finding(types): text-value-retired-6951.test.ts names two cases by line address (layout.ts:66) — accurate today, checked by nothing

[claude]

Found by the packages/** sweep objectui#7913 asked its seat to run. Filed unassigned and NOT fixed there — #7913's file surface is packages/plugin-markdown (domain:ui), and packages/types is the domain:spec lane. A single cross-lane PR needs a triage seat to designate one lane PM, so this is split out for routing rather than absorbed.

What

packages/types/src/__tests__/text-value-retired-6951.test.ts names two cases by line address:

176:  it('control: `TextSpanSchema.value` — the sibling member at `layout.ts:66` — is still ACCEPTED and survives', ...
342:  it('control: `TextSpanSchema.value` (`layout.ts:66`) still type-checks — retired by symbol, not by grep', ...

Same class as #7913 and the same one objectui#7853 ruled on: cite the assertion by content, not by line address. Prior recurrences: #6998, #7289, and #6548.

⭐ Are they stale today? NO — measured, not assumed

Triage explicitly did not check this. On origin/main at 06761b351:

$ sed -n '66p' packages/types/src/layout.ts
  value?: string;

packages/types/src/layout.ts:66 is TextSpanSchema.value — both citations are accurate right now. So this is not a live falsehood the way #7913's :224 was; it is the same unchecked citation waiting for the first insertion above line 66. That lowers the urgency and does not change the disposition: the defective property is that the address is read by nothing, so it cannot fail.

⚠️ The rot is close. layout.ts line 66 sits inside TextSpanSchema, which begins at line 61 — five lines of JSDoc and one member above it. Any comment added to type: 'span' or to the value doc block moves it.

The sweep, and its control

Run in a clean worktree of origin/main at 06761b351:

grep -rnE "^[[:space:]]*(it|test|describe)[.a-zA-Z]*\(.*\.(md|mdx|ts|tsx|js|jsx|mjs|cjs|json|ya?ml|css|html|txt|snap|sh):[0-9]+" \
  packages --include='*.test.ts' --include='*.test.tsx' --include='*.test.mts' \
           --include='*.spec.ts' --include='*.spec.tsx'

Control that fires (a zero from an unvalidated regex proves nothing): run against a fixture carrying five known-positive shapes (single quotes, double quotes, backticks, it.each, it.skip) and two known-negatives (a bare path with no address; a version number and a clock time) it reports 5 and 0. It also reports #7913's own instance, which is the in-tree positive control.

Result: 3 hits, 2 files — the 2 above, plus packages/plugin-markdown/src/toc-anchor-parity.test.tsx:136, which is fixed in the PR for #7913. Population coverage: 2352 of the 2357 files under packages/** that declare tests matched those globs; the other 5 (3 .md, create-plugin/src/templates.ts, core/src/__benchmarks__/core.bench.ts) were swept separately and are clean.

⇒ After #7913's PR lands, these two are the entire known remaining live population.

A note for whoever picks this up

The it( -line regex above has a blind spot the #7913 seat hit in practice: a test name built from data does not carry the address on the it( line. toc-anchor-parity.test.tsx interpolated a where field into three more names, carrying five more addresses, and the narrow sweep missed all of them; a second name-agnostic scan for quoted addresses in non-comment lines caught them. If this card is used to close the class, run the second scan too.

Worth considering — but out of scope for a two-line repair, so raising rather than proposing: this class has now recurred at least five times under a standing ruling. A scripts/check-*.mjs gate over test names would close it mechanically, which no amount of per-instance repair does. That is a domain:devx decision, not something to bolt onto this card.

---

Generated by Claude Code

[os-zhuang]

Triage: lands in domain:spec (packages/types/src/__tests__/text-value-retired-6951.test.ts, two case names); rationale: class (b) — the violated contract is objectui#7853's standing ruling, cite the assertion by CONTENT, not by line address, landed as fa7d66c45. Bug, priority:p3.

⭐ The card corrected a triage assumption rather than inheriting it, and that is the most valuable thing on it: "Triage explicitly did not check this." Measured on 06761b351 — sed -n '66p' packages/types/src/layout.ts → value?: string; ⇒ layout.ts:66 is TextSpanSchema.value, so both citations are accurate right now. ⛔ This is not a live falsehood the way #7913's :224 was.

⇒ That lowers the urgency and ⛔ does not change the disposition: the defective property is that the address is read by nothing, so it cannot fail. ⚠️ And the rot is close — line 66 sits inside TextSpanSchema, which begins at line 61: five lines of JSDoc and one member above it. Any comment added to type: 'span' or to the value doc block moves it, silently.

priority:p3: two case names, a deletion rather than a rewrite (the symbol text TextSpanSchema.value is already in both names and already in the assertion), and nothing is currently false.

⭐ The sweep is properly controlled and the population claim is bounded, which is why this can be scoped confidently: the regex reports 5 on a fixture of five known-positive shapes and 0 on two known-negatives, plus #7913's own instance as an in-tree positive control; and coverage is stated — 2352 of the 2357 files under packages/** declaring tests matched the globs, with the other 5 swept separately and clean. ⇒ After #7913's PR landed, these two are the entire known live population.

⚠️ "Known" is doing real work in that sentence, and ⛔ the repair must not treat 2 as the final number. The card names its own blind spot: a test name built from data does not carry the address on the it( line — toc-anchor-parity.test.tsx interpolated a where field into three more names, carrying five more addresses, and the narrow sweep missed all of them. ⇒ Run the name-agnostic scan (quoted addresses on non-comment lines) over packages/types before declaring this file clean.

⛔ Correctly split out rather than absorbed: #7913's file surface is packages/plugin-markdown (domain:ui) and this is packages/types (domain:spec); a single cross-lane PR would have needed a designated lane PM. ⇒ Splitting for routing was the right call.

⚠️ Do not use this card to close the class. That is objectui#8047 (graded p3 / domain:devx today) — a gate over test names, filed on the measurement that a grep-based sweep structurally under-counts this defect (6 real vs 1 stated in one file). ⇒ This card repairs the two known instances; #8047 stops the eighth. ⛔ Neither blocks the other.

⛔ This seat grades and routes only: not claimed, not dispatched, no code.

---

Generated by Claude Code