Oct 2023

3 commits
windows: msvc: Define STDIN/OUT/ERR_FILENO.
17 Oct 2023Nathan BossartREL_14_STABLE14.10REL_13_STABLE13.13REL_12_STABLE12.17REL_11_STABLE11.22
LLM analysis, 100% surewritten from the discussion, no matching patch2 of 2 reviewers saw the final version

No attachment changes src/include/port/win32_msvc/unistd.h; this commit implements the additional backpatch proposed after Windows failures #27 and approved #28. Its message explicitly credits Andres Freund as author of the reused definitions.

  • Tom Lanesaw the final versionLLM analysis, 100% sureApproved extending the existing Windows descriptor-definition fix to older branches before the commit #28.
  • Nathan Bossartsaw the final versioncommittercommitted the patch
Avoid calling proc_exit() in processes forked by system().
17 Oct 2023Nathan Bossartmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1REL_16_STABLE16.1REL_15_STABLE15.5REL_14_STABLE14.10REL_13_STABLE13.13+2 moreAlso in: Weird failure with latches in curculio on v15 (not this thread)
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #211 of 2 reviewers saw the final version

Takes the complete second component of #21 unchanged apart from git index metadata: the startup.c child guard, PANIC checks in ipc.c and proc.c, and write_stderr_signal_safe() in elog.c and elog.h. The attribution in thread #47353 to #528038 is superseded by this thread's matching October submission, which includes the subsequent diagnostic-helper and PANIC revisions.

  • Thomas Munrosaw an earlier versiontrailerReproduced the failure and verified the initial fix #2, before the signal-safe helper and production PANIC checks were added.
  • Andres Freundsaw the final versiontrailerReviewed PID checks and signal-safe output #5 #8 #10, requested PANIC checks #20, and addressed the resulting signal-safety question after the final submission #23.
Move extra code out of the Pre/PostRestoreCommand() section.
16 Oct 2023Nathan Bossartmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1REL_16_STABLE16.1REL_15_STABLE15.5Also in: Weird failure with latches in curculio on v15 (not this thread)
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #212 of 4 reviewers saw the final version

Takes the complete first component of #21, moving fflush() and wait-event reporting outside the Pre/PostRestoreCommand() section in src/backend/access/transam/xlogarchive.c, with only comment reflow. The attribution in thread #47353 to #528038 is superseded by this thread's matching October submission, which is later than that February source.

  • Michael Paquierno review foundtrailerDiscussed whether unsafe signal-handler calls remained #24, but this thread contains no identifiable review of the xlogarchive.c relocation.
  • Robert Haasno review foundtrailerThe commit credits Robert Haas, but he has no messages or identifiable review activity in this thread.
  • Thomas Munrosaw the final versiontrailerTested the initial submitted series #2, whose xlogarchive.c component subsequently changed only in comment formatting.
  • Andres Freundsaw the final versiontrailerGave an LGTM while reviewing the pending series #20, and its xlogarchive.c component reached the commit with only comment reflow.

Credits across 3 commits

Author
Nathan BossartLLM2
Andres Freund1
Reviewer
Andres Freund2
Thomas Munro2
Michael Paquier1
Nathan Bossartcommitter1
Robert Haas1
Reported by
Michael Paquier1
Sandeep Thakkar1
Thomas Munro1
Committer
Nathan Bossart3

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