Aug 2019

1 commit
Add default_table_access_method to postgresql.conf.sample.
16 Aug 2019Andres Freundmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1+1 more
LLM analysis, 100% surelanded as postedfrom patchset #1771 of 1 reviewer saw the final version

The single added default_table_access_method setting in src/backend/utils/misc/postgresql.conf.sample exactly matches #177.

  • Andres Freundsaw the final versioncommittercommitted the patch

May 2019

3 commits
Don't to predicate lock for analyze scans, refactor scan option passing.
19 May 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: Segfault on ANALYZE in SERIALIZABLE isolationAlso in: sample scans and predicate locking
LLM analysis, 96% surewritten from the discussion, no matching patch

The scan-option bitmask follows the API discussion in #147, #148, and #150; no attached patch implements it. The ANALYZE crash and sample-locking motivations also come from thread #40612 and thread #40615, neither of which has an attached source.

tableam: Avoid relying on relation size to determine validity of tids.
18 May 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #1660 of 1 reviewer saw the final version

Takes the TID-validity component of #166, adding table_tuple_tid_valid and replacing relation-size checks in nodeTidscan.c; the final code moves lazy scan initialization to TidListEval and uses table_rescan.

  • Ashwin Agrawalsaw an earlier versiontrailerReviewed both patches and questioned scan startup in #169 and #171; the landed code subsequently changed initialization and rescan handling.
tableam: Don't assume that every AM uses md.c style storage.
18 May 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1661 of 1 reviewer saw the final version

Takes the relation-size component of #166, routing non-heap size requests through the AM and supporting InvalidForkNumber for total size; differences are comment wording and formatting.

  • Ashwin Agrawalsaw the final versionLLM analysis, 92% sureCritiqued relation-size and sampling assumptions in #156 and examined both revised patches in #169, after which this component received only light edits.

Apr 2019

1 commit
Allow pg_class xid & multixid horizons to not be set.
24 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #151

Extends the horizon-assertion fix in #151 with its proposed relation_copy_for_cluster pointer arguments, allowing CLUSTER/VACUUM FULL to preserve invalid horizons.

Feb - Apr 2019

30 commits
tableam: comment and formatting fixes.
8 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1431 of 1 reviewer saw the final version

Commits only the tableam.h comment-correction component of #143, with pgindent reflow; the toy AM and scan-slot fixes are separate.

  • Andres Freundsaw the final versioncommittercommitted the patch
table: docs: fix typos and grammar.
5 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% suremodified before commitlight changesfrom patchset #1411 of 1 reviewer saw the final version

Uses Justin’s documentation patch in #141, omitting some wording edits and completing the unfinished shared-buffer/page-layout sentence.

  • Andres Freundsaw the final versioncommittercommitted the patch
tableam: Add table_multi_insert() and revamp/speed-up COPY FROM buffering.
4 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: COPY FROM WHEN condition (modified before commit)
LLM analysis, 98% surepart of a posted patchsetsignificant changesfrom patchset #130

Uses the table_multi_insert and slot-based COPY conversion from #130, combined with David Rowley’s buffering work from thread #39488. The landed copy.c implementation buffers multiple partitions and substantially exceeds this thread’s attached conversion.

tableam: basic documentation.
4 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #1400 of 2 reviewers saw the final version

Builds on Andres’s revision of Haribabu’s documentation in #140, adding generic-WAL/transaction guidance and clarifying storage.sgml’s scope beyond that draft.

  • Amit Langotesaw an earlier versionLLM analysis, 94% sureReviewed the documentation and corresponding API behavior in #51, before the later documentation restructuring.
  • Kyotaro Horiguchisaw an earlier versionLLM analysis, 98% sureReviewed API descriptions and syntax wording in #56, #57, and #122, with corrections incorporated before the chapter was restructured.
Only allow heap in a number of contrib modules.
1 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #130

Takes the contrib-restrictions component of #130, preserving heap-only checks in pgrowlocks/pgstattuple/pageinspect and adding further guard coverage.

tableam: Add table_finish_bulk_insert().
1 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetlight changesfrom patchset #130

Takes the finish_bulk_insert component of #130; the same callback replaces heap_sync, with TABLE_INSERT flag and ti_options naming cleanup plus expanded comments.

tableam: sample scan.
1 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #130

Uses the sample-scan component of #130, refining visibility checks, tuple accounting, and scan-loop/offset handling beneath tableam.

tableam: bitmap table scan.
1 Apr 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #1300 of 1 reviewer saw the final version

Uses the bitmap-scan component of #130, with revised skip-fetch/empty-tuple handling, callback documentation, and prefetch arrangement in nodeBitmapHeapscan.c.

  • Robert Haasno review foundtrailerThe commit credits an older review, but no identifiable bitmap-patch review by Robert appears in this thread; #149 is after the landing.
tableam: Move heap specific logic from estimate_rel_size below tableam.
31 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetlight changesfrom patchset #130

Takes planner size estimation from #130, preserving the heap callback while polishing its wrapper/comments and placing the existing block-size query in the index branch.

tableam: VACUUM and ANALYZE support.
31 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #130

Takes VACUUM/ANALYZE support from #130, changing the next-block callback to report whether a block should be sampled and clearing tuple slots appropriately.

tableam: relation creation, VACUUM FULL/CLUSTER, SET TABLESPACE.
29 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #130

Takes relation creation, rewrite, and tablespace support from #130, integrating CLUSTER progress reporting and adapting heap/table scan usage.

tableam: Support for an index build's initial table scan(s).
28 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetlight changesfrom patchset #130

Takes index-build scans from #130, with TableScanDesc/table_slot_create adaptations, a public validation wrapper, and comment cleanup.

Compute XID horizon for page level index vacuum on primary.
26 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: Minimal logical decoding on standbys (not this thread)Also in: Computing the conflict xid for index page-level-vacuum on primary (not this thread)
LLM analysis, 97% surepart of a posted patchsetmoderate changesfrom patchset #1300 of 1 reviewer saw the final version

Combines the horizon and table-AM dispatch components of #130 with Robert’s optimized prefetch version #96, removing obsolete WAL relfilenodes as explained in #132. This supersedes the older source claimed by thread #39814 and the original proposal in thread #39825.

  • Robert Haassaw an earlier versiontrailerReviewed prefetch state and heuristics in #92 and #94, then merged and polished the code in #96 before later table-AM/WAL integration.
tableam: Add table_get_latest_tid, to wrap heap_get_latest_tid.
26 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetlight changesfrom patchset #130

Takes table_get_latest_tid from #130, retaining the callback/wrapper and WHERE CURRENT OF integration with comment, formatting, and dependent heap_fetch API adjustments.

tableam: Add helper for indexes to check if a corresponding table tuples exist.
25 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #130

Takes the index-to-table existence-check component of #130, polishing the API and using table_slot_create with explicit table_index_fetch_begin state.

tableam: Add and use table_fetch_row_version().
25 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #1300 of 1 reviewer saw the final version

Takes fetch_row_version from #130, also removing the unused stats_relation parameter from heap_fetch and refining slot table-OID handling and API comments.

  • Haribabu Kommino review foundtrailerNo specific pre-commit review of the fetch_row_version component is identifiable here; Haribabu’s modification review in #125 concerns another component.
tableam: Use in CREATE TABLE AS and CREATE MATERIALIZED VIEW.
25 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #1300 of 1 reviewer saw the final version

Takes CREATE TABLE AS/materialized-view support from #130, dropping its extra target-slot allocation/copy/drop and passing the existing input slot directly to table insertion.

  • Haribabu Kommino review foundtrailerHaribabu supplied earlier CREATE AS materialization fixes in #40 and #41, but no identifiable review of the committed CTAS conversion appears here.
tableam: Add tuple_{insert, delete, update, lock} and use.
24 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: What to name the current heap after pluggable storage / what to rename? (not this thread)Also in: Pluggable storage
LLM analysis, 97% surepart of a posted patchsetlight changesfrom patchset #1300 of 1 reviewer saw the final version

The modification/EPQ component of #130 is the latest matching version; #131 describes the remaining edits as cosmetic. Although the attachment timestamp follows the recorded commit time slightly, its content and announcement establish the match; sources in thread #39860 and thread #35077 are older.

  • Dmitry Dolgovsaw an earlier versionLLM analysis, 91% sureTested and confirmed the concurrent-update invisible-tuple fix in #45, before the modification and EPQ paths were substantially reworked.
Ensure sufficient alignment for ParallelTableScanDescData in BTShared.
11 Mar 2019Andres 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 patch

The 32-bit alignment failure is diagnosed in #120 and fixed in #121 by aligned allocation of the parallel scan descriptor within BTShared storage; no matching attachment exists.

tableam: Add and use scan APIs.
11 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: Pluggable storage
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1190 of 1 reviewer saw the final version

The scan-API component of #119 matches the landed TableScanDesc, parallel-scan, index-fetch, and slot interfaces with only comment/whitespace cleanup; it is later than the sources in thread #35077.

  • Ashwin Agrawalsaw an earlier versionLLM analysis, 96% sureTesting in #102 exposed missing-callback crashes and prompted mandatory-callback assertions, before subsequent scan-interface consolidation.
Don't reuse slots between root and partition in ON CONFLICT ... UPDATE.
6 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surelanded as postedfrom patchset #1062 of 2 reviewers saw the final version

The changes to execPartition.c, nodeModifyTable.c, and execnodes.h exactly match #106 after whitespace normalization.

  • Robert Haassaw the final versiontrailerAndres acknowledges Robert’s review when posting the cleaned ON CONFLICT version in #106, after only light changes to #105.
  • David Rowleysaw the final versiontrailerDavid explicitly reports reading the cleaned patch in #107, before the unchanged version was committed.
tableam: Add pg_dump support.
6 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: Inheriting table AMs for partitioned tables
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1050 of 1 reviewer saw the final version

The pg_dump component of #105 matches exactly, incorporating Dmitry’s work from #85; thread #40222 has no attached source.

  • Amit Khandekarsaw an earlier versionLLM analysis, 99% sureFound archive compatibility, versioning, and buffer-lifetime issues in #66 and #71, before the integrated dump patch gained further changes and tests.
tableam: introduce table AM infrastructure.
6 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: Pluggable storageAlso in: Displaying and dumping of table access methodsAlso in: Inheriting table AMs for partitioned tables
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1050 of 1 reviewer saw the final version

Takes the DDL/catalog/psql infrastructure component of #105; only the catalog-version date and whitespace differ. That attachment is later than thread #35077, while thread #39944 and thread #40222 provide discussion but no competing patchsets.

  • Amit Khandekarsaw an earlier versionLLM analysis, 96% sureReviewed psql display, regression setup, and HIDE_TABLEAM semantics in #75 and #79, which were incorporated into the later infrastructure series.
Use a virtual rather than a heap slot in two places where that suffices.
2 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #78

Selects the virtual-slot changes for execPartition.c and replication/logical/worker.c from the broader slot-type-fixes component of #78, without changing those hunks.

Store tuples for EvalPlanQual in slots, rather than as HeapTuples.
1 Mar 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #78

Uses the EPQ slotification component of #78, polishing slot lifecycle and row-lock handling and converting RefetchForeignRow to fill the caller-provided slot.

Allow buffer tuple table slots to materialize after ExecStoreVirtualTuple().
28 Feb 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: pgsql: Use slots in trigger infrastructure, except for the actual invoc
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #78

Uses the buffer-slot virtualization component of #78, expanding materialization assertions, buffer-release handling, and comments; the redis_fdw discussion in thread #212443 has no competing attached patch.

Use slots in trigger infrastructure, except for the actual invocation.
27 Feb 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #78

Uses the trigger-slotification component of #78, refining lazy per-ResultRelInfo slots and callers while retaining the external HeapTuple trigger interface, as explained in #101.

Store table oid and tuple's tid in tuple slots directly.
27 Feb 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #78

Combines the slot table-OID/TID and caller-update components of #78, with further metadata propagation and cleanup; complete HeapTuple.t_tableOid removal was deferred.

Allow to use HeapTupleData embedded in [Buffer]HeapTupleTableSlot.
27 Feb 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #78

Takes the embedded HeapTupleData workspace component of #78; the include is moved to the header’s include block and the new field gains a comment.

Add ExecStorePinnedBufferHeapTuple.
27 Feb 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 98% surepart of a posted patchsetmoderate changesfrom patchset #78

Takes ExecStorePinnedBufferHeapTuple from #78, adding a caller that transfers an existing pin instead of pinning and immediately unpinning again.

Jan 2019

8 commits
Rename RelationData.rd_amroutine to rd_indam.
22 Jan 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #78

The rd_amroutine-to-rd_indam component of #78 is preserved; remaining differences are indentation.

Move remaining code from tqual.[ch] to heapam.h / heapam_visibility.c.
22 Jan 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #78

Combines the tqual.c relocation and heap-visibility declaration components of #78, moving heap-specific code to access/heap and heapam.h with minor cleanup.

Rephrase references to "time qualification".
22 Jan 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #78

The time-qualification wording component of #78 has identical substantive changed lines.

Move generic snapshot related code from tqual.h to snapmgr.h.
22 Jan 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #78

Takes the generic snapshot/snapmgr.h move from #78; the only substantive textual residual is a small comment wording change.

Change snapshot type to be determined by enum rather than callback.
22 Jan 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 moreAlso in: Pluggable storage
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #78

Takes the snapshot-enum conversion from #78, substantially expanding documentation of snapshot semantics and removing the obsolete callback typedef. The matching attachment is newer than thread #35077.

Remove superfluous tqual.h includes.
21 Jan 2019Andres Freundmaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #78

Takes the tqual.h include cleanup from #78 and removes one additional unnecessary include.

Don't duplicate parallel seqscan shmem sizing logic in nbtree.
15 Jan 2019Andres 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 patch

A preparatory change arising from the table-AM work introduced in #1: heap_parallelscan_estimate handles SnapshotAny and nbtree reuses it. It predates the first full attached series and has no matching earlier attachment.

Move vacuumlazy.c into access/heap.
15 Jan 2019Andres 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 patch

A preparatory separation of heap-specific code for the work introduced in #1, moving vacuumlazy.c into access/heap and adjusting build files; no matching attachment predates this commit.

Credits across 43 commits

Author
Andres Freund40
Haribabu Kommi7
Ashutosh Bapat3
Alvaro Herrera2
Amit Khandekar2
David Rowley1
Dimitri Golgov1
Dmitry Dolgov1
Heikki Linnakangas1
Justin Pryzby1
loosely based on earlier work by Haribabu Kommi1
Michael Paquier1
others1
Robert Haas1
Reviewer
Andres Freundcommitter3
Robert Haas3
Haribabu Kommi2
Ashwin Agrawal1
David Rowley1
Reported by
Heikki Linnakangas2
Joe Wildish1
Committer
Andres Freund43