Jun 2025

1 commit
Expand virtual generated columns for ALTER COLUMN TYPE
26 Jun 2025Richard Guomaster18beta2REL_19_STABLE18beta2REL_18_STABLE18beta2
LLM analysis, 100% surelanded as postedfrom patchset #1291 of 1 reviewer saw the final version

The tablecmds.c expansion and generated_virtual regression changes match #129 modulo rebasing; #131 confirms the push.

  • jian hesaw the final versiontrailerSuggested earlier expansion in ATPrepAlterColumnType() in #128 and explicitly approved the posted patch in #130.

Feb 2025

2 commits
Eliminate code duplication in replace_rte_variables callbacks
25 Feb 2025Richard Guomaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1231 of 1 reviewer saw the final version

Exactly matches the second component of #123, extracting ReplaceVarFromTargetList() across prepjointree.c and rewriteManip.c; the planner fix landed separately. The patch credits Dean Rasheed and Richard Guo, with Dean's shared-helper implementation originating in #114.

  • Jian Hesaw the final versiontrailerReviewed the shared helper's nesting-depth contract in #122, prompting documentation changes in #123 before the unchanged component was committed.
Expand virtual generated columns in the planner
25 Feb 2025Richard Guomaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1231 of 1 reviewer saw the final version

Applying the first component of #123 to the commit parent reproduces the committed files exactly, including planner expansion and generated_virtual tests; the second component landed separately. Richard Guo and Dean Rasheed are credited as joint authors.

  • Jian Hesaw the final versiontrailerTested grouping-set behavior in #116 and reviewed the near-final implementation and comments in #122, after which only small corrections and test additions followed.

Feb 2025

1 commit
Virtual generated columns
7 Feb 2025Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #951 of 2 reviewers saw the final version

Applying all three components of #95 reproduces the implementation except for whitespace and the catalog-version bump. The commit additionally tests virtual columns with REPLICA IDENTITY FULL in src/test/regress/sql/publication.sql and its expected output.

  • Jian Hesaw an earlier versiontrailerExtensively tested and supplied fixes through #80, but later dummy-RTE, RETURNING OLD/NEW, publication, and executor-validity changes were substantive.
  • Dean Rasheedsaw the final versiontrailerDirected the rewriting design, supplied RLS and RETURNING fixes, identified the validity-flag problem, and declared the final submission ready in #97.

Jan 2025

3 commits
Additional tests for stored generated columns
22 Jan 2025Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 98% surepart of a posted patchsetlight changesfrom patchset #85

Extracts the stored-column regression additions from #85, including whole-row, sublink, MERGE, privilege, domain, and composite-dependency cases, while omitting virtual-only tests. It adds RETURNING OLD/NEW coverage associated with #87 and further test adjustments.

Improve generated_stored test
21 Jan 2025Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surewritten from the discussion, no matching patch

The commit removes the gtest11s/gtest12s suffixes in src/test/regress/sql/generated_stored.sql and its expected output, implementing the separate, aligned test-file design proposed in #1. No attachment contains this renaming: #85 still uses the suffixed names, while #92 already has the cleanup in its base.

Improve generated_stored test
20 Jan 2025Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #851 of 1 reviewer saw the final version

Extracts the generated_stored catalog-sanity-check relocation present in #85, retaining the existing stored-only predicate because virtual columns had not yet landed. Jian proposed the relocation in #63, and Peter confirmed incorporating it in #72.

  • jian hesaw the final versionLLM analysis, 100% sureProposed moving the catalog sanity check after the tests in #63, a suggestion explicitly adopted in #72 and credited by the commit.

Dec 2024

1 commit
Fix handling of CREATE DOMAIN with GENERATED constraint syntax
3 Dec 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #541 of 1 reviewer saw the final version

The DefineDomain() change exactly matches #54: explicit GENERATED and IDENTITY diagnostics replace the switch default. Peter confirms committing it to master in #62.

  • jian hesaw the final versionLLM analysis, 94% sureReported the malformed CREATE DOMAIN diagnostic while testing #35 and identified DefineDomain() as the repair location in #42, directly leading to this small fix.

Aug 2024

3 commits
Put generated_stored test objects in a schema
29 Aug 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #172 of 2 reviewers saw the final version

Exactly matches the schema-isolation component of #17 in generated_stored.sql and its expected output; the feature and test-renaming components are separate changes.

  • Corey Huinkersaw the final versiontrailerEndorsed the separate stored/virtual test structure in #2, and this preparatory component remained unchanged through #17.
  • Tomasz Rybaksaw the final versiontrailerReviewed the schema-isolation patch and successfully ran incremental build and regression checks in #5, with no substantive subsequent change to this component.
Rename regress test generated to generated_stored
29 Aug 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #172 of 2 reviewers saw the final version

Matches the test-renaming component of #17: generated.sql and generated.out become generated_stored files, with the corresponding parallel_schedule edit.

  • Corey Huinkersaw the final versiontrailerExplicitly endorsed the stored/virtual test split in #2, and the renaming component did not materially change.
  • Tomasz Rybaksaw the final versiontrailerApproved the renaming as sensible preparation and reported successful build and regression testing in #5.
Small code simplification
21 Aug 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #171 of 1 reviewer saw the final version

Extracts the ATExecSetNotNull() attTup/GETSTRUCT cleanup embedded in #17, leaving out the virtual-column restriction that used it; remaining differences are formatting.

  • Tomasz Rybaksaw the final versionLLM analysis, 85% sureReported successful incremental builds and check-world runs for the complete series containing this cleanup in #5, although he had not yet performed a detailed review of the feature patch.

Jul 2024

3 commits
Remove useless code
1 Jul 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #31 of 1 reviewer saw the final version

Matches the complete BuildDescForRelation() cleanup component of #3, removing unused constraint construction and clarifying the function's limited purpose; the rest of the series remained pending.

  • Tomasz Rybaksaw the final versiontrailerInspected BuildDescForRelation() callers, confirmed that the removed constraint information was unused, and recommended applying the patch in #5.
Remove useless initializations
1 Jul 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #31 of 1 reviewer saw the final version

Matches the two-line relcache.c cleanup component of #3, removing redundant false assignments after MemoryContextAllocZero(); the other series components are excluded.

  • Tomasz Rybaksaw the final versiontrailerCompared neighboring MemoryContextAllocZero() usage and confirmed the cleanup's consistency in #5, alongside successful testing.
doc: Clarify that pg_attrdef also stores generation expressions
1 Jul 2024Peter Eisentrautmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Jian identified the missing pg_attrdef generation-expression documentation in #6, and Peter confirmed a separate documentation commit in #7. No attached patchset contains this catalogs.sgml change.

  • jian hesaw the final versionLLM analysis, 100% sureIdentified the exact pg_attrdef documentation omission in #6, which Peter corrected and acknowledged in #7.

Credits across 14 commits

Author
Peter Eisentraut8 LLM9
Richard Guo1 LLM3
Dean Rasheed2
Reviewer
jian he4
Tomasz Rybak4
Corey Huinker2
Dean Rasheed1
Tested by
Shlok Kyal1
Reported by
jian he2
Alexander Lakhin1
Co author
Dean Rasheed1
jian he1
Committer
Peter Eisentraut11
Richard Guo3

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