Jul 2023

1 commit
Fix oversight in handling of modifiedCols since f24523672d
2 Jul 2023Tomas Vondramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1REL_16_STABLE16beta3REL_15_STABLE15.4REL_14_STABLE14.9REL_13_STABLE13.12
LLM analysis, 100% suremodified before commitlight changesfrom patchset #301 of 1 reviewer saw the final version

The two trigger.c hunks match #30 in executable code, including afterTriggerCopyBitmap and the AfterTriggerSaveEvent call-site change. The committed version only expands the helper's explanatory comment.

  • Alexander Pyhalovsaw the final versionLLM analysis, 100% sureAfter reporting the AFTER UPDATE regression with a reproducer #28, he tested #30 and confirmed that it fixed the problem and looked good #31.

Jun 2023

1 commit
Use per-tuple context in ExecGetAllUpdatedCols
7 Jun 2023Tomas Vondramaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2REL_15_STABLE15.4REL_14_STABLE14.9REL_13_STABLE13.12+1 more
LLM analysis, 100% suremodified before commitlight changesfrom patchset #250 of 3 reviewers saw the final version

The execUtils.c change preserves the per-tuple allocation logic of #25, adding a comment requiring callers to copy the bitmap when a longer lifetime is needed and separating declaration from assignment. Tomas explicitly describes the added comment when announcing the commit #27.

  • Andres Freundsaw an earlier versiontrailerHe challenged explicit freeing #3 and reviewed the context-switch approach and cached-map lifetime hazards #18, before allocation control moved into ExecGetAllUpdatedCols #25.
  • Tom Lanesaw an earlier versiontrailerHe advocated shorter-lived contexts #2 and tested the PL/pgSQL allocation claim #14, before the series was reduced to the centralized bitmap fix #25.
  • Jakub Wartakno review foundtrailerHis contribution describes profiling the original leak and possible memory limits #10, but shows no review or testing of an attached patchset.

Credits across 2 commits

Author
Tomas VondraLLM2
Reviewer
Andres Freund1
Jakub Wartak1
Tom Lane1
Reported by
Alexander Pyhalov1
Tomas Vondra1
Committer
Tomas Vondra2

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