Skip to content

chore: release v0.5.2 - #16

Merged
MagicalTux merged 1 commit into
masterfrom
release-plz-2026-07-27T21-38-17Z
Jul 28, 2026
Merged

chore: release v0.5.2#16
MagicalTux merged 1 commit into
masterfrom
release-plz-2026-07-27T21-38-17Z

Conversation

@MagicalTux

Copy link
Copy Markdown
Member

🤖 New release

  • intl: 0.5.1 -> 0.5.2 (✓ API compatible changes)
Changelog

0.5.2 - 2026-07-27

Added

  • (collation) re-vendor locale tailorings from CLDR 48

Changed

  • (collation) re-vendor locale tailorings from CLDR 48. The rule table is now
    generated by codegen from data/cldr/48/collation/*.xml instead of a
    pre-filtered JSON blob, so Tailoring::for_locale carries CLDR's own
    standard rule for 78 locales (was 41). The first vendoring had filtered out
    every rule using [before]/[import]/expansions; the parser gained that
    syntax later, but the data was never regenerated, stranding Swedish, Finnish,
    Norwegian, Icelandic, Estonian, Turkish, Azerbaijani, Kazakh, Albanian,
    Ukrainian and Greenlandic on lossy hand-written approximations. Swedish now
    folds ü under y and æ/ø under ä/ö and orders þ/ð/đ, none of
    which the old &z < å < ä < ö expressed.

Fixed

  • (collation) Tailoring::for_locale("de") no longer returns the CLDR
    phonebook tailoring (ä = ae, ß = ss). German has no standard collation
    in CLDR — phonebook is de-u-co-phonebk — so plain de sorts in root order,
    matching Intl.Collator("de").
  • (collation) Tailoring::for_locale("ga") no longer invents a primary
    distinction for the long-vowel accents. CLDR states the root order is valid for
    Irish, where á is a secondary variant of a.
  • (collation) drop the Balochi (bal) tailoring, which inverted its own rule:
    it sorted ا before آ where the rule (and root order) put it after. The
    consistency gate had not caught it because its tokenizer skipped
    space-separated rules entirely.
  • (collation) the tests/collation_data_consistency gate now lexes the same
    grammar the runtime parser does ([before] resets, bracket options, <* star
    ranges, '…' quoting, \uXXXX escapes, / expansions) rather than scanning
    for relation characters, so it can validate the full unfiltered CLDR rule set
    instead of silently skipping the parts it could not tokenize.

Added

  • (collation) nb/nn inherit Norwegian (no) and tl inherits Filipino
    (fil) collation, as CLDR's locale inheritance specifies.


This PR was generated with release-plz.

@MagicalTux
MagicalTux merged commit 4164984 into master Jul 28, 2026
11 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant