Apr 2022

5 commits
Documentation for SQL/JSON features
8 Apr 2022Andrew Dunstanmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1Also in: SQL/JSON: functions (part of a posted patchset)
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #849 of 10 reviewers saw the final version

The combined JSON_TABLE and PLAN documentation from #84 exactly matches the committed 479-line JSON_TABLE section in doc/src/sgml/func.sgml. This is shared provenance with thread #38009, whose verdict attributes the functions documentation to #502994; this thread supplies the JSON_TABLE material from the later submission, while the commit also makes editorial changes to the functions and aggregate documentation.

  • Andres Freundsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in a6baa4badd, although no documentation review by him appears in this thread.
  • Alexander Korotkovsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in a6baa4badd, with no message in this thread.
  • Pavel Stehulesaw an earlier versionLLM analysis, 98% sureReviewed documentation in #34, requesting reproducible examples and separate basic and advanced explanations before later documentation expansion.
  • Andrew Alsupsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in a6baa4badd, with no message in this thread.
  • Erik Rijkerssaw the final versionLLM analysis, 98% sureRead the final full series' func.sgml and supplied prose and example corrections in #86, although many corrections were not incorporated.
  • Zhihong Yusaw the final versionLLM analysis, 95% sureReviewed PLAN DEFAULT documentation spelling in #40, corrected in #41, and is explicitly credited in a6baa4badd.
  • Himanshu Upadhyayasaw the final versionLLM analysis, 98% sureIdentified several JSON_TABLE documentation typos in #67, corrected before the documentation was extracted, and is explicitly credited in a6baa4badd.
  • Daniel Gustafssonsaw the final versionLLM analysis, 98% sureReviewed documentation spelling in #55 and reiterated outstanding corrections in #78, which #84 addressed.
  • Justin Pryzbysaw the final versionLLM analysis, 100% sureExplicitly credited in a6baa4badd; his visible message #37 is only a rebase-status notice.
  • Andrew Dunstansaw the final versioncommittercommitted the patch
Further improve jsonb_sqljson parallel test
6 Apr 2022Andrew Dunstanmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch3 of 3 reviewers saw the final version

Implements the parallel-planner settings proposed inline in #99, reduces the fixture to 50,000 rows, and drops it to resolve the recovery failure reported in #95. Andrew confirms this implementation in #100; no matching attachment exists.

  • Andres Freundsaw the final versionLLM analysis, 100% sureProposed the adopted parallelism settings and disabling leader participation in #99.
  • Stephen Frostsaw the final versionLLM analysis, 100% sureReported that retaining the unlogged test table broke recovery tests in #95, leading to the committed DROP TABLE.
  • Tom Lanesaw the final versionLLM analysis, 98% sureObjected to million-row fixtures in repeatedly executed core regression tests in #98, prompting the smaller retained test.
Reduce running time of jsonb_sqljson test
6 Apr 2022Andrew Dunstanmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch2 of 2 reviewers saw the final version

Responds to the test-runtime report in #91 by making test_parallel_jsonb_value unlogged and reducing its population from one million to 500,000 rows. Andrew reports the committed change in #94; no attached patch contains it.

  • Andres Freundsaw the final versionLLM analysis, 100% sureMeasured jsonb_sqljson as the slowest regression test and requested a runtime reduction in #91.
  • Tom Lanesaw the final versionLLM analysis, 91% sureCorroborated the performance concern with increased buildfarm cycle times in #92, before this reduction was committed.
PLAN clauses for JSON_TABLE
5 Apr 2022Andrew Dunstanmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #849 of 10 reviewers saw the final version

Combines the PLAN DEFAULT and PLAN components of #84, omitting documentation as announced in #88. Their combined net code and test changes match the commit after whitespace normalization.

  • Andres Freundsaw the final versionLLM analysis, 100% sureExplicitly credited in fadb48b00e, with no pre-commit PLAN review visible in this thread.
  • Alexander Korotkovsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in fadb48b00e, with no message in this thread.
  • Pavel Stehulesaw an earlier versionLLM analysis, 98% sureExamined the standard and requested separation of PLAN and PLAN DEFAULT in #24, before the later jsonb-only rework.
  • Andrew Alsupsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in fadb48b00e, with no message in this thread.
  • Erik Rijkerssaw the final versionLLM analysis, 85% sureReported substantial successful functionality and speed testing of the series in #50, after which the PLAN components received chiefly naming and cleanup changes.
  • Zhihong Yusaw the final versionLLM analysis, 92% sureReviewed PLAN enum naming, child counters, and disabled root-path validation in #40, with subsequent PLAN changes chiefly cleanup and rebasing.
  • Himanshu Upadhyayasaw the final versionLLM analysis, 100% sureCredited in fadb48b00e, while his visible review #67 explicitly concerns the separate base JSON_TABLE patch.
  • Daniel Gustafssonsaw the final versionLLM analysis, 96% sureQuestioned dead ordinal-handling code in #55 and #78, leading to its removal from the PLAN DEFAULT component in #84.
  • Justin Pryzbysaw the final versionLLM analysis, 100% sureExplicitly credited in fadb48b00e; #37 itself only records rebase status.
  • Andrew Dunstansaw the final versioncommittercommitted the patch
JSON_TABLE
4 Apr 2022Andrew Dunstanmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #845 of 10 reviewers saw the final version

Takes the base JSON_TABLE component of #84 without documentation or PLAN extensions. Remaining differences are whitespace, IS_JSON_* to JSON_*_OP naming, and equivalent conditional expansion in parse_expr.c.

  • Andres Freundsaw the final versionLLM analysis, 100% sureExplicitly credited in 4e34747c88, although his visible thread participation begins after that feature commit.
  • Alexander Korotkovsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in 4e34747c88, with no message in this thread.
  • Pavel Stehulesaw an earlier versionLLM analysis, 100% sureTested compatibility and requested EXISTS and formatting changes in #30, then reported successful check-world and performance testing in #34 before substantial later rework.
  • Andrew Alsupsaw the final versionLLM analysis, 100% sureExplicitly credited as a reviewer in 4e34747c88, with no message in this thread.
  • Erik Rijkerssaw an earlier versionLLM analysis, 99% sureReported missing build files in #44 and an LLVM failure during JSON_TABLE testing in #47, before later node-handling fixes.
  • Zhihong Yusaw an earlier versionLLM analysis, 98% sureReviewed memory ownership, expression comments, and list iteration in #40, before missing equality support was added in #83.
  • Himanshu Upadhyayasaw an earlier versionLLM analysis, 98% sureTested many JSON_TABLE scenarios and reviewed error behavior, hints, comments, and includes in #67, before the node-handling fixes in #83.
  • Daniel Gustafssonsaw an earlier versionLLM analysis, 98% sureReviewed conditionals, type handling, diagnostics, and comments in #55 and #78, before additional equality support arrived in #83.
  • Justin Pryzbysaw the final versionLLM analysis, 100% sureExplicitly credited in 4e34747c88; his visible contribution #37 is only a rebase-status notice.
  • Andrew Dunstansaw the final versioncommittercommitted the patch

Credits across 5 commits

Author
Nikita GlukhovLLM3
Reviewer
Andrew Dunstancommitter3
Committer
Andrew Dunstan5

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