Aug 2025

1 commit
Hide expensive pg_upgrade test behind PG_TEST_EXTRA
5 Aug 2025Álvaro Herreramaster19beta1REL_19_STABLE19beta1REL_18_STABLE18beta3
LLM analysis, 100% suremodified before commitlight changesfrom patchset #1012 of 2 reviewers saw the final version

Implements the PG_TEST_EXTRA guard from #101, renaming its option to regress_dump_restore and expanding the skip message. The commit also adds documentation in doc/src/sgml/regress.sgml, acknowledged in #106.

  • Daniel Gustafssonsaw the final versionLLM analysis, 100% sureExplicitly supported placing the submitted test behind PG_TEST_EXTRA in #102, before commitment.
  • Tom Lanesaw the final versionLLM analysis, 100% sureExplicitly endorsed the opt-in patch in #103, before commitment.

Apr 2025

4 commits
Set log_statement=none in t/002_pg_upgrade.pl
5 Apr 2025Álvaro Herreramaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 97% suremodified before commitlight changesfrom patchset #962 of 2 reviewers saw the final version

Preserves all three log_statement overrides from #96, with revised comments and spacing. The later series #97 adapted logging to a two-cluster optimization that did not land, whereas this commit retains the original three-cluster implementation.

  • Ashutosh Bapatsaw the final versionLLM analysis, 98% sureReported testing the logging change without a local timing improvement and carried it into his follow-up series in #97.
  • Andres Freundsaw the final versionLLM analysis, 98% sureProposed disabling statement logging after reporting Valgrind timeouts in #95 and welcomed the resulting log-volume reduction in #99.
002_pg_upgrade.pl: Move pg_dump test code for better stability
3 Apr 2025Álvaro Herreramaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1Also in: statistics import and export: another difference in dump/restore (not this thread)
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #921 of 1 reviewer saw the final version

Commits the test relocation and statistics re-enablement from #92 in src/bin/pg_upgrade/t/002_pg_upgrade.pl, with comment cleanup; its initialization-variable changes landed separately. This supersedes the attribution in thread #51340: that thread's #588164 only removed --no-statistics, whereas this thread's later patch contains the full committed stability fix.

  • Álvaro Herrerasaw the final versioncommittercommitted the patch
002_pg_upgrade.pl: rename some variables for clarity
3 Apr 2025Álvaro Herreramaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #921 of 1 reviewer saw the final version

Extracts the old_node_params/old_initdb_params renaming and separate new-node parameter variables from #92. The roundtrip relocation and statistics changes were reserved for the following commit.

  • Álvaro Herrerasaw the final versioncommittercommitted the patch
Verify roundtrip dump/restore of regression database
1 Apr 2025Álvaro Herreramaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #841 of 4 reviewers saw the final version

Combines the roundtrip test, directory-format default, and statistics exclusion represented by #84. The committer replaces the test subroutine with an inline SKIP block, changes comparison dumps to run_log(), removes adjustment assertions, and rewrites documentation, as acknowledged in #85.

  • Michael Paquiersaw an earlier versiontrailerReviewed runtime, normalization, helper extraction, and command invocation across #2, #13, #22, #34, and #35, with further scheduling objections in #70 before the final control-flow changes.
  • Daniel Gustafssonsaw an earlier versiontrailerProposed reusing pg_upgrade, reviewed version checks and diagnostics in #28 and #30, and supported the existing schedule in #82, before the committer's moderate implementation changes.
  • Tom Lanesaw an earlier versiontrailerExamined coverage limitations and runtime in #9 and #23, then challenged schedule filtering in #78, before the final implementation changes.
  • Álvaro Herrerasaw the final versiontrailerFound the locale failure, measured runtime, advocated one parallel format, and reviewed and edited the committed implementation as described in #45, #54, #65, and #85.

Feb 2025

1 commit
Refactor TAP test code for file comparisons into new routine in Utils.pm
9 Feb 2025Michael Paquiermaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #352 of 3 reviewers saw the final version

Commits the comparison-helper portion of Michael Paquier's two-patch submission #35, leaving the roundtrip test pending. Relative to that attachment, it removes unused result assignments and adds a comment; the commit credits Ashutosh Bapat, whose helper appeared in #26.

  • Daniel Gustafssonsaw an earlier versionLLM analysis, 98% sureReviewed compare_dumps() diagnostics and suggested checking for identical filenames in #28, then responded positively in #30, before the helper gained its generalized comparator interface.
  • Alvaro Herrerasaw the final versionLLM analysis, 88% sureWelcomed the extracted helper and explained how it could replace his nonstandard Test::Differences dependency in #36.
  • Michael Paquiersaw the final versioncommittercommitted the patch

Dec 2024

1 commit
Dump not-null constraints on inherited columns correctly
13 Dec 2024Álvaro Herreramaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1Also in: Difference in dump from original and restored database due to NOT NULL constraints on children (modified before commit)

Credits across 7 commits

Author
Ashutosh Bapat5
Alvaro HerreraLLM2
Reviewer
Alvaro Herreracommitter4
Michael Paquiercommitter2
Daniel Gustafsson1
Tom Lane1
Reported by
Ashutosh Bapat1
Committer
Alvaro Herrera6
Michael Paquier1

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