i18n: component-translation `submitLabel` copy key lost its only declared carrier when `element:form` retired (#9249) — decide retire vs re-anchor

[os-elon]

Found while retiring element:form at element grain (#9249, ADR-0049). Out of that card's scope — the translation face is a different schema with its own consumers (CLI i18n-extract, the resolver, their pins) — so recorded here unassigned for triage.

Measured (2026-08-21, objectstack c684d00cfc + the #9249 working tree)

• packages/spec/src/system/translation.zod.ts declares submitLabel on the per-component copy face, and its own "measured, not mirrored" table names exactly one declarer: element:form. That element is now retired whole (ADR-0049 candidate: element:form declares a full form contract with no renderer registration found in objectstack/objectui #9249) — every ElementFormProps key is a retiredKey() tombstone — so no spec-declared component carries submitLabel any more.
• The key is threaded through three co-pinned surfaces: PAGE_COMPONENT_COPY_KEYS in packages/spec/src/system/i18n-resolver.ts (shared with the CLI extractor by design), the submit alias in the component-translation alias table, and the resolver pin in i18n-resolver.test.ts (that fixture was retriaged in ADR-0049 candidate: element:form declares a full form contract with no renderer registration found in objectstack/objectui #9249 to carry the pin on a bespoke component type, which the face's own docblock documents as legal: "custom component types are legal, so these keys are also the route for a bespoke component that speaks the same vocabulary").
• The near-synonym on the LIVE form surface is object-form's submitText (I18nLabelSchema), which is not in the copy-key face at all.

Decision needed

1. Retire submitLabel from the component-translation face (drop the key, the submit alias, and its PAGE_COMPONENT_COPY_KEYS slot; ADR-0087 conversion for stored translation bundles if the face is conversion-covered) — consistent with the face's "measured, not mirrored" rule, at the cost of dropping the bespoke-component route for that one word; or
2. Keep it as the bespoke-component route (the documented ai_briefing pattern) and correct the table to say so instead of naming element:form; or
3. Re-anchor the face on the live form surface — add submitText (object-form, object-master-detail-form) to the copy face and retire submitLabel via alias — which widens the face and is its own accept-set change.

#9249 itself only corrects the two describes/table rows that advertised element:form as the carrier; it does not move the key.

Backlink: #9249.

[huangyiirene]

Triage: needs-user-decision + domain:spec — all three options change the component-translation face's accept set (a removal, a documented bespoke-only route, or a widening), which is the manual floor (public-contract shape). Lands in packages/spec (translation.zod.ts / i18n-resolver.ts / alias table) whichever way it is ruled; clause-② contract-review tier applies at dispatch time. Not dispatched while undecided.

<!-- os-decision-facets -->
一句话问题:表单「提交」按钮的翻译键,唯一挂靠的组件已整体退役(#9249),这个键现在无处安放 —— 删掉(A)、留作自定义组件的现成键位并改正说明表(B)、还是改挂到现役表单组件(C)。
① 项目长远合理性:A 缩小契约面,与该面「measured, not mirrored」的自定规则一致;B 留一个平台自身零载体的键;C 是面扩张(接受集变化,CLI 提取器与 pin 连带全动)。
② 实际业务拉动:实测平台内已无任何 spec 声明组件携带此键;唯一潜在读者是自定义组件路线(文档承认合法,但无实证使用)。零拉动默认 defer/remove。
③ 防 AI 犯错:留着无载体的键,AI 作者照抄进翻译包后静默无效、无错误无警告(与 @capabilities 决策卡同形);删除后写错处响亮拒绝。现役表单的近义词 submitText 已在活面上,混淆成本真实存在。
④ 创业阶段不扩散:每个已声明键都是永久维护义务;A 优于 B 优于 C。
推荐:A(退役,ADR-0087 转换覆盖存量翻译包);回退:B(只改说明表,不动面)。
本分析看不见什么:客户存量翻译包里是否已有人写了 submitLabel(需确认该面是否在 conversion 覆盖内并实测存量),以及自定义组件生态的真实使用面。
回一个字母即可;「A,但 X」= 附条件照 A 执行。

---

Generated by Claude Code

[huangyiirene]

Maintainer ruling recorded (2026-08-22, decision-inbox digest with the skills seat, session_01ApyDuQY2fkunMCqXiqvBhR)

The full decision inbox (46 cards) was presented with per-card four-axis recommendations; the maintainer accepted the batch, verbatim: 「接受所有」.

Ruled: Option A — retire submitLabel from the component-translation face. Drop the key, the submit alias, and its PAGE_COMPONENT_COPY_KEYS slot, with ADR-0087 conversion for stored translation bundles if the face is conversion-covered. This follows the face's own "measured, not mirrored" rule now that its only declared carrier (element:form) is retired whole per #9249, at the acknowledged cost of dropping the bespoke-component route for that one word. Per the triage note, the clause-② contract-review tier applies at dispatch time, and the dev should confirm conversion coverage and check stored bundles for existing submitLabel usage before removal.

State: needs-user-decision removed → pm:queue (ready for dispatch).

---

Generated by Claude Code

[os-sam]

Premise-refresh note before dispatch (spec seat, session_01RadETjNRLALFLhFA3xehZP): PR #11244 (Fixes #9249, MERGED 2026-08-23) touched this card's subject — element:form, submitLabel's only declared carrier, is now retired WHOLE at element grain, and the two translation.zod.ts describe/table strings that advertised the carrier were updated to say so, explicitly deferring the retire-vs-re-anchor decision to THIS card (verbatim in the merged source: "kept pending the #10926 decision (retire vs re-anchor); the bespoke-component route below still speaks it"). Whoever dispatches this card next: re-verify the face on the merged ref (origin/main ≥ 8da636880) — the carrier-absence premise is now STRONGER than when the card was filed (it was "lost its only declared carrier when element:form retired per #9249's eventual verdict"; that verdict has now landed), and the remaining open question is unchanged: retire the submitLabel copy key, or re-anchor it (the bespoke-component vocabulary route is the one consumer path still speaking it).

---

Generated by Claude Code