Repository navigation
[sweep] Two user-visible labels the authoring contract cannot express a translation for (#5377 #5728) #6489
Description
Sweep card (maintainer-directed, 2026-08-07). One claim, one PR, two Fixes.
Admission criterion
A user-visible label that the platform otherwise translates, which the authoring contract cannot express a translation for — the author has no legal way to write the localized value.
| # | Surface | Gap |
|---|---|---|
| #5377 | view tab labels (tabs[].label) |
no translation key and no resolver — the list-page tab bar is untranslatable in every locale |
| #5728 | component.zod.ts's I18nLabelSchema key |
declared as a plain string, while 3 platform pages author inline multilingual maps at 31 sites — i.e. real authoring already exceeds the declaration |
Both are authoring-surface gaps, and both are target:v17. #5728 is the sharper one: the declaration is already contradicted by in-repo authored metadata, so either the sites are wrong or the schema is.
Explicitly NOT members (same topic, different review face)
- i18n:
report与dataset完全没有 translator —— dataset 看板的表头/度量标签永远是英文(#4854 的范围外发现,拆单跟踪) #5199 (report/datasethave no translator at all) — that is runtime translator wiring, not an authoring-surface widening. Different mechanism, different reviewer question; it stays its own card. - [i18n] 为 flow 与 workflow 提供可作者化的标签和节点翻译契约 #4426 (flow / workflow authorable label + node translation contract) — currently
pm:on-hold; a held card may not ride a sweep.
Recording the exclusions because "i18n" is a topic, not an admission criterion — bundling by topic is exactly what the same-class rule forbids.
Direction (state per member)
Each member widens the authoring surface (add the key + resolver / accept the inline map form) or rules the current authoring wrong and corrects the 31 sites. #5728 cannot be settled without deciding which side is authoritative — say which, with the site count as evidence.
Scope discipline
- Zero changes beyond the two members;
git diff --statcorresponds to the checklist. - A widened schema may not shrink the refusal surface: shapes that are genuinely invalid keep rejecting, verbatim.
- Out-of-scope findings filed separately per PD chore: version packages #10.
Landing constraints
Touches packages/spec generated baselines ⇒ os-regen four-step序; check:authorable-surface green; ⛔ never hand-edit the anchor. If the ruling lands on "correct the 31 sites", the platform-page edits ride the same PR and appear in the checklist.
Review shape
By checklist: one line per member — surface | what an author can write today | what they can write after | which side was ruled authoritative.
Filed unassigned. Member issues keep pm:queue but are excluded from batch selection while this card is open; both close with the PR.
Activity