Oct 2024

1 commit
Doc: add detail about EXPLAIN's "Disabled" property
29 Oct 2024David Rowleymaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #1601 of 1 reviewer saw the final version

The perform.sgml changes exactly match #160. The commit credits Laurenz Albe and David Rowley as authors, consistent with Laurenz's inline precursor in #128.

  • Alvaro Herrerasaw the final versionLLM analysis, 83% sureHe endorsed Laurenz's inline documentation proposal in #143; the subsequent attached versions refined its wording and example.

Oct 2024

1 commit
Adjust EXPLAIN's output for disabled nodes
11 Oct 2024David Rowleymaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #1522 of 2 reviewers saw the final version

The implementation and existing test changes match #152, with the two collate.icu.utf8.out expectation updates supplied by Alena Rybakina in #153 folded in.

  • Laurenz Albesaw the final versiontrailerHe said the initial patch looked good in #109 and accepted the revised boolean presentation in #123, after which only polishing and test fixes followed.
  • Alena Rybakinasaw the final versiontrailerShe inspected special-child handling in #138, tested the revision, and supplied missing ICU expectations for the final patch in #153.

Sep 2024

1 commit
Fix order of parameters in a cost_sort call
9 Sep 2024Richard Guomaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Alexander Lakhin's UBSan report in #101 identifies reversed input_disabled_nodes and input_cost arguments in label_sort_with_costsize; the commit makes that exact argument-order correction and credits him as author.

  • Richard Guosaw the final versioncommittercommitted the patch

Aug 2024

2 commits
Show number of disabled nodes in EXPLAIN ANALYZE output.
21 Aug 2024Robert Haasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #752 of 3 reviewers saw the final version

The changed lines exactly match the EXPLAIN component of #75, including Plan.disabled_nodes propagation and expected output. The companion planner-cost component landed separately.

  • Andres Freundsaw the final versionLLM analysis, 99% sureThe commit explicitly credits him as a reviewer, although his visible review in #55 concerns the companion planner-cost change.
  • Heikki Linnakangassaw an earlier versionLLM analysis, 98% sureHe questioned COSTS OFF behavior and proposed per-node markers in #60, before later fixes added missing Plan count propagation.
  • David Rowleysaw the final versionLLM analysis, 100% sureHe explicitly reviewed both components in #74 and advocated marking individual disabled nodes; the EXPLAIN component then remained unchanged through #75.
Treat number of disabled nodes in a path as a separate cost metric.
21 Aug 2024Robert Haasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #751 of 3 reviewers saw the final version

The changed lines exactly match the planner-cost component of #75, covering disabled_nodes propagation, path comparisons, join prechecks, FDWs, and tests. The EXPLAIN component was committed separately.

  • Andres Freundsaw an earlier versionLLM analysis, 100% sureHe benchmarked the added fuzzy-cost comparison branch in #55, before subsequent propagation and join-count corrections.
  • Heikki Linnakangassaw an earlier versionLLM analysis, 100% sureHe reviewed forced-TID counting, branch annotations, debugging remnants, and comments in #60, before later logic fixes.
  • David Rowleysaw the final versionLLM analysis, 100% sureHis detailed review in #74 drove propagation and redundant-count fixes, and he continued examining the final revision's join prechecks in #76 and #78.

Jul 2024

1 commit
Remove grotty use of disable_cost for TID scan plans.
22 Jul 2024Robert Haasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #591 of 1 reviewer saw the final version

The changed lines exactly match the forced-TID component of #59: create_tidscan_paths signals that only its CURRENT OF path is valid, and costsize.c drops the penalty trick. The other three series components are absent.

  • Heikki Linnakangassaw the final versionLLM analysis, 100% sureHe explicitly called the TID patch ready for commit in #60, and the commit credits his review.

Credits across 6 commits

Author
Robert HaasLLM3
David Rowley1 LLM2
Alexander Lakhin1
Laurenz Albe1
Reviewer
Alena Rybakina1
Laurenz Albe1
Richard Guocommitter1
Committer
Robert Haas3
David Rowley2
Richard Guo1

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