Jan 2024

6 commits
Combine FSM updates for prune and no-prune cases.
26 Jan 2024Robert Haasmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #942 of 3 reviewers saw the final version

Matches the FSM consolidation logic in #94; differences are comments explaining cleanup-lock processing and missed FSM updates under failsafe or index-vacuum bypass. Robert explicitly confirms editing and committing that version in #101.

  • Andres Freundsaw an earlier versionLLM analysis, 90% sureIn #10, Andres requested a common buffer-release and FSM-update location, which Melanie implemented in #19 before subsequent control-flow revisions.
  • Peter Geoghegansaw the final versionLLM analysis, 97% surePeter examined FSM-vacuum constraints in #74 and #75 and clarified the failsafe comment in #95, with only light changes after the final attachment.
  • Robert Haassaw the final versioncommittercommitted the patch
Remove LVPagePruneState.
18 Jan 2024Robert Haasmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #811 of 1 reviewer saw the final version

Takes the LVPagePruneState-removal component of #81, including its typedefs.list deletion. Robert narrowed has_lpdead_items scope, removed its redundant initialization, and adjusted comments and line wrapping, as described in #90.

  • Robert Haassaw the final versioncommittercommitted the patch
Move VM update code from lazy_scan_heap() to lazy_scan_prune().
18 Jan 2024Robert Haasmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #811 of 2 reviewers saw the final version

Takes only the VM-relocation component of #81; its added and removed lines match exactly, moving visibility-map processing from lazy_scan_heap() into lazy_scan_prune().

  • Andres Freundsaw an earlier versionLLM analysis, 88% sureAndres proposed moving prune-specific processing into lazy_scan_prune() in #10, and Melanie supplied that restructuring in #19.
  • Robert Haassaw the final versioncommittercommitted the patch
Optimize vacuuming of relations with no indexes.
18 Jan 2024Robert Haasmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #811 of 3 reviewers saw the final version

Takes the first component of #81, with matching added and removed lines across pruneheap.c, vacuumlazy.c, and heapam.h; the remaining refactoring components landed separately.

  • Andres Freundsaw an earlier versionLLM analysis, 100% sureAndres reviewed branch costs, misleading replay parameters, and nested dead-item handling in #9, #10, and #16, before substantive validation and FSM revisions.
  • Peter Geoghegansaw an earlier versionLLM analysis, 97% surePeter corrected the FSM-update interpretation in #49 and #52 and examined vacuum cadence in #75, before the final condition change in #81.
  • Robert Haassaw the final versioncommittercommitted the patch
Be more consistent about whether to update the FSM while vacuuming.
16 Jan 2024Robert Haasmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 100% surelanded as postedfrom patchset #651 of 2 reviewers saw the final version

The single patch in #65 has exactly matching added and removed lines, including the has_lpdead_items output parameter and consistent do_index_vacuuming check; #67 announces its commit.

  • Peter Geoghegansaw an earlier versionLLM analysis, 100% surePeter identified the invalid prunestate.has_lpdead_items reference in #57, prompting the dedicated output parameter in the next revision.
  • Melanie Plagemansaw the final versionLLM analysis, 100% sureMelanie reviewed comments and the output-parameter design in #60, #62, and #64, then approved the committed version in #66.
Remove hastup from LVPagePruneState.
11 Jan 2024Robert Haasmaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #393 of 3 reviewers saw the final version

Extracts hastup removal from #39, updating LVRelState->nonempty_pages directly in both helpers. The only added/removed-line discrepancy is adapting local declarations because this component landed before the pruning patch.

  • Andres Freundsaw the final versionLLM analysis, 97% sureAndres questioned moving truncation bookkeeping in #10 and #16, then withdrew his objection in #35; the committed component retained that design with light revisions.
  • Michael Paquiersaw the final versionLLM analysis, 99% sureMichael checked caller coverage and requested a more accurate truncation comment in #28, following his favorable assessment in #6.
  • Robert Haassaw the final versioncommittercommitted the patch

Credits across 6 commits

Author
Melanie PlagemanLLM5
Robert HaasLLM1
Reviewer
Robert Haascommitter5
Committer
Robert Haas6

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