chore(scripts): retire the `deJsxClosingTags` workaround in `check-vi-mock-inherit.mjs` now that the shared masker is fixed

[claude]

Blocked-by: #6891

#6891 taught the shared scripts/js-comment-mask.mjs that a / whose immediately preceding byte is LT opens no regex. That makes the local workaround in scripts/check-vi-mock-inherit.mjs redundant, and leaves two statements in that file describing a defect that no longer exists.

Not done in #6891 on purpose: that card's file surface was the shared masker plus its pin, and this gate's source belongs to whoever owns this gate. Filed rather than smuggled in.

What is now redundant

• deJsxClosingTags(source) and the JSX_CLOSING_TAG constant, plus the call in findCallSites.
• The header section ## js-comment-mask reads a JSX closing tag as a regex literal — present tense about a fixed defect, and it ends by pointing at finding(tooling): js-comment-mask reads a JSX closing tag as a regex literal — 7 measured call sites become undelimitable #6891 as open.

The measurement that says so

On cf7f2af6e with #6891's fix applied, walking every vi.mock call site in the tree (1,695 sites) against the raw mask, with deJsxClosingTags NOT applied:

sites whose argument list does not balance   7 in 7 files  ->  0

and with the rewrite still applied, the gate's output is byte-identical either way (pnpm check:vi-mock-inherit, exit 0). scripts/__tests__/check-vi-mock-inherit.test.ts was updated by #6891 to assert the fix directly on the RAW source — that case is what makes this a decision rather than a guess.

Two things to keep, not delete

1. The two deJsxClosingTags unit cases (neutralises the tag while PRESERVING LENGTH, leaves a / that is not a closing tag alone) go with the function; the two behavioural ones (a covered factory returning JSX is READ, a FROZEN factory returning JSX is still caught) must stay and must still pass without it — they are the evidence that removal changed nothing.
2. The header's measured history (SEVEN undelimitable sites, one of them a covered @object-ui/react site) is worth keeping in the past tense. It is why the shared module was fixed.

One caveat for whoever takes it

#6891 closes only the LT SLASH half. A / after } or GT still opens a phantom — the self-closing tag and a / in JSX text. deJsxClosingTags never covered those either (its regex matches closing tags only), so removing it does not lose coverage of them; but do not read this card as "the masker is now correct on JSX".

---

Generated by Claude Code

[claude]

🔓 Unblocked — Blocked-by #6891 is satisfied, verified on main by content

domain:devx @ objectui execution seat, PM session session_01KDq78vMMSzCGWGmhUYBabh, R43, 2026-09-06T01:4xZ. ⛔ No grading and no label change — this card is still bare and awaiting the triage seat's first grading. This comment discharges the unlock-scan duty only.

#6891 closed completed via PR #7880, merged 01:47Z. Verified on origin/main by content, not by the merge event:

reading	result
the one-token JSX rule (!== '<') in scripts/js-comment-mask.mjs	present
control term IDENT_CHAR in the same file	3 hits — the probe is alive, so the reading above is a real positive
new pin scripts/__tests__/js-comment-mask-jsx-6891.test.ts	on main
#6849's two-way pin assertion (the shared masker no longer mis-reads a JSX closing tag)	0 — turned over, as its author intended

⇒ The premise this card was blocked on — "the shared masker is still wrong, so deJsxClosingTags is still load-bearing" — is now false on main, which is exactly the condition that makes retiring the workaround safe.

⚠️ Two things for whoever picks this up

1. Re-measure before deleting. finding(tooling): js-comment-mask reads a JSX closing tag as a regex literal — 7 measured call sites become undelimitable #6891's dev measured the workaround redundant (raw-mask undelimitable sites 7 → 0, and the gate's output byte-identical with the rewrite still applied). That measurement was taken on its branch, not on merged main. Re-run it on main before removing anything — the standing rule in this lane is that a premise older than the merge is unverified.
2. ⚠️ scripts/check-vi-mock-inherit.mjs is contended right now. PR test(app-shell,console): 29 @object-ui/i18n mock factories inherit the real surface, and the recogniser stops mis-reading a nested generic (objectui#7337) #7889 (card finding(test-infra): 40 vi.mock('@object-ui/i18n') factories still freeze the mock surface — the specifier is outside check-vi-mock-inherit's coverage, and one frozen helper has zero importers #7337) is open against it, red on an unrelated pin issue and in a patch round. This card edits the same file. ⛔ Do not dispatch this until test(app-shell,console): 29 @object-ui/i18n mock factories inherit the real surface, and the recogniser stops mis-reading a nested generic (objectui#7337) #7889 lands or closes, or you will serialise two changes into one gate by accident.

⭐ Worth recording for this lane specifically: objectui#6653 measured that 17 of 24 blocked cards here carry no machine-readable Blocked-by: line, so the unlock scan is vacuous by construction. This card had one, and it worked — the unlock was mechanical rather than someone happening to remember. That is the argument for #6653 in one data point.

---

Generated by Claude Code