May 2021

2 commits
Be more verbose when the postmaster unexpectedly quits.
23 May 2021Tom Lanemaster14beta2REL_19_STABLE14beta2REL_18_STABLE14beta2REL_17_STABLE14beta2REL_16_STABLE14beta2REL_15_STABLE14beta2REL_14_STABLE14beta2
LLM analysis, 100% surelanded as postedfrom patchset #22

All three hunks in src/backend/postmaster/postmaster.c match #22; the patch and commit differ only in index-hash abbreviation.

Fix access to no-longer-open relcache entry in logical-rep worker.
23 May 2021Tom Lanemaster14beta2REL_19_STABLE14beta2REL_18_STABLE14beta2REL_17_STABLE14beta2REL_16_STABLE14beta2REL_15_STABLE14beta2REL_14_STABLE14beta2REL_13_STABLE13.4
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #200 of 3 reviewers saw the final version

Preserves the worker.c implementation from #20, with declaration relocation and explanatory comment changes. It additionally adds Assert(!RelationHasReferenceCountZero(rel)) in src/backend/commands/trigger.c, a diagnostic logic change absent from the attachment.

  • Amit Langotesaw an earlier versionLLM analysis, 95% sureTested the initial fix under CLOBBER_CACHE_ALWAYS in #3, advocated explicit ResultRelInfo arguments in #15, and discussed backpatch prerequisites for #20 in #23, before the additional trigger assertion was committed.
  • Michael Paquiersaw an earlier versionLLM analysis, 93% sureReviewed state-field alternatives and snapshot handling in #13 and assessed the debugging value of #17 in #18, before the active-relation API was revised.
  • Amit Kapilasaw an earlier versionLLM analysis, 96% sureFlagged the maintenance risk of mutable activeRelInfo in #19, a concern addressed by the explicit relation arguments in #20.

Credits across 2 commits

Author
Tom LaneLLM2
Committer
Tom Lane2

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