Feb 2019

1 commit
Fix memory leak when inserting tuple at relation creation for CTAS
27 Feb 2019Michael Paquiermaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surelanded as postedfrom patchset #441 of 1 reviewer saw the final version

The createas.c diff exactly matches #44, adding the comment and heap_freetuple(tuple) before intorel_receive() returns; Michael Paquier confirms committing the fix in #45.

  • Michael Paquiersaw the final versioncommittercommitted the patch

Nov 2018

1 commit
Build HashState's hashkeys expression with the correct parent.
7 Nov 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surewritten from the discussion, no matching patch

Creates separate rhclauses initialized with the HashState parent in nodeHashjoin.c; no available attachment contains this change. It addresses the Hash/HashJoin slot mismatch investigated in #39, before the revised slot inference in #41.

Oct 2018

2 commits
Move TupleTableSlots boolean member into one flag variable.
16 Oct 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 82% surepart of a posted patchsetmoderate changesfrom patchset #172 of 2 reviewers saw the final version

Extracts the flag consolidation introduced in #1 and the LLVM slow-flag load/OR/store changes in #17, the latest matching attachment. The standalone commit retains the original slot structure, adds TTS_FLAG_* constants and read macros, and credits Ashutosh Bapat jointly with Andres Freund.

  • Kyotaro Horiguchisaw the final versionLLM analysis, 86% sureReviewed the archived flag patch in #30, recommending open-coded setters as used in the commit while also discussing the unchosen bitfield alternative.
  • Tom Lanesaw the final versionLLM analysis, 88% sureChallenged the posted flag conversion's masking cost and size benefit in #31, prompting Andres's benchmark-based response in #36; the reviewed archive has no indexed patchset.
Move generic slot support functions from heaptuple.c into execTuples.c.
15 Oct 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 97% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Moves slot_getattr(), slot_getsomeattrs(), and slot_getmissingattrs() into execTuples.c while retaining heap-specific deforming in heaptuple.c. This standalone extraction is requested in #21 and supplied in the archive described by #32; the available prototype instead replaces accessors with callback wrappers.

  • Andres Freundsaw the final versioncommittercommitted the patch

Sep 2018

4 commits
Remove absolete function TupleDescGetSlot().
25 Sep 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 95% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Deletes TupleDescGetSlot() from execTuples.c, matching the standalone removal described in #5 and its push announcement in #27. The available prototype #1 only disables the function with a preprocessor guard, and the later available series already assumes its removal.

  • Andres Freundsaw the final versiontrailerReviewed and committed Ashutosh's archived removal patch, announcing its push in #27; as committer he covers the final version.
Remove function list from prologue of execTuples.c.
25 Sep 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Removes the stale interface-routine inventory from execTuples.c and relocates its file header. Andres requested the separate cleanup in #21, Ashutosh supplied it in the archive described by #22, and Andres announced an extended committed version in #27; no indexed attachment contains that cleanup.

  • Andres Freundsaw the final versioncommittercommitted the patch
Split ExecStoreTuple into ExecStoreHeapTuple and ExecStoreBufferHeapTuple.
25 Sep 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Splits ExecStoreTuple() into heap and buffer-heap entry points and converts their callers, implementing the proposal explicitly left undone in #1. The actual patch evolved through archived submissions, including the buffer-unpin correction in #22, and was pushed with minor polishing in #27.

  • Andres Freundsaw the final versiontrailerFound the missing release of an existing buffer pin in #21, obtained the correction in #22, and committed the polished archived patch as announced in #27.
Change TupleTableSlot->tts_nvalid to type AttrNumber.
25 Sep 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 93% suremodified before commitmoderate changesfrom patchset #61 of 1 reviewer saw the final version

Reworks the LLVM attribute-width correction from #6: it uses l_int16_const() instead of introducing TypeAttrNumber/l_attrnumber_const(), also fixes switch-case constants, and changes tts_nvalid to AttrNumber. #15 reports the crash resolved, and #27 confirms this preparation was committed separately from accidentally combined changes.

  • Andres Freundsaw the final versioncommittercommitted the patch

Aug 2018

1 commit
Deduplicate code between slot_getallattrs() and slot_getsomeattrs().
23 Aug 2018Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 91% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Replaces slot_getallattrs() with an inline call to slot_getsomeattrs(), retaining the existing non-abstract slot implementation. The commit credits Ashutosh's series, but no indexed attachment contains this standalone implementation: #1 instead calls the new getsomeattrs callback directly, and #41 already has the committed wrapper as context.

  • Andres Freundsaw the final versiontrailerThe commit explicitly credits Andres's review of Ashutosh's cleanup, and his role as committer establishes final coverage despite the absence of its archived patch from the indexed submissions.

Credits across 9 commits

Author
Ashutosh Bapat7
Andres Freund2
Jeff Janes1
Reviewer
Andres Freundcommitter6
Michael Paquiercommitter1
Committer
Andres Freund8
Michael Paquier1