment.
Add PostgreSQL::Test::Cluster::config_data()
8 Feb 2022Michael Paquiermaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1Also in: GUC flags (part of a posted patchset)

Credits across 7 commits

Author
Michael Paquier4 LLM6
Reviewer
Andres Freund1
Andrew Dunstan1
Justin Pryzby1
Noah Misch1
Rachel Heaton1
Tom Lane1
Reported by
Noah Misch1
Committer
Michael Paquier6
Thomas Munro1

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