Nov 2021

1 commit
Restore lock level to set vacuum flags
11 Nov 2021Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1REL_14_STABLE14.2
LLM analysis, 100% surewritten from the discussion, no matching patch

Andres identified the omitted ReplicationSlotRelease lock restoration in #62, and Alvaro agreed to fix it in #63; the commit makes that exact LW_SHARED-to-LW_EXCLUSIVE change in src/backend/replication/slot.c without an attached patch.

Nov 2020

6 commits
Restore lock level to update statusFlags
26 Nov 2020Alvaro Herreramaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 99% surelanded as postedfrom patchset #55

Matches #55 in all functional changes: exclusive flag-update locks, vacuum snapshot ordering and proc.h documentation; the only additional edit rewraps an existing procarray.c comment.

Avoid spurious waits in concurrent indexing
25 Nov 2020Alvaro Herreramaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #290 of 1 reviewer saw the final version

Lands the CREATE INDEX portion of #29, leaving its REINDEX flag-setting extension out; set_indexsafe_procflags() changes from LW_SHARED to LW_EXCLUSIVE and receives comment edits.

  • Michael Paquiersaw an earlier versiontrailerMichael reviewed safety, helper extraction, transaction context, assertions and naming in #8, #11, #25 and #28, then advocated exclusive locking in #33, before the final lock-mode change.
Centralize logic for skipping useless ereport/elog calls.
24 Nov 2020Tom Lanemaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #441 of 1 reviewer saw the final version

Combines Tom's draft #43 with Alvaro's later #44, including its ProcSleep guard and client-policy factoring; the final elog.c implementation adds inline policy helpers, an explicit ERROR check and pg_re_throw cleanup, with both authors credited.

  • Tom Lanesaw the final versioncommittercommitted the patch
Don't hold ProcArrayLock longer than needed in rare cases
23 Nov 2020Alvaro Herreramaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #380 of 1 reviewer saw the final version

Takes the shortened ProcSleep critical section from #38, but retains LW_EXCLUSIVE and omits its direct message-level guard after Tom's objection in #39; logging suppression lands separately.

  • Michael Paquiersaw an earlier versiontrailerMichael reviewed the debugging-message treatment and proposed skipping unnecessary formatting in #34 and #35, before the broader ProcSleep restructuring in #38.
Relax lock level for setting PGPROC->statusFlags
18 Nov 2020Alvaro Herreramaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #271 of 1 reviewer saw the final version

Lands the lock-relaxation component of #27, adding locking documentation in proc.h, a stronger ProcArrayEndTransactionInternal assertion and a stale comment-name correction.

  • Michael Paquiersaw the final versiontrailerMichael requested explicit documentation of shared-lock safety in #28, which was added before committing, with only comments and diagnostic assertion changes beyond the reviewed patch.
Rename PGPROC->vacuumFlags to statusFlags
16 Nov 2020Alvaro Herreramaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1Also in: Proposal: GetOldestXminExtend for ignoring arbitrary vacuum flags
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #201 of 1 reviewer saw the final version

Lands the nine-file vacuumFlags-to-statusFlags prerequisite from #20, with small proc.h comment and formatting edits; this November 2020 attachment is later than every supplied patchset in thread #36026, which supplied earlier naming discussion.

  • Álvaro Herrerasaw the final versiontrailerAlvaro announced review with intent to push in #22 and committed the renamed fields, establishing final coverage as the committer.

Credits across 7 commits

Author
Alvaro Herrera3 LLM5
Dmitry Dolgov2
Reviewer
Michael Paquier3
Alvaro Herrera1
Tom Lanecommitter1
Committer
Alvaro Herrera6
Tom Lane1

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