Oct 2021

5 commits
Initialize variable to placate compiler.
25 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Nathan reports the compiler warning and proposes initializing oldXLogAllowed to zero in #183; the commit makes precisely that one-line change in CreateCheckPoint, and Robert confirms it in #184. No attached patch contains this fix.

  • Nathan Bossartsaw the final versionLLM analysis, 100% sureReported the uninitialized-variable warning and proposed the exact initialization adopted by the discussion-derived fix in #183.
StartupXLOG: Don't repeatedly disable/enable local xlog insertion.
25 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #1811 of 1 reviewer saw the final version

Combines the LocalSetXLogInsertAllowed return-value change and toggle removals from the first two components of #181, retaining save/restore only in CreateCheckPoint rather than StartupXLOG. Robert explicitly reports reordering and editing those components in #182.

  • Robert Haassaw the final versioncommittercommitted the patch
StartupXLOG: Call CleanupAfterArchiveRecovery after XLogReportParameters.
25 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1811 of 1 reviewer saw the final version

Takes the CleanupAfterArchiveRecovery call relocation from the second component of #181, placing it after XLogReportParameters in src/backend/access/transam/xlog.c. The other local-WAL-permission changes were committed separately, as confirmed in #182.

  • Robert Haassaw the final versioncommittercommitted the patch
Postpone some end-of-recovery operations related to allowing WAL.
14 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1770 of 2 reviewers saw the final version

Matches the second component of #177, postponing end-of-recovery WAL operations in StartupXLOG, with only an expanded comment explaining lastReplayedEndRecPtr validity. The patch and commit credit Robert Haas and Amul Sul, and #178 confirms the landing.

  • Andres Freundsaw an earlier versionLLM analysis, 86% sureConfirmed the deferred-startup-WAL design proposed for #83 in #85, with the specific operation movement recorded in #84 before substantial restructuring.
  • Dilip Kumarsaw an earlier versionLLM analysis, 93% sureReviewed startup WAL-operation grouping in #88 and #97, questioning promotion-specific work and insisting that only necessary operations be deferred before later restructuring.
Refactor some end-of-recovery code out of StartupXLOG().
13 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1770 of 2 reviewers saw the final version

The first component of #177 matches the committed extraction of PerformRecoveryXLogAction and CleanupAfterArchiveRecovery modulo patch offsets and whitespace. Only that component of the four-patch series landed here, with Robert Haas and Amul Sul credited as authors.

  • Andres Freundsaw an earlier versionLLM analysis, 80% sureConfirmed the startup-WAL extraction design for the earlier series in #85, as detailed in #84, before Robert divided it into independently committable helpers.
  • Dilip Kumarsaw an earlier versionLLM analysis, 90% sureExamined which recovery and promotion operations belonged in the extracted startup code in #88 and #97, before the helper structure was substantially rewritten.

Jan 2021

1 commit
Allow for error or refusal while absorbing a ProcSignalBarrier.
18 Jan 2021Robert Haasmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #781 of 1 reviewer saw the final version

Matches the barrier prerequisite in #78 modulo whitespace, including volatile flags, bit scanning, and retry restoration in procsignal.c; the test module and WAL-prohibition feature were excluded. Robert authored the prerequisite in #67, and the commit also credits forwarding sender Amul's review in #68.

  • Andres Freundsaw the final versionLLM analysis, 99% sureRequested volatile flags, bit scanning, reset handling, and race explanations in #60, then approved the revised loop in #74, after which the committed barrier component changed only lightly.

Credits across 6 commits

Author
Robert HaasLLM3
Amul SulLLM2
Reviewer
Robert Haascommitter2
Committer
Robert Haas6

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