[finding] @objectstack/rest registers four generic synonyms the standard catalog already covers (CONFLICT, NOT_FOUND, FORBIDDEN, INTERNAL) — contract call, not a cleanup #8211

Activity

1. hotlong

   added

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

   on Aug 12, 2026
2. hotlong commented on Aug 12, 2026

   

   hotlong

   on Aug 12, 2026 – with Claude

   ContributorAuthor

   More actions

   PM note — domain:cli seat (#6024): C endorsed, and route it with #8087. Not claimed, not queued.

   The diagnosis is the finding, not the four synonyms

   The ledger's header rule says to use the catalog instead of registering a synonym, but that rule is PROSE ONLY — the mechanical gate rejects only a literal StandardErrorCode member, which is why four accumulated without a decision.

   That sentence is the card. CONFLICT, NOT_FOUND, FORBIDDEN, INTERNAL did not accumulate because four authors each disagreed with the rule; they accumulated because nothing was checking. A rule that lives only in a header comment is a rule with a silent-drift channel attached, and four is simply how many drifted through before someone routed an arm past a typed writer and noticed.

   This is the same finding this lane has filed all day in other costumes — a check that cannot fail, a gate whose success condition a broken config satisfies, a ledger whose prose and mechanism disagree. ADR-0049's enforce-or-remove is the general form: declared and enforced must be the same fact, not two facts that happen to agree.

   Endorsing C, for the reason given

   • A leaves the channel open — the fifth synonym lands the same way, and the next author copies a rule the codebase demonstrably does not follow.
   • B spends real client breakage (the SDK surfaces err.code verbatim; consumer sweeps across objectui and cloud) on vocabulary tidiness, with no measured pull — after fix(rest): converge the record-sharing family onto the ADR-0112 D5 envelope (#8111) #8212 all four are registered and parse, so nobody is hitting anything today.
   • C converts the prose into something enforced. It prevents the mistake at authoring time rather than relying on a reader noticing a comment, and it stops the fifth synonym landing while the fate of the existing four is still open.

   The sequencing in the recommendation is right too: C now, B only if a specific code proves confusing in practice — with a measured victim, not on principle.

   ⚠️ One thing whoever implements C must get right, because this lane has been burned by it repeatedly: a "semantic synonym" detector that never fires is worse than no detector. Whatever heuristic it uses, the gate needs a pin proving it rejects a newly-introduced synonym of a standard member — not merely that it passes on today's tree. Today's tree passing is exactly what the prose rule also did.

   Route this with #8087

   These are one family: what happens when a producer's chosen error.code and the standard catalog say the same thing differently. #8087 is the same question at the dispatcher door and is already with the maintainer, where my recommendation is B delivered as a gate — structurally the same answer as C here. I have cross-referenced this card there.

   Answering them together avoids ruling the general case at one door and re-litigating it at the other. Leaving finding/unassigned as filed; not urgent, and correctly so.

   ---

   Generated by Claude Code
3. hotlong

   added

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   domain:specDomain lane — seat card indexed by label:pm:seatDomain lane — seat card indexed by label:pm:seat

   auto-adjudicated

   and removed

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

   on Aug 12, 2026
4. hotlong

   added the

   Task

   issue type on Aug 12, 2026
5. hotlong commented on Aug 12, 2026

   

   hotlong

   on Aug 12, 2026 – with Claude

   ContributorAuthor

   More actions

   Triage — auto-adjudicated (triage-seat adjudication lane, maintainer authorization of 2026-08-12: high-confidence cases where "AI 会有明确的建议…这个是不是可以让分诊就处理掉", with feature/ADR/protocol classes reserved for the maintainer). Ruling: Option C — extend the ledger admission gate to reject a semantic synonym of a standard-catalog member unless explicitly waived with a recorded reason; the existing four (CONFLICT, NOT_FOUND, FORBIDDEN, INTERNAL) are grandfathered via explicit waiver entries; B is deferred until a specific code has a measured victim; A (status quo) is rejected as the shape the ledger's own prose already forbids. ⇒ pm:queue, domain:spec, type Task.

   Why this clears the adjudication confidence gate:

   • Not on the human floor: no wire change (all four codes stay registered and parse), no new capability, no ADR — this converts an already-declared prose rule into its mechanical form, i.e. the restore-declared=enforced family. The genuine contract call (fate of the existing four) is explicitly deferred, not decided.
   • Four prongs, all pointing at C: ① coherence — closes the silent-drift channel on the closed vocabulary; ② pull — zero user impact either way today (post-[finding] registerSharingEndpoints / respondSharingError still answers the bare-string error dialect (ADR-0112) — the sharing half split out of #8073 #8111/fix(rest): converge the record-sharing family onto the ADR-0112 D5 envelope (#8111) #8212 everything parses), so B's client breakage buys nothing measurable; ③ AI-error-resistance — the textbook case: authoring-time refusal over a header comment nobody re-reads; ④ scope — one gate change vs. cross-repo consumer sweeps.
   • Sibling ruling, same direction: the maintainer's 2026-08-12 ruling on The dispatcher emits unregistered error.code values verbatim — three suites pin bodies that ApiErrorSchema would reject #8087 ("B delivered as a gate", at the dispatcher door) already chose mechanical-enforcement-over-prose for this exact family; this inherits that reasoning one door over, per the same-family meta-rule. The domain:cli seat's 20:36Z comment independently endorses C.
   • Veto window, not a permission gate: this ruling goes in the round report's adjudication list; the maintainer can overturn it there or here.

   Dispatch constraints: lands in packages/spec/src/api/error-code-ledger.zod.ts (waiver shape) + error-code-ledger.test.ts (gate) ⇒ spec semantic lane, claude-fable-5 mandatory (ledger admission behaviour changes). The gate MUST carry a pin proving it rejects a newly-introduced synonym — a detector that only passes on today's tree is the failure mode this card exists to close (per the cli seat's comment). Reconcile wording/mechanism with #8087's ruled body-parsing gate and PR #8088's resolver so the two doors state one rule.

   ---

   Generated by Claude Code
6. akarma-synetal

   added 2 commits that reference this issue on Aug 17, 2026

   feat(spec): mechanical standard-synonym admission gate on the error-c…

   ...

   36d90fc

   feat(spec): the object-* block family enters ComponentPropsMap — typo…

   ...

   8f20201
7. akarma-synetal

   added 4 commits that reference this issue on Oct 7, 2026

   fix(runtime)!: an app-authored body may not bind a hook to, or write,…

   ...

   bd70706

   fix(spec): the error-code waiver reasons and the auth-feature registr…

   ...

   36e4647

   fix(plugin-auth): phone send-otp with no deliverable SMS service answ…

   ...

   a43d90a

   test(spec): the first api/ file group's test titles state each cited …

   ...

   5a22eb5