i18n: @objectstack/core's exported resolveLocale is a verbatim second copy of @objectstack/spec's resolveBundleLocale — one rule, two places to edit #16085

Activity

1. claude

   added

   findingRecorded observation — held, not dispatchable until the findings triage round grades itRecorded observation — held, not dispatchable until the findings triage round grades it

   domain:engineRetired 2026-10-22, alias of domain:server (engine surface); in-flight cards keep it until they landRetired 2026-10-22, alias of domain:server (engine surface); in-flight cards keep it until they land

   i18n

   on Sep 5, 2026
2. os-zhuang commented on Sep 7, 2026

   

   os-zhuang

   on Sep 7, 2026 – with Claude

   Contributor

   More actions

   Triage: closing not planned — outside the three admission classes (#16351, as refined by PR #16557), on the card's own measurement: ⛔ "The two produce the same answer for every input this repo can construct, so nothing is wrong today. It is a maintenance hazard, not a live bug."

   Against each class: ⛔ (a) no defect — the two agree on every constructible input; ⛔ (b) nothing declares "one rule, one place" for this; ⛔ (c) not applicable. Under the boundary now in the rule (「不完整 vs 错误」) a verbatim duplicate that agrees is neither.

   ⭐ The identification is thorough and worth keeping: same four steps, same order, same comments naming the same examples (zh-cn→zh-CN, zh-CN→zh, zh→zh-CN), differing only in being list-shaped vs record-shaped. ⇒ Not a resemblance — a copy, established at the level of its comments.

   承接者: ⛔ none — and I checked both candidates rather than assuming.

   • i18n: the validation-message bridge matches the locale EXACTLY while every document translator does variant fallback — accept-language: zh refuses in English on the same response whose dataset labels are Chinese #15757 is fenced: this is "the third copy that card's premise implies, spotted on the way; it is out of that card's scope and is not changed by it."
   • i18n: the core in-memory i18n fallback never receives the declared i18n.fallbackLocale — its t() falls back to defaultLocale only, so one declaration answers differently under the memory fallback and under service-i18n #15694 looked like the natural vehicle — same file, packages/core/src/fallbacks/memory-i18n.ts. ⚠️ It is closed completed, by merged PR fix(core,runtime,spec): the in-memory i18n fallback learns the declared i18n.fallbackLocale #16088 (2026-09-06). ⇒ The file was opened, the fallback learned fallbackLocale, and the duplicate was correctly left alone as out of scope. The carrier existed, was used for something else, and is gone.

   ⇒ That is the eighth closure in this sweep where the acceptance-notes fallback has no carrier, and it is the clearest illustration of why: a card being open on the same file is not a carrier, because a well-fenced dispatch does not widen. ⛔ The emptiness is structural, not an oversight by anyone.

   What I am uncertain about, stated because it is the one thing that could make this the wrong call: the duplicate is exported from @objectstack/core, so it is reachable by anything depending on that package, and a future consumer picking it up would be adopting the copy rather than the rule. ⛔ Not measured whether any consumer does today, in this repo or in objectui / cloud. If one does, the maintenance hazard acquires a named victim and the grade changes.

   Reopen conditions (any, no approval needed):

   1. The negotiation rule changes — a new step, a change to which variant wins when several match, an ordering fix. At that moment the edit must be made twice with nothing failing if only one is done, and the divergence is a live defect the moment it lands. ⇒ Whoever edits resolveBundleLocale should be sent here.
   2. Any consumer is measured importing resolveLocale from @objectstack/core.
   3. Someone opens memory-i18n.ts for another reason and can take the delegating one-liner as a free rider — ⛔ not work anyone should schedule, but the right moment if it arrives.

   Reversibility: fully reversible; nothing decays while the rule is unchanged. ⭐ And the suggested shape survives here for whenever it is taken: have resolveLocale delegate to resolveBundleLocale rather than re-derive the steps — a delegating one-liner makes divergence impossible rather than merely unlikely, which is the right form and is strictly better than adding a test that the two agree.

   ⛔ Correctly distinct from #15694 (same file, different mechanism — the memory fallback never receiving i18n.fallbackLocale): collapsing the duplicate would not have fixed that, and fixing that did not collapse this. ⛔ Neither is a duplicate of the other.

   ---

   Generated by Claude Code
3. os-zhuang

   closed this as not plannedon Sep 7, 2026