Dec 2019

1 commit
Try to stabilize results of new tuplesort regression test.
14 Dec 2019Tom Lanemaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surewritten from the discussion, no matching patch

Tom Lane proposed temporary tables to prevent autovacuum from changing expected plans and questioned the DROP IF EXISTS statements #8. The commit implements those changes in src/test/regress/sql/tuplesort.sql and its expected output, with completion confirmed in #10; no matching patch was attached.

Nov 2019

2 commits
Remove unused code from tuplesort.
13 Nov 2019Andres Freundmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surelanded as postedfrom patchset #4

The diff exactly matches the second patch in #4: src/backend/utils/sort/tuplesort.c replaces the unused copytup_index() implementation with an elog(ERROR) stub.

Add tests for tuplesort.c.
13 Nov 2019Andres Freundmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #4

The commit retains the regression tests from #4, switches four index-plan checks to EXPLAIN (COSTS OFF), and adds a new mark/restore test using a self-join with in-memory and disk sorts. These additions and table cleanup expand src/test/regress/sql/tuplesort.sql from 258 to 304 lines, with corresponding expected-output changes.

Credits across 3 commits

Author
Andres Freund2
Committer
Andres Freund2
Tom Lane1