Sep 2022

2 commits
walmethods.c/h: Make WalWriteMethod more object-oriented.
19 Sep 2022Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #13 of 3 reviewers saw the final version

ebfb814f7c takes the complete second patch of the two-patch series in #1, introducing WalWriteMethodOps, per-instance state and callback-based cleanup. Differences are rebase adjustments for the preceding filename fix in src/bin/pg_basebackup/receivelog.c and existing Z_DEFAULT_STRATEGY changes in walmethods.c, consistent with #5.

  • Peter Eisentrautsaw the final versionLLM analysis, 88% surePeter endorsed the proposed direction in #2, and the WalWriteMethod patch from #1 subsequently received only rebase adjustments.
  • velagandula sravan kumarsaw the final versionLLM analysis, 92% sureAfter a quick examination, Sravan said the refactoring in #1 made sense in #3, with only rebase adjustments subsequently made to this component.
  • Kyotaro Horiguchisaw the final versionLLM analysis, 95% sureKyotaro endorsed the design in #4 and discussed the remaining abstraction problems, while the WalWriteMethod component of #1 landed with only rebase adjustments.
walmethods.c/h: Make Walfile a struct, rather than a void *
19 Sep 2022Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #10 of 3 reviewers saw the final version

4fd1479494 takes the first patch of the two-patch series in #1, making Walfile a structure and removing get_current_pos(). Beyond that submitted component, src/bin/pg_basebackup/receivelog.c copies the pathname before closing and clearing walfile so archive marking can safely use it, and corrects the open_walfile comment; #5 confirms pre-commit fixes.

  • Peter Eisentrautsaw an earlier versionLLM analysis, 88% surePeter endorsed the direction of #1 in #2, before the unposted close_walfile logic fix confirmed in #5.
  • velagandula sravan kumarsaw an earlier versionLLM analysis, 97% sureSravan explicitly supported preventing Walfile pointer-type mistakes and reported examining the patch in #3, before the pre-commit logic fix described in #5.
  • Kyotaro Horiguchisaw an earlier versionLLM analysis, 95% sureKyotaro endorsed the refactoring in #4, before the unposted filename-lifetime fix to the Walfile component of #1.

Credits across 2 commits

Author
Robert HaasLLM2
Committer
Robert Haas2

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