Mar - Apr 2020

7 commits
Fix RELCACHE_FORCE_RELEASE issue
11 Apr 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surewritten from the discussion, no matching patch

Matches Peter's inline CreateTupleDescCopy fix in #78, preserving tuple descriptors beyond RelationClose() in maybe_send_schema(). Peter explicitly confirms in #80 that he committed his patch without Amit's broader rearrangement from #79.

Allow publishing partition changes via ancestors
8 Apr 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #711 of 3 reviewers saw the final version

Matches the complete ancestor-publication patch in #71, with documentation and formatting edits, a catalog-version bump, and omission of the extra root CheckValidResultRel() call in nodeModifyTable.c.

  • Rafia Sabihsaw an earlier versiontrailerRafia reviewed the series in #27 and #36, before the later ancestor-selection and TRUNCATE logic changes.
  • Peter Eisentrautsaw the final versiontrailerPeter shaped the publication option, reviewed coverage and ancestor-selection code, supplied fixes in #69, and reviewed the final version as committer.
  • Petr Jelineksaw an earlier versiontrailerPetr participated in reviewing the posted series and its option naming in #54, before the final ancestor-selection rewrite.
Add logical replication support to replicate into partitioned tables
6 Apr 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #561 of 3 reviewers saw the final version

Commits the subscriber-routing component of #56, leaving ancestor publication pending. The implementation closely matches, with documentation and test rewrites, including omission of an UPDATE test later restored in #75.

  • Rafia Sabihsaw an earlier versiontrailerRafia reviewed tuple-routing naming and the broader series in #27 and #36, before substantial UPDATE and partition-map corrections.
  • Peter Eisentrautsaw the final versiontrailerPeter identified missing routing coverage and partition-map issues in #49 and #53, then reviewed the committed version as committer.
  • Petr Jelineksaw an earlier versiontrailerPetr examined the subscriber implementation and memory-context handling in #54, before the hash-table and UPDATE fixes in #56.
Refactor code to look up local replication tuple
1 Apr 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #521 of 1 reviewer saw the final version

Commits only the tuple-lookup refactoring from #52, which repeats that component of #51. Changes are limited to comment placement and moving slot creation slightly earlier within FindReplTupleInLocalRel().

  • Peter Eisentrautsaw the final versioncommittercommitted the patch
Some refactoring of logical/worker.c
24 Mar 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #481 of 2 reviewers saw the final version

Commits only the worker.c helper-extraction component of #48, with local variable renaming and comment cleanup; the two feature patches remained pending as confirmed in #49.

  • Rafia Sabihsaw an earlier versiontrailerRafia reviewed the series in #27 and #36, before helper placement and index-opening order were revised following #47.
  • Peter Eisentrautsaw the final versiontrailerPeter requested helper naming, placement, and index-opening corrections in #47 and reviewed the resulting patch as committer.
Prepare to support non-tables in publications
19 Mar 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #411 of 1 reviewer saw the final version

Uses Peter's extracted tablesync.c and logicalproto.h patch from #41, retaining explicit COPY column lists but removing the remote publishability query and error check after #42 and #44.

  • Amit Langotesaw the final versiontrailerAmit questioned the redundant error check in #42, examined the synchronization assertion in #43, and explicitly agreed to the final check removal in #45.
Support adding partitioned tables to publication
10 Mar 2020Peter Eisentrautmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #381 of 2 reviewers saw the final version

Commits the publication-membership component of #38, replacing integer partition-mode constants with PublicationPartOpt and polishing documentation and comments. Peter identifies this extraction and these edits in #39.

  • Rafia Sabihsaw an earlier versiontrailerRafia tested initial synchronization and reviewed membership behavior through #36, after which GetPublicationRelations() was reworked to implement the requested partition-expansion modes.
  • Peter Eisentrautsaw the final versiontrailerPeter drove the ancestor-lookup redesign and publication API simplification in #9, #30, and #35, then reviewed and committed the final component in #39.

Credits across 7 commits

Author
Amit Langote5
Peter EisentrautLLM1
Reviewer
Peter Eisentrautcommitter5
Rafia Sabih4
Petr Jelinek2
Amit Langote1
Committer
Peter Eisentraut7

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