May 2022

1 commit
Repurpose PROC_COPYABLE_FLAGS as PROC_XMIN_FLAGS
19 May 2022Alvaro Herreramaster15beta2REL_19_STABLE15beta2REL_18_STABLE15beta2REL_17_STABLE15beta2REL_16_STABLE15beta2REL_15_STABLE15beta2REL_14_STABLE14.4REL_13_STABLE13.8
LLM analysis, 100% surelanded as postedfrom patchset #861 of 1 reviewer saw the final version

The canonical commit's procarray.c and proc.h changes exactly match #86, including an identical stable patch-id. The later submission #88 supplies stable-branch adaptations rather than a replacement development-branch patch.

  • Masahiko Sawadasaw the final versiontrailerMasahiko explained the original assertion and supported selective copying #85, then reviewed the revised mask's completeness #87; the committed implementation matches #86.

Apr 2022

3 commits
Tighten ComputeXidHorizons' handling of walsenders.
15 Apr 2022Tom Lanemaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #780 of 1 reviewer saw the final version

The commit retains the flag and restored-xmin exception from #78, but restricts flag setting in InitWalSender to MyDatabaseId == InvalidOid and adds an xmin assertion. It also replaces the XXX comments and reorganizes ComputeXidHorizons' equivalent tests, as announced in #80.

  • Andres Freundsaw an earlier versionLLM analysis, 96% sureAndres supplied the hot_standby_feedback rationale requested by #78, which Tom incorporated into the comments #79, but the committed version additionally restricts which walsenders set the flag.
Adjust VACUUM's removable cutoff log message.
15 Apr 2022Peter Geogheganmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch2 of 2 reviewers saw the final version

Tom proposed the replacement removable-cutoff wording inline #59, and Peter agreed to implement it #68. The commit changes that wording, capitalization and instrumentation placement in vacuumlazy.c; it does not implement the DEBUG1 attachment #27.

  • Tom Lanesaw the final versionLLM analysis, 100% sureTom identified the misleading interpretation, supplied the wording adopted by the commit, and questioned lowercase XID spelling #59.
  • Andres Freundsaw the final versionLLM analysis, 84% sureAndres reviewed the log-message semantics, distinguishing OldestXmin's age from relfrozenxid advancement during the discussion leading to the wording fix #72.
Temporarily add some probes of tenk1's relallvisible in create_index.sql.
14 Apr 2022Tom Lanemaster15beta1REL_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

Tom proposed probing tenk1's relallvisible in the opening report #1 and announced the temporary queries #23. The commit adds those probes to create_index.sql and its expected output before any patchset was attached.

  • David Rowleysaw the final versionLLM analysis, 80% sureDavid reported a similar relallvisible discrepancy and supported confirming relallvisible as the cause before pursuing a fix, the diagnostic approach implemented by the probes #4.

Credits across 4 commits

Author
Alvaro HerreraLLM1
Peter Geoghegan1
Tom LaneLLM1
Reviewer
Masahiko Sawada1
Committer
Tom Lane2
Alvaro Herrera1
Peter Geoghegan1

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