Apr 2026

1 commit
Doc: use "an SQL" consistently rather than "a SQL"
12 Apr 2026David Rowleymaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surewritten from the discussion, no matching patch

The commit applies the convention proposed in #1 to one occurrence in doc/src/sgml/xfunc.sgml, changing “build a SQL string” to “build an SQL string.” None of the three attachments contains this hunk, and the commit explicitly cites the established precedent and this discussion.

Apr 2024

1 commit
Doc: use "an SQL" instead of "a SQL"
9 Apr 2024David Rowleymaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surelanded as postedfrom patchset #30

The single doc/src/sgml/ddl.sgml hunk exactly matches #30, apart from git index hashes; David Rowley confirms pushing it in #31.

Apr 2023

1 commit
Doc: use "an SQL" consistently rather than "a SQL"
11 Apr 2023David Rowleymaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #281 of 1 reviewer saw the final version

All three doc/src/sgml/ecpg.sgml hunks exactly match #28, with only differing git index hash lengths. Michael Paquier endorsed the corrections in #29 before the commit.

  • Michael Paquiersaw the final versionLLM analysis, 99% sureMichael endorsed the three ECPG corrections with “Good catches!” before they were committed unchanged #29 #28.

Apr 2022

1 commit
Doc: use "an SQL" consistently rather than "a SQL"
20 Apr 2022David Rowleymaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch

The commit implements the convention established in #1 through four documentation corrections in doc/src/sgml/func.sgml, doc/src/sgml/ref/select.sgml and doc/src/sgml/xfunc.sgml, including occurrences hidden by markup. It landed in April 2022, before #28, and none of the attachments contains these corrections.

Jun 2021

1 commit
Use the correct article for abbreviations
11 Jun 2021David Rowleymaster14beta2REL_19_STABLE14beta2REL_18_STABLE14beta2REL_17_STABLE14beta2REL_16_STABLE14beta2REL_15_STABLE14beta2REL_14_STABLE14beta2
LLM analysis, 100% suremodified before commitlight changesfrom patchset #57 of 7 reviewers saw the final version

The commit retains all changes from #5, adding the SHA1 and SASLResponse wording corrections proposed in #8 and three matching pgbench test expectations in src/bin/pgbench/t/001_pgbench_with_server.pl. The resulting diff has 75 hunks across 35 files, versus 71 hunks across 33 files in the attachment.

  • Peter Eisentrautsaw the final versionLLM analysis, 94% surePeter supported the SQL convention and explained that pronounceable FIFO, SASL and SHA take “a,” guidance reflected in the attachment and final additions #2 #5 #8.
  • Dave Pagesaw the final versionLLM analysis, 98% sureDave proposed extending the cleanup to user-visible error messages while leaving code comments alone, which David implemented in the attachment #3 #5.
  • Daniel Gustafssonsaw the final versionLLM analysis, 78% sureDaniel endorsed the proposed consistency between documentation and user-facing messages before that scope was implemented in the attachment #4 #5.
  • Roberto Mellosaw the final versionLLM analysis, 78% sureRoberto gave a positive consistency-focused response after the initial patch submission, which subsequently received only light additions #6 #5.
  • Alvaro Herrerasaw the final versionLLM analysis, 94% sureAlvaro supplied a multiline search and potential missed occurrences, which David checked against the patch and its deliberate exclusion of code comments #7 #10.
  • Isaac Morlandsaw the final versionLLM analysis, 96% sureIsaac caught the incorrect proposed change to “a MIT” in the inline additions, and David acknowledged the correction before committing #8 #9 #21.
  • Tom Lanesaw the final versionLLM analysis, 93% sureTom reviewed the cleanup critically, raising backpatching costs and doubts about maintaining consistency, then withdrew his objection without endorsing the policy #11 #13 #15.

Credits across 5 commits

Author
David Rowley3 LLM4
Committer
David Rowley5

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