Nov 2026

6 commits
Return NULL for unknown pgstattuple estimate
19 Nov 2026Melanie Plagemanmasterunreleased
LLM analysis, 100% surelanded as postedfrom patchset #16

b443bc7406 exactly matches the four-file diff attached in #16, including the unknown-count flag, NULL result, regression expectation and documentation. Lukas Fittl's support in #15 preceded this patch and is credited as Suggested-by, with no subsequent qualifying review.

Fix pgstattuple_approx handling of unknown tuple estimates
18 Nov 2026Melanie PlagemanmasterunreleasedREL_19_STABLEunreleasedREL_18_STABLEunreleasedREL_17_STABLEunreleasedREL_16_STABLEunreleasedREL_15_STABLEunreleased
LLM analysis, 100% surelanded as postedfrom patchset #102 of 2 reviewers saw the final version

8c15c2cb5c exactly matches the four-file diff in #10, clamping the double estimate before unsigned assignment and adding the regression test and zero-result documentation. Melanie confirmed committing and backpatching it in #14.

  • Lukas Fittlsaw the final versiontrailerIn #11, Lukas reviewed #10, endorsed its documentation and explained why returning NULL on stable branches could break existing clients.
  • Zsolt Parragisaw the final versiontrailerIn #12, Zsolt approved #10 and compared its clamp with the double-valued clamp in vacuumlazy.c, explaining why that other use should remain unchanged.
Exclude new pages from VACUUM reltuples estimates
13 Nov 2026Melanie PlagemanmasterunreleasedREL_19_STABLE19.0
LLM analysis, 100% suremodified before commitlight changesfrom patchset #31 of 1 reviewer saw the final version

206e71935e preserves all implementation and test changes from #3, removing only the stale negative-return comment in src/backend/commands/vacuum.c and revising the commit message. The sender, Melanie Plageman, is the commit author, with Vaibhav Jain also credited as co-author.

  • Zsolt Parragisaw the final versionLLM analysis, 98% sureZsolt examined the estimator and pgstattuple changes in #4 and identified the stale comment in #7, whose removal was the only code-diff adjustment between #3 and the commit.
Scale VACUUM reltuples estimates after relation growth
12 Nov 2026Melanie PlagemanmasterunreleasedREL_19_STABLE19.0
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #21 of 1 reviewer saw the final version

eacabd7d03 extracts the unchanged relation-size guard and accompanying comment from the first patch in #2, touching only src/backend/commands/vacuum.c. The commit explicitly credits Vaibhav Jain as author.

  • Melanie Plagemansaw the final versiontrailerMelanie reviewed the reported failure modes in #1, prepared the revised split in #2, and committed the final change.
Avoid extrapolating reltuples from tiny VACUUM samples
12 Nov 2026Melanie PlagemanmasterunreleasedREL_19_STABLE19.0
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #21 of 1 reviewer saw the final version

1a001d930e extracts the unknown-density tiny-sample guard from the first patch in #2 and the vacuumlazy.c explanatory comment from its second patch, with only wording and whitespace adjustments. It leaves out the other fixes and tests and explicitly credits Vaibhav Jain as author.

  • Melanie Plagemansaw the final versiontrailerMelanie endorsed returning unknown reltuples while reviewing #1, supplied the revised implementation in #2, and committed the final change.
Preserve n_live_tup when VACUUM cannot estimate it
12 Nov 2026Melanie PlagemanmasterunreleasedREL_19_STABLE19.0
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #21 of 1 reviewer saw the final version

49931d8fdc combines the pgstat_relation.c conditional update from the first patch in #2 with the vacuumlazy.c removal of the zero clamp from its second patch. It adds Assert(livetuples >= -1), omits the remaining fixes and tests, and explicitly credits Vaibhav Jain as author.

  • Melanie Plagemansaw the final versiontrailerMelanie reviewed the statistics-preservation proposal in #1, incorporated it into #2, and committed the final change.

Credits across 6 commits

Author
Vaibhav Jain3
Melanie Plageman1 LLM2
Reviewer
Melanie Plageman3
Lukas Fittl1
Zsolt Parragi1
Reported by
Vaibhav Jain4
Zsolt Parragi1
Co author
Melanie Plageman1
Vaibhav Jain1
Committer
Melanie Plageman6

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