Mar - Apr 2026

3 commits
Give an 'options' parameter to tuple_delete/_update
1 Apr 2026Álvaro Herreramaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #143 of 4 reviewers saw the final version

db89a47115 takes the complete update/delete API component of #14, excluding its caller-type repair and flag cleanup. Its six-file diff matches that component apart from retaining a blank line in src/include/access/tableam.h.

  • Nathan Bossartsaw an earlier versiontrailerNathan read and endorsed #1 in #2, before the API was revised to retain a separate wait boolean and reposition the options argument.
  • Zsolt Parragisaw the final versiontrailerZsolt identified a missed caller and stale comments in #7, then reviewed #11 in #12, prompting documentation and unused-parameter annotation fixes.
  • Chao Lisaw the final versiontrailerChao reviewed #11 in #13 and requested update/delete parameter documentation corrections that were incorporated in #14.
  • Antonin Houskasaw the final versiontrailerAntonin explicitly reported reviewing the series #14 in #15, raising only a question about the separate insert-flag cleanup.
Fix vicinity of tuple_insert to use uint32, not int, for options
1 Apr 2026Álvaro Herreramaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #142 of 2 reviewers saw the final version

ec2f81766a takes only the preliminary caller-type repair from #14. Its 13-file diff has the same stable patch ID as that attachment, covering TOAST helpers, heap buffer allocation, COPY, and other insert callers.

  • Chao Lisaw the final versionLLM analysis, 99% sureWhile reviewing #11, Chao identified remaining signed options in RelationGetBufferForTuple and raw_heap_insert in #13, directly prompting the mechanical repair in #14.
  • Antonin Houskasaw the final versionLLM analysis, 98% sureAntonin explicitly reviewed the series #14 in #15 before this component was committed, with his only question concerning another component.
Have table_insert and siblings use an unsigned type for options
30 Mar 2026Álvaro Herreramaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #92 of 2 reviewers saw the final version

1bd6f22f43 extracts the insert-side int-to-bits32 conversion and private-bit documentation from #9. Applying this commit followed by the update/delete patch in #10 produces exactly the same six files as the first patch in #9.

  • Nathan Bossartsaw the final versionLLM analysis, 87% sureAfter reviewing #1 in #2, Nathan explicitly accepted changing insert options to bits32 in #4, the type conversion subsequently committed.
  • Andres Freundsaw the final versionLLM analysis, 95% sureResponding to #1, Andres requested unsigned option types in #5 and specifically endorsed making table_tuple_insert options unsigned in #6, which #9 implemented.

Credits across 3 commits

Author
Alvaro Herrera1 LLM3
Reviewer
Antonin Houska1
Chao Li1
Nathan Bossart1
Zsolt Parragi1
Reported by
Chao Li1
Committer
Alvaro Herrera3

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