Apr 2022

1 commit
Truncate line pointer array during heap pruning.
7 Apr 2022Peter Geogheganmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #361 of 4 reviewers saw the final version

Commits the PageRepairFragmentation() truncation from #36, but omits zero-pointer truncation in PageTruncateLinePointerArray(), removes clobbering, and replaces the maxoff recalculation with an explanatory comment. Peter explicitly lists these omissions and revisions in #40; the diff also renames lastUsed and adds an assertion.

  • Tom Lanesaw an earlier versionLLM analysis, 95% sureIdentified stale-TID access risks in #5 and required the bounds-check audit supplied in #11, well before the subsequent implementation and scope changes.
  • Simon Riggssaw an earlier versionLLM analysis, 98% sureSupported pruning-time truncation, requested separating zero-pointer truncation, and examined WAL replay implications in #19, #21, and #31, before later instrumentation and maxoff revisions.
  • Robert Haassaw an earlier versionLLM analysis, 86% sureReviewed the performance tradeoff in #27, supporting inexpensive pruning-time reclamation piggybacked on existing WAL records before the later implementation revisions.
  • Peter Geoghegansaw the final versioncommittercommitted the patch

Apr 2021

1 commit
Truncate line pointer array during VACUUM.
7 Apr 2021Peter Geogheganmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1Also in: New IndexAM API controlling index vacuum strategies (part of a posted patchset)

Credits across 2 commits

Author
Matthias van de Meent2
Peter Geoghegan1
Reviewer
Peter Geoghegancommitter2
Masahiko Sawada1
Committer
Peter Geoghegan2