Aug 2022

1 commit
MERGE docs adjustments
12 Aug 2022Alvaro Herreramaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1REL_15_STABLE15beta4Also in: MERGE and parsing with prepared statements
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

The mvcc.sgml change implements the wording proposed in #119; the merge.sgml sentence removal and example alignment are the additional changes enumerated in #120. No attached patchset contains these edits.

  • Justin Pryzbysaw the final versionLLM analysis, 100% sureQuestioned the concurrency wording in #117 and reviewed the replacement wording and remaining documentation edits in #120.

Jun 2022

1 commit
Link to MVCC docs in MERGE docs
15 Jun 2022Alvaro Herreramaster15beta2REL_19_STABLE15beta2REL_18_STABLE15beta2REL_17_STABLE15beta2REL_16_STABLE15beta2REL_15_STABLE15beta2
LLM analysis, 100% suremodified before commitlight changesfrom patchset #1152 of 2 reviewers saw the final version

Contains the documentation changes from #115, with the MERGE paragraph moved earlier in mvcc.sgml and an additional markup correction; #116 explicitly describes the paragraph move.

  • Justin Pryzbysaw the final versionLLM analysis, 99% sureProposed the missing cross-reference in #101 and reviewed its placement in #108, after which only documentation refinements followed.
  • Peter Eisentrautsaw the final versionLLM analysis, 99% sureSaid Álvaro's documentation patch looked acceptable in #112, before the subsequent wording and paragraph-placement refinements.

May 2022

2 commits
Fix EXPLAIN MERGE output when no tuples are processed
18 May 2022Alvaro Herreramaster15beta2REL_19_STABLE15beta2REL_18_STABLE15beta2REL_17_STABLE15beta2REL_16_STABLE15beta2REL_15_STABLE15beta2
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Responds directly to #109 by nesting the total > 0 check inside the text-format branch in explain.c and adding a zero-row regression case; #110 confirms the fix.

  • Tom Lanesaw the final versionLLM analysis, 100% sureIdentified the misplaced conditional and specified that zero-row text output should print no tuple summary in #109, before this corrective commit.
Make EXPLAIN MERGE output format more compact
18 May 2022Alvaro Herreramaster15beta2REL_19_STABLE15beta2REL_18_STABLE15beta2REL_17_STABLE15beta2REL_16_STABLE15beta2REL_15_STABLE15beta2
LLM analysis, 100% surewritten from the discussion, no matching patch2 of 2 reviewers saw the final version

Implements the compact EXPLAIN output discussed in #101 and #105, choosing lowercase key=value counters with zero elision instead of the inline patch's colon format; #106 confirms that choice.

  • Justin Pryzbysaw the final versionLLM analysis, 98% sureProposed compact tuple counts and zero elision in #101 and analyzed existing EXPLAIN formatting conventions in #105.
  • Laurenz Albesaw the final versionLLM analysis, 95% sureEndorsed the proposed compact output alternatives in #103 before the implementation was committed.

Mar 2022

3 commits
Add support for MERGE SQL command
28 Mar 2022Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1Also in: MERGE SQL Statement for PG11
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #970 of 8 reviewers saw the final version

Matches the final MERGE attachment #97 except for documentation formatting and a local executor rewrite passing tuple identity from ExecModifyTable into ExecMerge, including null-identity handling; #98 confirms that final change.

  • Peter Eisentrautsaw an earlier versiontrailerReviewed SQL-standard checks in #52 and supplied fixups in #93, before the counter, grammar and final tuple-identity changes.
  • Andres Freundsaw an earlier versiontrailerReviewed executor organization, concurrency, documentation and instrumentation in #60, prompting substantial subsequent restructuring.
  • Peter Geogheganno review foundtrailerThe commit credits earlier-version review, but Peter Geoghegan has no review messages in this supplied thread.
  • Robert Haasno review foundtrailerAsked whether objections from previous discussions had been addressed in #9, but the supplied thread shows no review of an attached version by him.
  • Japin Lisaw an earlier versiontrailerDiagnosed namespace visibility in #51, corrected an assertion in #56 and reviewed trigger/documentation details in #86, before later executor changes.
  • Justin Pryzbysaw an earlier versiontrailerSupplied language fixes, reported sqlsmith results in #34, exposed serialization failures in #62 and checked node ordering in #84, before later functional revisions.
  • Tomas Vondrasaw an earlier versiontrailerReviewed documentation, executor organization, rule handling and a Valgrind failure in #3, before the target-representation and executor redesigns.
  • Zhihong Yusaw an earlier versiontrailerReviewed assertions, control flow, memory-context handling and executor helper interfaces in #24, #26, #47, #61, #72, #77 and #80, before subsequent functional changes.
Split ExecUpdate and ExecDelete into reusable pieces
17 Mar 2022Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #890 of 3 reviewers saw the final version

Takes only the executor-refactoring member of #89, leaving MERGE for later; applying that member to the commit parent shows only comment/whitespace edits in nodeModifyTable.c, plus committed typedef-list additions.

  • Amit Langotesaw an earlier versiontrailerReviewed context fields, cross-partition state and helper responsibilities in #85 and #87, before update-result and epilogue logic changed in #88.
  • Japin Lisaw an earlier versiontrailerReviewed transition-capture handling in the executor split in #86, before the subsequent update-result and index-maintenance changes.
  • Zhihong Yusaw an earlier versiontrailerSuggested clearer slot naming and direct trigger-result returns in #80, before the later changes to update-result tracking and epilogue behavior.
Add API of sorts for transition table handling in trigger.c
11 Mar 2022Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1Also in: MERGE SQL Statement for PG11
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #781 of 1 reviewer saw the final version

Extracts the transition-table prerequisite from #78, whose header explicitly credits Pavan Deolasee, although he has no message in this thread. Comparison against the commit shows helper renaming, relocation and formatting in trigger.c without a behavioral redesign.

  • Álvaro Herrerasaw the final versiontrailerSeparated and revised the transition-table prerequisite through #78 and committed the final version, establishing final coverage.

Credits across 7 commits

Author
Alvaro Herrera1 LLM3
Pavan Deolasee2
Amit Langote1
Simon Riggs1
Reviewer
Japin Li2
Zhihong Yu2
Alvaro Herrera1
Amit Langote1
Andres Freund1
Justin Pryzby1
Peter Eisentraut1
Peter Geoghegan1
Robert Haas1
Tomas Vondra1
Committer
Alvaro Herrera7

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