Nov 2022

1 commit
Fix initialization of pg_stat_get_lastscan()
8 Nov 2022Michael Paquiermaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #411 of 3 reviewers saw the final version

The pgstatfuncs.c fix and both regression-test hunks exactly match #41. Michael Paquier forwarded the revised fix, while the commit credits Dave Page as author and #42 confirms its application.

  • Robert Treatsaw an earlier versionLLM analysis, 97% sureReported reproducible epoch timestamps for unscanned relations in #38, prompting the subsequent accessor correction; no review of the final fix is recorded.
  • Peter Eisentrautsaw an earlier versionLLM analysis, 86% sureDave credited his diagnosis of the committed implementation's bug in #40, corroborated by the Analyzed-by credit in d7744d50a5, before the accessor was corrected.
  • Michael Paquiersaw the final versioncommittercommitted the patch

Oct 2022

2 commits
pgstat: Track time of the last scan of a relation
14 Oct 2022Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #341 of 3 reviewers saw the final version

Combines the feature and regression-test components of #34, leaving its xact.c prerequisite to a separate commit. Comparing added and removed lines shows only the catalog-version date changed within those components.

  • Andres Freundsaw the final versiontrailerChallenged timestamp overhead, developed the stop-time approach, revised and tested #34, and committed the final implementation as confirmed in #36.
  • Bruce Momjiansaw an earlier versiontrailerQuestioned alternatives and timestamp costs, endorsed stop-time semantics in #17, and advocated removing the GUC in #19, before subsequent logic changes.
  • Vik Fearingsaw an earlier versiontrailerInvestigated timestamp availability with Dave as reported in #23 and approved #26 in #27, before #34 changed the accessor's zero-timestamp behavior.
Have GetCurrentTransactionStopTimestamp() set xactStopTimestamp if unset
14 Oct 2022Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #342 of 2 reviewers saw the final version

Exactly matches the xact.c prerequisite component of #34, including timestamp caching, updated callers, comments, and the transaction-state assertion. The feature and test components of that submission were committed separately.

  • Andres Freundsaw the final versiontrailerRequested a separate prerequisite commit in #28, added explanatory comments and an assertion in #34, and committed the result.
  • Vik Fearingsaw the final versiontrailerApproved the timestamp behavior in #27, after which the prerequisite received only comments, a defensive assertion, and separation into its own patch.

Credits across 3 commits

Author
Dave Page3
Reviewer
Andres Freund2
Vik Fearing2
Bruce Momjian1
Michael Paquiercommitter1
Reported by
Robert Treat1
Committer
Andres Freund2
Michael Paquier1