Dec 2021

2 commits
Some RELKIND macro refactoring
3 Dec 2021Peter Eisentrautmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 98% surepart of a posted patchsetlight changesfrom patchset #82 of 2 reviewers saw the final version

Commits the macro-refactoring portion of #8, excluding its proposed removal of pg_dump checks. Beyond rebase context, the remaining changes are a defensive RELKIND_HAS_STORAGE branch with Assert(false) and use of OidIsValid in src/backend/catalog/heap.c, as requested in #11.

  • Michael Paquiersaw the final versiontrailerIn #9 he supported the macro refactoring while requesting preservation of storage safety checks, and in #14 he accepted the proposed adjustment.
  • Alvaro Herrerasaw the final versiontrailerHe proposed table-AM-first handling in #2 and supplied the final storage guard and OidIsValid correction while judging the patch positively in #11.
pg_dump: Add missing relkind case
2 Dec 2021Peter Eisentrautmaster15beta1REL_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

Adds RELKIND_MATVIEW to guessConstraintInheritance() in src/bin/pg_dump/pg_dump.c, whereas the attached series proposed deleting that check. Peter proposed the replacement in #10, Michael endorsed it in #14, and Peter confirmed its commit in #15.

  • Michael Paquiersaw the final versiontrailerHe argued for retaining the explicit pg_dump checks in #9 and endorsed adding the missing materialized-view case in #14.

Aug 2021

1 commit
Fix handling of partitioned index in RelationGetNumberOfBlocksInFork()
26 Aug 2021Peter Eisentrautmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surelanded as postedfrom patchset #42 of 2 reviewers saw the final version

The two hunks in src/backend/storage/buffer/bufmgr.c exactly match #4, moving RELKIND_PARTITIONED_INDEX from the smgrnblocks cases to the non-storage assertion cases.

  • Michael Paquiersaw the final versiontrailerHe confirmed the correction in #5 and discussed expressing the invariant with RELKIND_HAS_STORAGE.
  • Alvaro Herrerasaw the final versiontrailerHe confirmed the mistake and reported finding no obvious caller passing a partitioned index during his call-site inspection in #6.

Credits across 3 commits

Author
Peter EisentrautLLM2
Reviewer
Michael Paquier3
Alvaro Herrera2
Committer
Peter Eisentraut3

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