Mar 2026

1 commit
In pg_dumpall, don't skip role GRANTs with dangling grantor OIDs.
2 Mar 2026Tom Lanemaster19beta1REL_19_STABLE19beta1REL_18_STABLE18.4REL_17_STABLE17.10REL_16_STABLE16.14
LLM analysis, 100% suremodified before commitlight changesfrom patchset #261 of 1 reviewer saw the final version

The src/bin/pg_dump/pg_dumpall.c changes match #26, with dump_grantor renamed to dump_this_grantor and adjustments for the target branch's allocation and output APIs. Robert requested clearer naming in #27 and #29, and Tom confirmed committing that change in #30.

  • Robert Haassaw the final versionLLM analysis, 100% sureRobert said the patch looked good on a quick read and requested clearer variable naming in #27, with his preference confirmed in #29 and adopted in #30.

Feb 2025

2 commits
Avoid race condition between "GRANT role" and "DROP ROLE".
21 Feb 2025Tom Lanemaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #71 of 1 reviewer saw the final version

This commits the backend, documentation, and regression-test portion of #7, leaving its separate pg_dumpall patch to another commit. The substantive edit makes CreateRole's CommandCounterIncrement unconditional, as discussed in #11 and accepted in #12.

  • Laurenz Albesaw the final versiontrailerLaurenz reviewed the series in #8, questioned the command-counter change, and explicitly accepted the final unconditional-call simplification in #12.
Fix pg_dumpall to cope with dangling OIDs in pg_auth_members.
21 Feb 2025Tom Lanemaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1REL_17_STABLE17.5REL_16_STABLE16.9REL_15_STABLE15.13REL_14_STABLE14.18REL_13_STABLE13.21
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #70 of 2 reviewers saw the final version

This commits only the pg_dumpall portion of #7, adapted to the supplied commit's simpler dumpRoleMembership loop with positional columns and per-row checks for missing roles and members. Unlike the attachment, it retains the older handling of missing grantors and changes the diagnostic wording from dangling to orphaned.

  • Laurenz Albesaw an earlier versionLLM analysis, 95% sureLaurenz said both patches looked fine in #8, before the supplied commit's branch-specific loop and grantor-handling adaptations.
  • Alvaro Herrerasaw an earlier versionLLM analysis, 90% sureAlvaro proposed suppressing invalid GRANT output in pg_dumpall in #5, which Tom implemented with explicit warnings in #7, before the supplied commit's further adaptations.

Credits across 3 commits

Author
Tom Lane2 LLM3
Reviewer
Laurenz Albe1
Reported by
Virender Singla2
Robert Haas1
Committer
Tom Lane3

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