Dec 2023

3 commits
Add asserts to bimapset manipulation functions
27 Dec 2023Alexander Korotkovmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surelanded as postedfrom patchset #102 of 4 reviewers saw the final version

The complete diff exactly matches the first patch in #10, adding Bitmapset NodeTag assertions throughout src/backend/nodes/bitmapset.c.

  • Richard Guono review foundtrailerRichard endorsed reallocation and supplied the SJE fix in #9, but no review of the NodeTag assertions appears in the thread.
  • Andres Freundno review foundtrailerAndres discussed forced reallocation and the SJE bug in #3, but no review of the added NodeTag assertions appears in the thread.
  • Ashutosh Bapatsaw the final versiontrailerAshutosh reviewed the interleaved assertions and reallocation changes in #8 and requested the split subsequently delivered in #10.
  • Andrei Lepikhovsaw the final versiontrailerAndrei explicitly said the first patch looked good and could be applied in #15.
REALLOCATE_BITMAPSETS manual compile-time option
27 Dec 2023Alexander Korotkovmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surelanded as postedfrom patchset #103 of 4 reviewers saw the final version

The complete diff exactly matches the second patch in #10, adding REALLOCATE_BITMAPSETS instrumentation in src/backend/nodes/bitmapset.c and its switch in src/include/pg_config_manual.h.

  • Richard Guosaw the final versiontrailerRichard endorsed the proposed reallocation mode in #9, after #7, and its subsequent separation into #10 preserved the implementation.
  • Andres Freundno review foundtrailerAndres proposed and prototyped forced reallocation in #3, but the thread contains no review by him of Alexander's attached implementation.
  • Ashutosh Bapatsaw the final versiontrailerAshutosh examined the combined instrumentation in #7 and requested its separation from assertions in #8, which #10 implemented.
  • Andrei Lepikhovsaw the final versiontrailerAndrei reviewed the second patch in #15, accepting its purpose while raising memory concerns and suggesting separate consideration from the hotfix.
Make replace_relid() leave argument unmodified
27 Dec 2023Alexander Korotkovmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surelanded as postedfrom patchset #105 of 5 reviewers saw the final version

Matches the third patch in #10: all changed lines in analyzejoins.c, join.sql and join.out are identical, with only rebased context and offsets. That patch reproduces Richard's implementation and test from #9, which Alexander explicitly acknowledged when incorporating it.

  • Richard Guosaw the final versiontrailerRichard supplied the corrected implementation in #9, which was incorporated unchanged into #10, so his involvement was authorship rather than independent review.
  • Andres Freundsaw the final versiontrailerAndres exposed the missing deletion-path copy through testing in #3 and assessed the final copying approach's memory and maintenance implications in #13.
  • Ashutosh Bapatsaw the final versiontrailerAshutosh challenged the final patchset's memory overhead and unnecessary copying in #12, prompting the measurements reported in #22.
  • Andrei Lepikhovsaw the final versiontrailerAndrei endorsed the third patch as the needed, lower-risk SJE fix in #15.
  • Alexander Korotkovsaw the final versioncommittercommitted the patch

Credits across 3 commits

Author
Alexander KorotkovLLM2
Richard GuoLLM1
Reviewer
Andrei Lepikhov3
Andres Freund3
Ashutosh Bapat3
Richard Guo3
Alexander Korotkovcommitter1
Reported by
Richard Guo1
Committer
Alexander Korotkov3

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