Jul - Aug 2022

3 commits
Revert recent changes to 002_pg_upgrade.pl.
4 Aug 2022Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1REL_15_STABLE15beta3reverts 1 commit here
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Removes the added VACUUM FULL and horizon checks from src/bin/pg_upgrade/t/002_pg_upgrade.pl; none of the indexed attachments proposes this revert. Robert proposed restoring the original test in #87, Tom agreed in #88, and Robert confirmed completion in #92.

  • Tom Lanesaw the final versionLLM analysis, 100% sureTom explicitly endorsed reverting the test changes before the release because there was insufficient time to establish a reliable replacement #88.
Revise test case added in 43746996399541ecb5c7b188725a5f097c15ceae.
30 Jul 2022Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1REL_15_STABLE15beta3
LLM analysis, 99% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Implements Robert's proposal in #45, whose text attachment is not an indexed patchset: use safe_psql(), query the regression database after VACUUM FULL, add ORDER BY, and revise failure diagnostics. This is a follow-up to the test in #37, not a subset of that attachment.

  • Tom Lanesaw the final versionLLM analysis, 100% sureTom identified the unsafe psql invocation in #42 and approved Robert's replacement and accompanying test improvements in #46, matching the commit's explicit review credit.
Use TRUNCATE to preserve relfilenode for pg_largeobject + index.
28 Jul 2022Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1REL_15_STABLE15beta3
LLM analysis, 100% suremodified before commitlight changesfrom patchset #271 of 1 reviewer saw the final version

Combines both patches from #27 across tablecmds.c, md.c, relcache.c, and pg_dump.c. Comparing added and removed lines shows only the correction of 'pg_largebject' to 'pg_largeobject' in a relcache.c comment.

  • Bruce Momjiansaw the final versionLLM analysis, 96% sureBruce endorsed the IsBinaryUpgrade-based design in #29 and supported applying it to both master and PostgreSQL 15 in #32.

Jul 2022

1 commit
Preserve relfilenode of pg_largeobject and its index across pg_upgrade.
8 Jul 2022Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1REL_15_STABLE15beta3
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #123 of 3 reviewers saw the final version

Exactly retains the pg_dump.c and 002_pg_upgrade.pl changes from #12 while omitting all orphan-file cleanup changes in relfilenumber.c. Robert explicitly confirmed that scope reduction in #20.

  • Shruthi Gowdasaw the final versionLLM analysis, 100% sureShruthi reproduced the failure and verified that #4 fixed it in #6; the committed portion differs from that tested version only in its generated comment.
  • Bruce Momjiansaw the final versionLLM analysis, 93% sureBruce checked the system-versus-user relfilenode reasoning behind the retained fix and explained why preserving pg_largeobject's old relfilenode was safe in #19.
  • Robert Haassaw the final versioncommittercommitted the patch

Credits across 4 commits

Author
Justin PryzbyLLM1
Robert HaasLLM1
Reviewer
Robert Haascommitter1
Committer
Robert Haas4

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