mirror of
https://github.com/rmcrackan/Libation.git
synced 2026-05-31 12:02:54 -04:00
The de-CH row hard-coded U+2019 as the expected thousands separator, which only matches certain .NET/ICU/CLDR data versions. Linux .NET hosts return U+0027 (ASCII apostrophe) for the same culture, so every Linux CI run failed this test regardless of the actual change under test. Resolve U+2019 in DataRow expectations to the runtime culture's NumberGroupSeparator before comparison so the test stays stable across hosts while still verifying the engine respects culture-specific formatting. - Added DataRows for de-DE (period) and ja-JP (comma) - Added Samplerate_template_uses_culture_NumberGroupSeparator as an explicit regression guard that asserts the engine uses whatever the runtime CultureInfo reports Fixes #1813. Verified locally: 584 tests pass, 0 failed (565 succeeded, 19 skipped Windows-only). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>