Apr - May 2024

7 commits
Revert structural changes to not-null constraints
13 May 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1reverts 8 commits here
LLM analysis, 99% suremodified before commitlight changesfrom patchset #142

Implements the revert attached in #142; comparison of resulting files shows preservation of unrelated foreach modernizations, typechecked list access, domain documentation, and formatting.

Fix inconsistencies in error messages
9 May 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1Also in: pg17 issues with not-null contraints (written from the discussion, no matching patch)
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #139

The commit takes the first patch of #139 unchanged, correcting error wording in pg_constraint.c, tablecmds.c and inherit.out. It combines that contribution with version-comment corrections in src/bin/pg_dump/common.c and pg_dump.c originating in thread #49465, consistent with that thread's from_discussion verdict for its portion.

Disallow direct change of NO INHERIT of not-null constraints
2 May 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1341 of 1 reviewer saw the final version

Commits the first patch of #134, preserving the backend implementation while expanding inheritance null-validation tests and clarifying comments.

  • Alexander Lakhinsaw the final versionLLM analysis, 96% sureReported the missing-lock assertion in #129, examined the revised behavior in #132, and acknowledged the supplied fixes before this commit in #135.
Disallow NO INHERIT not-null constraints on partitioned tables
2 May 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1340 of 1 reviewer saw the final version

Matches the second patch of #134, rejecting NO INHERIT constraints on partitioned tables and omitting their unnecessary pg_dump temporary constraints; only comment wrapping differs.

  • Alexander Lakhinsaw an earlier versionLLM analysis, 95% sureFound the partition-detach failure while examining #131 in #132, before the new prohibition was implemented in #134.
Fix restore of not-null constraints with inheritance
18 Apr 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1Also in: pg17 issues with not-null contraints (modified before commit)
ATTACH PARTITION: Don't match a PK with a UNIQUE constraint
15 Apr 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #125

Takes the second patch of #125, but changes get_constraint_type() to error on a missing catalog tuple instead of returning zero and adds regression coverage.

Fix propagating attnotnull in multiple inheritance
15 Apr 2024Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1251 of 3 reviewers saw the final version

Takes the first patch of #125, retaining its CommandCounterIncrement placement in set_attnotnull while revising comments and extending and renaming regression cases.

  • Alexander Lakhinsaw an earlier versionLLM analysis, 91% sureReported a concrete multiple-inheritance self-update failure in the landed implementation in #119, prompting the later CCI fix.
  • Michael Paquiersaw an earlier versionLLM analysis, 87% sureDiagnosed the reported failure as a missing CommandCounterIncrement in #120, which became the fix in #122.
  • jian hesaw the final versionLLM analysis, 97% sureExamined the CCI change and reported regression behavior with an alternative placement in #126, although that placement was rejected.

Nov 2023

1 commit
Check stack depth in new recursive functions
8 Nov 2023Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surewritten from the discussion, no matching patch0 of 1 reviewer saw the final version

Implements the stack-depth protection requested in #116 without an attached patch; #117 explains adding explicit checks to four recursive functions in tablecmds.c.

  • Alexander Lakhinsaw an earlier versionLLM analysis, 100% sureSupplied deep-inheritance crash reproducers and requested check_stack_depth calls in #116, well before the thread's final patchset.

Sep 2023

2 commits
Update information_schema definition for not-null constraints
7 Sep 2023Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1Also in: information_schema and not-null constraints (modified before commit)
Fix not-null constraint test
1 Sep 2023Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #1100 of 1 reviewer saw the final version

Revises #110 by replacing its !HeapTupleIsValid condition with HeapTupleIsValid && !connoinherit and adding two regression cases; #112 explicitly describes the further correction.

  • Alexander Lakhinsaw an earlier versionLLM analysis, 100% sureTested #110 and confirmed that it stopped the crash in #111, before the committed condition was corrected further.

Jul 2023

1 commit
parallel_schedule: add comment on event_trigger test dependency
12 Jul 2023Alvaro Herreramaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #631 of 1 reviewer saw the final version

The raw attachment in #63 contains the exact committed src/test/regress/parallel_schedule comment hunk, despite the similarity scores missing that match; #64 subsequently removes it.

  • Peter Eisentrautsaw the final versiontrailerExplicitly called the schedule change correct and requested a separate commit in #60; that hunk remained unchanged through #63.

Apr 2023

4 commits
Revert "Catalog NOT NULL constraints" and fallout
12 Apr 2023Alvaro Herreramaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1reverts 1 commit here
LLM analysis, 100% surewritten from the discussion, no matching patch0 of 2 reviewers saw the final version

No revert attachment was posted; #54 proposed reverting because of upgrade failures and restore scans, and #55 accepted that course.

  • Tom Lanesaw an earlier versionLLM analysis, 100% sureExplained the upgrade and restore-performance regressions and recommended reverting in #54, before the later reimplementation.
  • Andres Freundsaw an earlier versionLLM analysis, 95% sureReported the cross-version upgrade regression in #52 that led to the revert discussion.
Adjust contrib/sepgsql regression test expected outputs.
7 Apr 2023Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surewritten from the discussion, no matching patch0 of 1 reviewer saw the final version

Tom updated contrib/sepgsql expected output after Andres reported buildfarm fallout in #47; #48 explains accepting the changed catalog-access logging without an attached patch.

  • Andres Freundsaw an earlier versionLLM analysis, 97% sureIdentified the sepgsql regression failure in #47 before the output correction was committed.
Fix table name clash in recently introduced test
7 Apr 2023Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surewritten from the discussion, no matching patch0 of 1 reviewer saw the final version

Andres diagnosed the parallel inherit.sql/triggers.sql table-name collision in #43 and announced the unattached child-to-inh_child rename in #45.

  • Justin Pryzbysaw an earlier versionLLM analysis, 94% sureFlagged generic regression-table names in #38 and reiterated test problems in #41, anticipating the collision fixed here.
Catalog NOT NULL constraints
7 Apr 2023Alvaro Herreramaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1Also in: Make PG's "NOT NULL"s and attnotnull ("is_nullable") conform to SQL-2011Also in: cataloguing NOT NULL constraintsAlso in: cataloguing NOT NULL constraintsAlso in: starting to review the Extend NOT NULL representation to pg_constraint patchAlso in: ALTER TABLE...ALTER COLUMN vs inheritance
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #400 of 2 reviewers saw the final version

Commits #40 with additional changes to AddRelationNotNullConstraints list processing, StoreRelNotNull's inheritance-count type, NOT NULL verification paths, and regression expectations.

  • Justin Pryzbysaw an earlier versiontrailerReviewed display, APIs, resources, and tests in #29, #38, and #41, but further backend logic changes followed before commit.
  • Peter Eisentrautsaw an earlier versiontrailerReviewed representation, syntax, display, and tests in #15, #30, and #33, before the later NO INHERIT and pg_dump implementation.

Credits across 15 commits

Author
Alvaro Herrera8 LLM10
Bernd Helmle1
Reviewer
Peter Eisentraut2
Justin Pryzby1
Reported by
Alexander Lakhin4
Andrew Bille1
Justin Pryzby1
Co author
Alvaro Herrera1
Peter Eisentraut1
Committer
Alvaro Herrera13
Andres Freund1
Tom Lane1

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