Mar 2024

1 commit
Introduce "builtin" collation provider.
14 Mar 2024Jeff Davismaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1Also in: Built-in CTYPE provider (modified before commit)

May - Jun 2023

12 commits
ICU: do not convert locale 'C' to 'en-US-u-va-posix'.
21 Jun 2023Jeff Davismaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2Also in: pg_collation.collversion for C.UTF-8 (not this thread)
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #412 of 2 reviewers saw the final version

Re-lands the conversion-removal component of #41 in src/backend/utils/adt/pg_locale.c and src/bin/initdb/initdb.c, retaining validation-disable/reset adjustments while omitting ICU-version-dependent C-locale tests. I disagree with the from_discussion attribution in thread #47764: its supplied verdict identifies no matching attachment, whereas this thread contains the implementation that the commit explicitly describes as an un-revert.

  • Peter Eisentrautsaw the final versionLLM analysis, 96% sureReviewed the conversion removal in #34, explaining its consequences for explicit libc versus ICU selection; only light changes followed in the committed component.
  • Tom Lanesaw the final versionLLM analysis, 99% sureAnalyzed buildfarm results for the first landing and supplied the regression diff in #43, identifying the version-dependent tests omitted from the second landing.
initdb: change default --locale-provider back to libc.
21 Jun 2023Jeff Davismaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2Also in: collation-related loose ends before beta2 (part of a posted patchset)
CREATE DATABASE: make LOCALE apply to all collation providers.
16 Jun 2023Jeff Davismaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2
LLM analysis, 100% suremodified before commitlight changesfrom patchset #1121 of 1 reviewer saw the final version

The complete LOCALE patch matches #112 with the documentation and Perl-test fixup from #114 folded in. The backend and frontend behavior is unchanged from the complete submission, as confirmed by the commit announcement in #115.

  • Peter Eisentrautsaw the final versiontrailerRequested removal of redundant createdb locale expansion in #96, reviewed the narrowed scope, and approved #112 while supplying the final documentation/test fixes in #114.
Remove read-only server settings lc_collate and lc_ctype
7 Jun 2023Peter Eisentrautmaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2
LLM analysis, 100% suremodified before commitlight changesfrom patchset #471 of 1 reviewer saw the final version

All changed lines match #47, removing the two GUCs and replacing test lookups with pg_database queries. The commit message softens the claim that the settings are meaningless, following #71 and the explicit announcement in #94.

  • Jeff Davissaw the final versiontrailerAccepted the mechanical patch in #71 and requested the adopted commit-message clarification that LC_COLLATE and LC_CTYPE still describe the server environment.
ICU: use uloc_getDefault() for initdb.
26 May 2023Jeff Davismaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2
LLM analysis, 100% surelanded as postedfrom patchset #781 of 1 reviewer saw the final version

The initdb.c diff exactly matches Daniel Verite's attachment in #78, replacing default-collator inspection with pg_strdup(uloc_getDefault()). Although git records Jeff Davis as author, the source sender and commit's Author credit identify Daniel as the patch author.

  • Jeff Davissaw the final versioncommittercommitted the patch
Doc fixes for commit 1e16af8ab5.
26 May 2023Jeff Davismaster16beta2REL_19_STABLE16beta2REL_18_STABLE16beta2REL_17_STABLE16beta2REL_16_STABLE16beta2
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

The charset.sgml cleanup implements Peter Eisentraut's unattached review in #69: section spacing, lowercase parameter names, table introductions, and reordered settings. Jeff confirms applying the feedback in #77.

  • Peter Eisentrautsaw the final versionLLM analysis, 100% sureSupplied the concrete documentation corrections in #69 that became this follow-up commit without an attached patch.
Doc: fix some rendering problems in the PDF docs build.
21 May 2023Tom Lanemaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #66

Includes the charset.sgml column-width hints from #66, then adds widths for the settings table and rewrites its combining-accent example. It also replaces an unsupported character in release-16.sgml.

Doc improvements for language tags and custom ICU collations.
18 May 2023Jeff Davismaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #581 of 1 reviewer saw the final version

Matches the charset.sgml documentation in #58 except for a missing SQL quote correction and relocation of the co settings-table row. The commit announcement is #60.

  • Jonathan S. Katzsaw the final versiontrailerProvided detailed wording and terminology corrections in #56, then reread the revised documentation and said it looked good in #59.
ICU: check for U_STRING_NOT_TERMINATED_WARNING.
17 May 2023Jeff Davismaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surelanded as postedfrom patchset #521 of 1 reviewer saw the final version

The changed lines exactly match #52, checking U_STRING_NOT_TERMINATED_WARNING and adjusting language-tag allocation loops in pg_locale.c and initdb.c.

  • Alexander Lakhinsaw the final versionLLM analysis, 100% sureReported the ASAN overflow in #50 and reviewed the proposed fix's icu_from_uchar() check and backpatch implications in #54.
Reduce icu_validation_level default to WARNING.
17 May 2023Jeff Davismaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #491 of 1 reviewer saw the final version

Exactly matches the validation-default component of #49, changing ERROR to WARNING in configuration, initialization, documentation, and regression tests; the other six patches are not included.

  • Peter Eisentrautsaw the final versionLLM analysis, 99% sureQuestioned the rationale and requested concrete scenarios for reducing validation severity in #51; this was substantive review, although Jeff later acknowledged missing the objection before committing.
Revert "ICU: do not convert locale 'C' to 'en-US-u-va-posix'."
9 May 2023Jeff Davismaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1reverts 1 commit here
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Exactly reverses f7faa9976c, following Tom Lane's buildfarm report in #43 and Jeff's decision in #44. No revert patch was attached.

  • Tom Lanesaw the final versionLLM analysis, 100% sureProvided the concrete buildfarm failure and regression diff in #43 that prompted the unattached revert.
ICU: do not convert locale 'C' to 'en-US-u-va-posix'.
8 May 2023Jeff Davismaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #411 of 1 reviewer saw the final version

Lands only the first component of #41, removing the C/POSIX conversion in pg_locale.c and initdb.c. Its implementation matches the attachment; regression inputs and corresponding expected messages change lowercase c to uppercase C.

  • Peter Eisentrautsaw the final versionLLM analysis, 99% sureEndorsed the removal in #34, explaining that users could then explicitly choose libc or a native ICU locale; only light test edits followed.

Apr 2023

1 commit
Fix initdb --no-locale.
21 Apr 2023Jeff Davismaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1
LLM analysis, 100% surewritten from the discussion, no matching patch2 of 2 reviewers saw the final version

Adds locale_provider = COLLPROVIDER_LIBC to initdb's --no-locale option handling, implementing the report in #12 and confirmation in #13. It predates the first attachment and is announced in #24.

  • Andrew Gierthsaw the final versionLLM analysis, 98% sureReported in #12 that --no-locale incorrectly selected an environment-derived ICU locale, prompting this one-line fix.
  • Tom Lanesaw the final versionLLM analysis, 100% sureIndependently reproduced the --no-locale failure and supplied initdb output in #13 before the fix was committed.

Credits across 14 commits

Author
Jeff DavisLLM8
Daniel Verite1
Peter EisentrautLLM1
Tom LaneLLM1
Reviewer
Jeff Daviscommitter2
Peter Eisentraut2
Daniel Vérité1
Jeremy Schneider1
Jonathan S. Katz1
Reported by
Alexander Lakhin1
Andrew Gierth1
Committer
Jeff Davis12
Peter Eisentraut1
Tom Lane1

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