Apr 2026

3 commits
Update .abi-compliance-history for change to enum ProcSignalReason
16 Apr 2026Fujii MasaoREL_17_STABLE17.10
LLM analysis, 99% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Adds only a .abi-compliance-history exception for the increased NUM_PROCSIGNALS value; no matching attached patch is available. This follows the anticipated history update in #40 and the compatibility reasoning accepted in #42.

  • Nisha Moondsaw the final versionLLM analysis, 90% sureIn #41, Nisha checked core extensions for NUM_PROCSIGNALS use and explained why increasing its value was acceptable, supplying the rationale accepted in #42 and recorded by this exception.
Fix ABI break by moving PROCSIG_SLOTSYNC_MESSAGE in ProcSignalReason
9 Apr 2026Fujii MasaoREL_18_STABLE18.4REL_17_STABLE17.10
LLM analysis, 84% surelanded as postedfrom patchset #401 of 1 reviewer saw the final version

The PostgreSQL 18 commit moves PROCSIG_SLOTSYNC_MESSAGE to the end of ProcSignalReason and updates the NUM_PROCSIGNALS macro, matching the branch-specific repair described in #39 and submitted in #40. Raw attachments are unavailable because extraction failed, preventing exact textual verification; the subsequent revision discussed in #42 concerns PostgreSQL 17.

  • Nisha Moondsaw the final versiontrailerIn #41, Nisha reviewed enum ordering and NUM_PROCSIGNALS compatibility, checked extension usage, and suggested retaining NUM_PROCSIGNALS inside the PostgreSQL 17 enum.
Fix slotsync worker blocking promotion when stuck in wait
8 Apr 2026Fujii Masaomaster19beta1REL_19_STABLE19beta1REL_18_STABLE18.4REL_17_STABLE17.10
LLM analysis, 100% surelanded as postedfrom patchset #342 of 4 reviewers saw the final version

The five-file implementation matches #34 modulo rebased context and blank-line placement in postgres.c; the added and removed substantive lines are unchanged. Both the submitted patch header and commit credit Nisha Moond, whose implementation first appears in #13, although Fujii sent the final submission.

  • shveta maliksaw an earlier versiontrailerIn #14, #22, and #27, Shveta reviewed redundant handling, interrupt flags, comments, and late errors; the subsequent IsSyncingReplicationSlots() guard changed behavior.
  • Amit Kapilasaw the final versiontrailerAmit challenged backend termination in #4, proposed the procsignal design in #12, refined interrupt handling in #17 and #18, and assessed final backpatch safety in #35 and #37.
  • Zhijie Housaw an earlier versiontrailerIn #26, Zhijie identified unnecessary errors after synchronization finishes, prompting the behavioral guard added in #30, without a subsequent review of its implementation.
  • Fujii Masaosaw the final versiontrailerFujii reviewed and tested successive versions, requested naming and logging refinements, approved the final code in #34, and committed it as confirmed in #38.

Credits across 3 commits

Author
Fujii Masao1
Nisha Moond1
Reviewer
Amit Kapila1
Fujii Masao1
Nisha Moond1
shveta malik1
Zhijie Hou (Fujitsu)1
Committer
Fujii Masao3