Uh oh!
There was an error while loading. Please reload this page.
fix(i18n): remove five dead objectView/appDesigner locale keys, add reverse sweep - #4731
Merged
Merged
Conversation
…everse sweep console.objectView.striped/bordered/virtualScroll and appDesigner.stripedRows/ bordered had no t()/tt() call site anywhere in the repo, in any of the ten packs. The console.objectView three describe grid options objectstack#7176 retired upstream and objectui stopped forwarding in #4649; the appDesigner pair was independently dead by name coincidence only. Adds scripts/check-i18n-dead-keys.mjs, a report-only reverse sweep (pack key set minus referenced key set) that reuses check-i18n-call-site-keys.mjs's own AST walk (extended, not duplicated, to also collect referenced keys/branches/ dynamic-template heads during the same pass) plus a whole-repo text safety net that demotes indirectly-referenced keys to a NEEDS-REVIEW tier instead of false-confirming them dead. Not wired into CI - report-only by design. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RnQd8iMMUwXQEV1crFmQiQ
The latest updates on your projects. Learn more about Vercel for GitHub. |
yinlianghui
commented
Aug 15, 2026
CollaboratorAuthor
Review verdict: ACCEPT (reviewer of record, PM session Verified against this PR, not the report: fifteen changed files; Will flip ready + enable auto-merge once Lint, Type Check, and Test shards 1–4 conclude Generated by Claude Code |
Contributor
✅ Console Performance Budget
📦 Bundle Size Report
Size Limits
|
yinlianghui
marked this pull request as ready for review
August 15, 2026 19:59
Uh oh!
There was an error while loading. Please reload this page.
This was referenced Aug 15, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes#4658
What changed
Deleted five locale keys with no
t()/tt()call site anywhere in the repo,from all ten packs (
enplus the nine translations, soall-locales-key-parity.test.tsstays green):console.objectView.striped,console.objectView.bordered,console.objectView.virtualScroll— grid optionsobjectstack#7176retired upstream (maintainer-ruled 2026-08-10) and objectui stopped
declaring/forwarding in
#4649. Dead permanently, not merely dormant.appDesigner.stripedRows,appDesigner.bordered— textually similar nameonly; independently dead. No control in
AppCreationWizard.tsx, no entryin
useDesignerTranslation.ts'sDESIGNER_DEFAULT_TRANSLATIONSfallbackmap, no mention anywhere in the repo outside the locale packs.
Added
scripts/check-i18n-dead-keys.mjs, the reverse-sweep the card askedfor: pack key set minus referenced key set. Report-only — exit 0 always
unless
--strictis passed (nothing calls--strict; not wired into anyworkflow). It reuses
check-i18n-call-site-keys.mjs's own AST walk ratherthan a second parser:
analyze()was extended (not duplicated) to alsocollect, during the same pass that already classifies every call site,
referencedKeys(literal + plural-suffixed),referencedBranches(
returnObjects: trueconsumption), anddynamicHeads(template-key staticprefixes — the call-site gate's own
missing-prefixclass read in reverse).A candidate the AST pass cannot explain also gets a text safety net: one
grep -rFpass over the whole repo (not justpackages/+apps/, which isthe AST walk's scope) for the literal dotted key string, excluding the locale
packs themselves. Zero hits anywhere → CONFIRMED; some hit elsewhere (an
indirect
labelKey: 'foo.bar'property, a doc mention, a test fixture) →NEEDS-REVIEW, not confirmed dead. This is what lets the report tell a
truly-dead key apart from one only the AST pass's own blind spots (calls
outside
packages/+apps/, keys reached through a variable rather than aliteral argument) made look dead.
Premise verification (issue's mechanism assumptions)
confirmed independently three ways: (a)
check-i18n-dead-keys.mjs's ASTpass found no
t()/tt()call site; (b) the text safety net found zerotextual occurrences of any of the five dotted key strings anywhere outside
the locale packs (not even in comments/docs/tests); (c) direct greps for
the bare identifiers (
striped,bordered,virtualScroll,stripedRows) confirm the only non-locale, non-unrelated hits are theobjectql.zod.tscomment documenting the upstream retirement and theblock-config.tscomment documenting the#4649field removal — bothcorroborating evidence, neither a consumer.
check-i18n-call-site-keys.mjs's parsing machinery reused, notduplicated —
analyze()was extended in place; the new script importsanalyze()andcollectEnKeys()directly. The one place a second,independent pass was necessary (the text safety net) is documented as such
in the new script's header, with the specific gap it closes (indirect
references, and
examples//e2e/— measured empty oft()/tt()callstoday, so the AST walk's
packages/+apps/-only scope costs nothing onthis run, but is flagged as a scope note rather than silently matched).
this class — confirmed by reading each gate's own header (already
documented there, cross-checked against behavior):
all-locales-key-parityis pack-vs-pack (full parity on a key nobody wants);
check-i18n-*-keysis call-site-to-
en(asks whether a wanted key exists, never whether anexisting key is wanted);
check-i18n-en-driftfires onenVALUE change(a key that never changes never registers, and removal is explicitly
parity's business per its own output — see the
check:i18n-driftrunbelow).
Sweep report / measured false-positive rate
Run against this PR's tree (
d3d028631): 2927 pack keys, 13776 referenced.626 candidates, 384 CONFIRMED, 242 NEEDS-REVIEW (38.7%) — i.e. a naive
AST-only reverse sweep (no text safety net) would have been wrong about
2 in 5 of its own candidates. That NEEDS-REVIEW rate is this PR's answer to
"is this worth red-ifying yet": no — even with the text safety net, a
CONFIRMED verdict still needs a human before deletion, so
--strictstaysunwired.
Spot-verification beyond the five keys actually deleted here (sampled, not
exhaustive — the full 384 is far bigger than this card's scope):
workflow.*(54/58 confirmed) traces toProcessDesigner.tsx, whichhardcodes English directly and imports no translation hook at all;
publicForm.demo.*(36/36 confirmed) has zero footprint anywhere includingoutside
t()calls. Filed as#4730(unassigned,findinglabel,observation-class) rather than expanded in this PR — out of
#4658'sfive-key scope, and
#4730is not addressed here.Tests
New self-test (
scripts/__tests__/check-i18n-dead-keys.test.ts, 16 tests):positive controls for all three "still live" escape hatches (literal+plural-
suffix,
returnObjectsbranch, dynamic-template head — each must keep a keyOUT of the candidate set), a dead-key positive control (must be CONFIRMED), an
indirect-reference control (must be NEEDS-REVIEW, not CONFIRMED), and the
locale-pack self-reference exclusion (a key's own definition line in
en.tsmust never itself count as a textual hit — the one bug that would silently
empty the CONFIRMED tier entirely; caught during development when an earlier
docstring draft used a real target key as an illustrative example and
self-polluted its own report entry, fixed by not spelling a real candidate
key in the script's own comments).
All gate/test output above and in this body is from the same commit —
git rev-parse --short HEAD=d3d028631, run after the final commit(re-run if this PR gets a post-review push).
Notes
check-i18n-call-site-keys.mjswas extended, not forked:analyze()gained three new Set fields (
referencedKeys,referencedBranches,dynamicHeads), populated inline at the exact points the existing walkalready computes the underlying facts. No existing finding/counter logic
changed — pinned by the pre-existing self-test suite staying green
unmodified (97 tests, same file).
scripts/changes do not need a changeset (check-changeset-presence.mjsconfirms: 10 files under a released package's
src/, 0 under an ignoredpackage —
scripts/isn't under any package'ssrc/at all).--strictexists on the new script deliberately, so a future decision tored-ify this needs no rewrite — just wiring, once
#4730's backlog isworked down enough that CONFIRMED stops meaning "hundreds, mostly
unreviewed."
Generated by Claude Code
Generated by Claude Code