Dec 2022

29 commits
Convert contrib/intarray's bqarr_in() to report errors softly
28 Dec 2022Andrew Dunstanmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1730 of 2 reviewers saw the final version

Extracts contrib/intarray from #173, retaining the state-carried error context but restoring makepol's integer return convention and using ERR directly instead of a separate result parameter.

  • Tom Lanesaw an earlier versionLLM analysis, 98% sureRequested removal of unnecessary result parameters in #170 and is explicitly credited by the commit, with subsequent implementation changes.
  • Amul Sulsaw an earlier versionLLM analysis, 96% sureSupplied the accepted state-carried escontext revision in #173 and is explicitly credited as a reviewer, before makepol's return interface was simplified.
Convert contrib/ltree's input functions to report errors softly
28 Dec 2022Andrew Dunstanmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1730 of 2 reviewers saw the final version

Extracts the ltree portion of #173, which revises Andrew's #168. Committed ltree_io.c and ltxtquery_io.c remove boolean/output-parameter interfaces and preserve existing diagnostics through SOFT_ERROR_OCCURRED checks.

  • Tom Lanesaw an earlier versionLLM analysis, 98% sureRequested simpler return-value interfaces in #170 and is explicitly credited by the commit, but parser-state and control-flow revisions followed.
  • Amul Sulsaw an earlier versionLLM analysis, 96% sureReviewed Andrew's implementation and supplied the parser-state revision in #173, receiving explicit review credit, before further return-value simplification.
Convert hstore_in to report errors softly.
27 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surewritten from the discussion, no matching patch

No attachment changes contrib/hstore; Tom volunteers the conversion in #170 and reports completion in #172. The commit updates hstore_io.c error propagation, messages, and regression coverage.

Simplify the implementations of the to_reg* functions.
27 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #1671 of 1 reviewer saw the final version

Applying #167 to the commit's parent reproduces regproc.c and func.sgml exactly, including safe input wrappers and numeric OID acceptance.

  • Andrew Dunstansaw the final versionLLM analysis, 99% sureExplicitly endorsed the wrapper simplification and numeric OID behavior in #169.
Convert the reg* input functions to report (most) errors softly.
27 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #1620 of 1 reviewer saw the final version

Matches #162 except that misc.c keeps invalid type specifications as hard errors by calling parseTypeString with NULL rather than propagating the validation function's soft-error context.

  • Andrew Dunstansaw an earlier versionLLM analysis, 93% sureAccepted the proposed scope and remaining hard-error cases in #164, before the committed misc.c error-handling adjustment.
Convert tsqueryin and tsvectorin to report errors softly.
27 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #165

Applying #165 to the commit's parent reproduces all eight changed files exactly, including parser-state contexts and suppression of empty-query NOTICEs.

Detect bad input for types xid, xid8, and cid.
27 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #163

Uses #163 with consolidated errno/endptr validation in numutils.c and corrects cidin's diagnostic type name from xid to cid in xid.c.

Convert enum_in() to report errors softly.
25 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 96% surewritten from the discussion, no matching patch

No indexed attachment changes enum.c; the linked commit converts enum_in and adds enum tests, explaining that the initial datatype survey missed enumerated types.

Convert jsonpath's input function to report errors softly
24 Dec 2022Andrew Dunstanmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% suremodified before commitmoderate changesfrom patchset #1491 of 1 reviewer saw the final version

Uses the complete jsonpath conversion in #149, incorporating Tom's pg_regfree delta from #159, removing obsolete REG_NOSUB handling, and making small Datum/comment corrections.

  • Tom Lanesaw the final versionLLM analysis, 99% sureIdentified the regex hard-error gap in #147, reviewed the replacement, and supplied the final leak fix with a reproducer in #159.
Convert contrib/seg's input function to report errors softly
23 Dec 2022Andrew Dunstanmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #1451 of 1 reviewer saw the final version

Matches #145 with the float-parser diagnostic type changed from real to seg and an explanatory comment added to seg_yyerror.

  • Tom Lanesaw the final versionLLM analysis, 100% sureRequested the diagnostic-name and comment changes in #146, which Andrew confirmed adopting in #148.
Convert xml_in to report errors softly.
16 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #136

Extracts the first patch in #136; xml.c changes match, with additional regression coverage for builds without libxml.

Clean up dubious error handling in wellformed_xml().
16 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #136

Exactly extracts the second patch in #136, replacing broad exception trapping in wellformed_xml and xml_is_document with explicit soft-error handling.

Convert range_in and multirange_in to report errors softly.
15 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #1311 of 1 reviewer saw the final version

Applying both patches from #131 to the commit's parent reproduces every affected file exactly; the commit combines input parsing and canonical-function conversion.

  • Amul Sulsaw the final versionLLM analysis, 98% sureQuestioned remaining hard errors in range helpers in #132 and accepted the internal-error distinction in #134.
Convert a few more datatype input functions to report errors softly.
15 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 98% surewritten from the discussion, no matching patch

The bpchar, varchar, and bytea conversions are absent from the attachments and explicitly excluded from Amul's submission in #126; this commit implements them in varchar.c, varlena.c, and encode.c.

Convert a few more datatype input functions to report errors softly.
14 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1261 of 1 reviewer saw the final version

Extracts cash_in and uuid_in from #126; code changes are essentially retained, with light ereturn notation and test adjustments.

  • Tom Lanesaw the final versioncommittercommitted the patch
Convert a few more datatype input functions to report errors softly.
14 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1261 of 1 reviewer saw the final version

Extracts the ACL, integer/OID vector, OID, LSN, snapshot, and TID conversions from #126. Adds missing soft-error propagation through acl.c's getid and role lookups and another oid.c error path.

  • Tom Lanesaw the final versioncommittercommitted the patch
Convert the geometric input functions to report errors softly.
14 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1261 of 1 reviewer saw the final version

Extracts geometry conversions from #126, reworks single_decode and line parsing, and drops the proposed float arithmetic error-context changes as explained in #130.

  • Tom Lanesaw the final versioncommittercommitted the patch
Convert a few more datatype input functions to report errors softly.
14 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1261 of 1 reviewer saw the final version

Extracts bit, varbit, inet/cidr, and MAC conversions from #126. Reworks mac8.c to collect hex-decoding failure and report it centrally instead of continuing after a helper's soft error.

  • Tom Lanesaw the final versioncommittercommitted the patch
Convert domain_in to report errors softly.
11 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #1161 of 1 reviewer saw the final version

Applying #116 to the commit's parent exactly reproduces domains.c, CREATE DOMAIN documentation, and the domain regression files.

  • Alvaro Herrerasaw the final versionLLM analysis, 80% sureApproved the proposed CHECK-error boundary before attachment in #110, and the committed implementation preserves that policy.
Convert json_in and jsonb_in to report errors softly.
11 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1151 of 2 reviewers saw the final version

Extracts JSON/JSONB conversion from #115, separating the JsonSemAction API commit and adding Unicode-conversion soft errors discussed in #118. The commit explicitly credits Andrew Dunstan and Tom Lane.

  • Pavel Stehulesaw an earlier versionLLM analysis, 88% sureEndorsed the original JSON patch in #113, before JSONB, semantic-action, and Unicode-conversion changes expanded it.
  • Tom Lanesaw the final versioncommittercommitted the patch
Change JsonSemAction to allow non-throw error reporting.
11 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1151 of 1 reviewer saw the final version

Extracts the JsonSemAction status-return API from #115, updating jsonapi.c and existing callbacks while leaving actual JSON input soft-error behavior for the next commit.

  • Andrew Dunstansaw the final versionLLM analysis, 96% sureReviewed Tom's semantic-action API extension and said it looked good in #117.
Restructure soft-error handling in formatting.c.
10 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surewritten from the discussion, no matching patch0 of 1 reviewer saw the final version

No attachment contains the formatting.c conversion; Andres proposed replacing its ad hoc error mechanism in #102, Tom agreed in #103, and the commit replaces it with errsave/ereturn.

  • Andres Freundsaw an earlier versionLLM analysis, 83% sureProposed the formatting.c conversion in #102, which Tom accepted in #103, without a subsequent review of the finished implementation.
Convert datetime input functions to use "soft" error reporting.
9 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surewritten from the discussion, no matching patch

Implements the datetime conversion Tom undertook in #103 and reported complete in #109; no attachment contains its date.c, timestamp.c, and datetime.c changes.

Allow DateTimeParseError to handle bad-timezone error messages.
9 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surewritten from the discussion, no matching patch

Unattached prerequisite to the datetime work announced in #103, introducing DateTimeErrorExtra and routing timezone failures through DateTimeParseError.

Convert a few datatype input functions to use "soft" error reporting.
9 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1000 of 1 reviewer saw the final version

Combines the datatype, numeric, and cube portions of #100; substantive conversion code is retained, while rowtype tests use the existing complex type and add overflow coverage.

  • Andres Freundsaw an earlier versionLLM analysis, 99% sureReproduced the record_in resource leak in #60 and reviewed later cleanup in #93, before additional numeric and cube conversions were submitted.
Add test scaffolding for soft error reporting from input functions.
9 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #1000 of 2 reviewers saw the final version

Extracts SQL validation scaffolding from #100, adding ValidIOData caching of constant type-name parsing and a catalog-version bump.

  • Andres Freundsaw an earlier versionLLM analysis, 98% sureRequested hard-error testing in #58 and reviewed the scaffolding in #93, before textual type specifications and constant-argument caching were added.
  • David G. Johnstonsaw an earlier versionLLM analysis, 90% sureReviewed validation-function wording and interface choices in #75 and #76, before the later type-name API and caching changes.
Create infrastructure for "soft" error reporting.
9 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1003 of 4 reviewers saw the final version

Extracts the infrastructure portion of #100, preserving ErrorSaveContext, errsave/ereturn, and InputFunctionCallSafe; additionally registers ErrorSaveContext in pgindent's typedef list.

  • Andres Freundsaw the final versionLLM analysis, 99% sureReviewed error-stack behavior and context declarations in #93 and confirmed the final forward-declaration approach in #99.
  • Robert Haassaw the final versionLLM analysis, 98% sureDrove the errsave/ereturn distinction in #41 and the adopted struct Node declaration approach in #98.
  • David G. Johnstonsaw the final versionLLM analysis, 94% sureProposed soft-error terminology and documentation corrections in #75, which Tom explicitly adopted in #89.
  • Corey Huinkersaw an earlier versionLLM analysis, 90% sureReviewed the early context interface and naming in #27, before the returning macro, boolean wrapper API, and error-stack refactoring.
Minor code refactoring in elog.c (no functional change).
7 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #681 of 2 reviewers saw the final version

Exactly extracts the preliminary elog.c refactoring from #68, creating shared error-stack initialization and location/domain helpers.

  • Andres Freundsaw an earlier versionLLM analysis, 97% sureRequested extraction of duplicated elog.c code in #56, which the commit explicitly credits, before the standalone refactoring was posted.
  • Andrew Dunstansaw the final versionLLM analysis, 94% sureReviewed the submitted series containing the unchanged preliminary refactoring and said it looked good in #70.
Doc: subdivide System Information Functions and Operators.
7 Dec 2022Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Adds sect2 subdivisions to func.sgml following the documentation discussion in #74 and #81; no attachment contains this standalone restructuring.

  • David G. Johnstonsaw the final versionLLM analysis, 94% sureDiscussed alternative documentation structures and accepted the sect2 approach in #81 before it was committed.

Credits across 29 commits

Author
Tom LaneLLM13
Andrew DunstanLLM5
Amul SulLLM4
Reviewer
Tom Lanecommitter5
Committer
Tom Lane25
Andrew Dunstan4

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