Jan 2021

3 commits
Throw error when assigning jsonb scalar instead of a composite object
31 Jan 2021Alexander Korotkovmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1Also in: [PATCH] Generic type subscriptionAlso in: jsonb array-style subscriptionAlso in: jsonb array-style subscripting
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #2713 of 16 reviewers saw the final version

Takes the scalar-error patch from #271, preserving its jsonfuncs.c checks, tests, and documentation with only whitespace/comment wrapping changes. This source is newer than the prototypes in thread #35251, thread #34033, and thread #33337.

  • Tom Laneno review foundtrailerHis earlier infrastructure and coercion reviews do not establish review of the later scalar-error patch; #275 reports a failure after the commit.
  • Arthur Zakirovno review foundtrailerHis reviews concern earlier subscripting implementations, before the scalar-error patch was introduced in #254.
  • Pavel Stehulesaw the final versiontrailerHe reviewed scalar-error behavior and documentation, then reported successful check-world and documentation tests of #271 in #272.
  • Dian M Faysaw the final versiontrailerShe verified the update/trigger problem in #262 and supplied final documentation corrections in #270, followed only by light changes.
  • Andrew Dunstanno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Chapman Flackno review foundtrailerHis discussion beginning at #222 concerns type assertions, not the subsequent scalar-error implementation.
  • Merlin Moncureno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Peter Geogheganno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Alvaro Herrerano review foundtrailerHis substantive work here concerned the renaming patch, committed in #120, rather than scalar-error handling.
  • Jim Nasbyno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Josh Berkusno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Victor Wagnerno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Aleksander Alekseevno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Robert Haasno review foundtrailerHis discussion at #79 concerns type dependencies and predates this scalar-error feature.
  • Oleg Bartunovno review foundtrailerHis only message here, #54, is a bare agreement and establishes no patch review.
  • Alexander Korotkovsaw the final versioncommittercommitted the patch
Filling array gaps during jsonb subscripting
31 Jan 2021Alexander Korotkovmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1Also in: [PATCH] Generic type subscriptionAlso in: jsonb array-style subscriptionAlso in: jsonb array-style subscripting
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #2713 of 16 reviewers saw the final version

Takes the gap-filling patch from #271, including JB_PATH_FILL_GAPS, JB_PATH_CONSISTENT_POSITION, missing-container creation, and tests, with formatting/comment edits only. The final implementation here supersedes the prototypes in thread #35251, thread #34033, and thread #33337.

  • Tom Lanesaw an earlier versiontrailerHis coercion review in #160 caused removal of the polymorphic implementation supporting the earlier gap-filling design, which was subsequently rewritten.
  • Arthur Zakirovno review foundtrailerHis reviews predate the gap-filling implementation introduced with #140.
  • Pavel Stehulesaw the final versiontrailerHe repeatedly tested padding and missing-path creation, exposed corruption, and validated the final #271 series in #272.
  • Dian M Faysaw the final versiontrailerHer tests and documentation reviews clarified container creation and NULL padding, culminating in #270, which was incorporated with only light subsequent changes.
  • Andrew Dunstanno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Chapman Flackno review foundtrailerHis type-assertion discussion beginning at #222 does not review gap filling.
  • Merlin Moncureno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Peter Geogheganno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Alvaro Herrerano review foundtrailerHis build checks and commit activity concern the earlier renaming portion, not the gap-filling implementation.
  • Jim Nasbyno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Josh Berkusno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Victor Wagnerno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Aleksander Alekseevno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Robert Haasno review foundtrailerHis type-dependency discussion at #79 predates and does not review gap filling.
  • Oleg Bartunovno review foundtrailerHis only message here, #54, is a bare agreement and establishes no patch review.
  • Alexander Korotkovsaw the final versioncommittercommitted the patch
Implementation of subscripting for jsonb
31 Jan 2021Alexander Korotkovmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1Also in: [PATCH] Generic type subscriptionAlso in: jsonb array-style subscriptionAlso in: jsonb array-style subscripting
LLM analysis, 100% surepart of a posted patchsetmoderate changesfrom patchset #2712 of 16 reviewers saw the final version

Takes the base jsonb implementation from #271, changing JsonbToJsonbValue to initialize caller-owned storage, updating callers, retaining setPath as static, and adding catalog/formatting maintenance. This latest matching series supersedes thread #35251, thread #34033, and thread #33337.

  • Tom Lanesaw an earlier versiontrailerHe reviewed execution and coercion design, notably rejecting original-type tracking in #235 and proposing the implemented implicit-coercion rules in #241, before further substantial changes.
  • Arthur Zakirovsaw an earlier versiontrailerHe tested and reviewed early jsonb/subscripting versions, including NULL behavior in #61 and interface cleanup in #88, before later API rewrites.
  • Pavel Stehulesaw the final versiontrailerHe tested the final source series #271 and reported successful compilation, check-world, and documentation builds in #272.
  • Dian M Faysaw an earlier versiontrailerShe tested assignment behavior and revised documentation through #270, after which the committed base implementation received moderate allocation/API edits.
  • Andrew Dunstanno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Chapman Flackno review foundtrailerHis discussion at #222 and following messages explores a possible type-assertion design without establishing review of an attached jsonb patch.
  • Merlin Moncureno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Peter Geogheganno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Alvaro Herrerano review foundtrailerHis substantive checks concern the renaming attachment of #114, not its jsonb implementation.
  • Jim Nasbyno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Josh Berkusno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Victor Wagnerno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Aleksander Alekseevno review foundtrailerNo review message by this credited reviewer appears in this thread.
  • Robert Haassaw an earlier versiontrailerHe examined the existing generic subscripting proposal's type/dependency semantics in #79 and #83, before extensive implementation changes.
  • Oleg Bartunovno review foundtrailerHis only message here, #54, is a bare agreement and establishes no patch review.
  • Alexander Korotkovsaw the final versioncommittercommitted the patch

Dec 2020

6 commits
Allow subscripting of hstore values.
11 Dec 2020Tom Lanemaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #212

Takes the hstore attachment of #212, preserving its implementation and tests while adding hstore usage documentation and a CREATE TYPE documentation reference.

Allow ALTER TYPE to update an existing type's typsubscript value.
11 Dec 2020Tom Lanemaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #212

Exactly matches the ALTER TYPE SUBSCRIPT attachment of the two-patch submission #212, including typecmds.c and documentation changes; the hstore attachment landed separately.

Support subscripting of arbitrary types, not only arrays.
9 Dec 2020Tom Lanemaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1Also in: [PATCH] Generic type subscription
LLM analysis, 98% suremodified before commitmoderate changesfrom patchset #2071 of 6 reviewers saw the final version

Combines the full core implementation #204 with the latest matching MAXDIM delta #207, then updates LLVM callback typing and typelem documentation; the commit credits Dmitry's work, extensively rewritten by Tom. These attachments supersede the older series in thread #35251.

  • Arthur Zakirovsaw an earlier versionLLM analysis, 99% sureHe tested and corrected executor, dependency, and interface code through #88, and the commit explicitly credits his review before the later core rewrite.
  • Peter Eisentrautsaw an earlier versionLLM analysis, 99% sureHis review in #50 argued for storing subscripting as a type property, adopted in #56, and the commit explicitly credits him.
  • Pavel Stehulesaw an earlier versionLLM analysis, 99% sureHe tested builds, documentation, regressions, and array performance in #150, #177, and #192 before Tom's substantial rewrite.
  • Andres Freundsaw an earlier versionLLM analysis, 99% sureHis reviews in #74 and #200 shaped callback dispatch and retained distinct opcodes, before further semantic and LLVM changes.
  • Justin Pryzbysaw an earlier versionLLM analysis, 94% sureHe supplied concrete core comment corrections in #157 that Dmitry incorporated in #158, before the later rewrite.
  • Tom Lanesaw the final versioncommittercommitted the patch
jit: Reference function pointer types via llvmjit_types.c.
9 Dec 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1981 of 1 reviewer saw the final version

Extracts the function-pointer-type refactor from #198 across llvmjit.c, llvmjit_expr.c, llvmjit_types.c, and llvmjit.h; the boolean bugfix was committed separately and redundant extern declarations were removed.

  • Tom Lanesaw the final versiontrailerHe explicitly studied the patch in #203, questioned redundant extern declarations, and requested the corrected version be committed.
Teach contain_leaked_vars that assignment SubscriptingRefs are leaky.
8 Dec 2020Tom Lanemaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1REL_13_STABLE13.2+4 more
LLM analysis, 87% surepart of a posted patchsetmoderate changesfrom patchset #197

Extracts the assignment-leakproofness correction introduced in #197, relocating its assignment test from nodeFuncs.c/check_functions_in_node into clauses.c/contain_leaked_vars_walker and retaining the Var-sensitive check for backpatching.

jit: Correct parameter type for generated expression evaluation functions.
8 Dec 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1REL_13_STABLE13.2+2 more
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #198

Separates the boolean correctness fix contained in #198: removes the same LLVMBuildTrunc and fixes the existing signature to use TypeStorageBool rather than taking the prototype's larger signature refactor.

Feb 2019

1 commit
Renaming for new subscripting mechanism
1 Feb 2019Alvaro Herreramaster12beta1REL_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 #1143 of 3 reviewers saw the final version

Takes the renaming attachment from #114, changing ArrayRef and related parser/executor names to SubscriptingRef terminology; Alvaro confirms the separate landing and pgindent edits in #120.

  • Tom Lanesaw the final versiontrailerHis naming and structural reviews in #71 and #73 cover the renaming subsequently extracted and landed with light adjustments.
  • Arthur Zakirovsaw the final versiontrailerHis detailed naming and unused-state review in #88 covers the extracted renaming work, followed by rebasing and formatting changes.
  • Alvaro Herrerasaw the final versioncommittercommitted the patch

Credits across 10 commits

Author
Dmitry Dolgov1 LLM5
Tom LaneLLM3
Andres Freund2
Reviewer
Tom Lanecommitter6
Alvaro Herreracommitter4
Arthur Zakirov4
Aleksander Alekseev3
Alexander Korotkovcommitter3
Andrew Dunstan3
Chapman Flack3
Dian M Fay3
Jim Nasby3
Josh Berkus3
Merlin Moncure3
Oleg Bartunov3
Pavel Stehule3
Peter Geoghegan3
Robert Haas3
Victor Wagner3
Committer
Tom Lane4
Alexander Korotkov3
Andres Freund2
Alvaro Herrera1

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