test(components,scripts): retire the ui:icon pin's membership half, record the third zero - #7604
Merged
os-sam merged 3 commits intoSep 4, 2026
Merged
Conversation
…ecord the third zero objectui#5936 asked for the consumer of a component registration's `icon` meta before extending the icon gate to it. Three first-party populations were measured and none has one (objectui, objectstack, cloud), so the gate is not extended — adjudicated 2026-09-04. Records that where the next reader hits it (the gate's own "Deliberate boundaries"), with the three readings, their controls, the adjudication, and the caveat that three zeros are not a proof of absence. Retires the membership half of the `ui:icon` local pin, which was kept at objectui#5633 on the premise that the palette lives outside this repo. It guarded 1 of the 45 registrations declaring an `icon` meta while the gate is adjudicated not to cover the other 44, and its hand-copied tokeniser had already gone stale against the seam objectui#5935 / PR #7491 created. The coupling half stays: it needs no external vocabulary and cannot drift. The published `ComponentMeta.icon` key is untouched. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KbJQ1y1J12nZxYzFWhP8Q3
…cation The gate suite asserted the `ui:icon` pin "is kept, and says why", matching the literal string the pin carried. objectui#5936 retired that pin's membership half, so the assertion pinned a claim the adjudication removed. It now pins what must not go silently instead: the gate header carries the three readings, the adjudication, and the not-a-proof-of-absence caveat; and the pin file survives without re-importing the `icons` record. Matched against a whitespace-normalised header, because these are assertions about prose. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KbJQ1y1J12nZxYzFWhP8Q3
Test and tooling only — the diff is confined to packages/components/src/__tests__/ and scripts/. The empty frontmatter rests on that, NOT on the "no consumer anywhere" premise the card's retirement half turns on. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KbJQ1y1J12nZxYzFWhP8Q3
Contributor
✅ Console Performance Budget
The eager closure is every chunk the entry reaches through static imports — what the browser fetches and parses before the app renders. The entry chunk on its own is a small fraction of it. 📦 Bundle Size Report
Size Limits
|
This was referenced Sep 4, 2026
os-sam
marked this pull request as ready for review
September 4, 2026 06:14
os-sam
deleted the
claude/issue-5936-registration-icon-meta-dead-surface
branch
September 4, 2026 06:29
This was referenced Sep 4, 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 free
to 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 #5936
Clause-②: no — verified from the diff, not assumed. Four files: a changeset, one test file, one gate-test file, and a comment block in a gate script.
packages/typesis not reached (0 files), so the publishedComponentMeta.iconkey is untouched.What this card asked, and what came back
#5936 asked for the consumer of a component registration's
iconmeta before the lucide icon gate was extended to that population. The answer is the third zero — no consumer in objectui, objectstack, or cloud — relayed onto the card in comment 5536103214. ⇒ The gate is NOT extended (adjudicated 2026-09-04), and scope item 4's disposition is what is left.Retirement reading taken: the NARROW one. The gate-extension plan and the
ui:iconpin's membership half. ⛔ The published meta key is not removed, not deprecated, not narrowed — that is the maintainer floor and it was fenced out of this dispatch. The argument for and against the wide reading is in the report on the card rather than acted on here.I re-derived objectui's zero rather than inheriting it
The card's zero was two grep shapes. Registry meta is reachable through five read paths, so I censused all five at
24e027e93:.icon?getMeta()form.tsx:453reads.labelling,page.tsx:463reads.namespace/.isContainer/.inputs,Registry.ts:717is the definition,:760reads.deprecatedgetPublicConfigs()react-page.tsx:75, readstypeandisContainergetConfig()/getAllConfigs()/getNamespaceComponents()Controls fired on the same tree:
getMeta(occurs 51 times,grep -i iconreturns 4898 lines. The six rawmeta?.iconhits are all local lookup tables (WIDGET_TYPE_META,STATUS_META) whoseiconis a React component, not a registration meta string — plus two self-references inside the pin's own prose.⭐ A finding the card's framing did not predict. The pin's membership half covers two rows —
meta.iconand the glyph input'sdefaultValue— and the card reads as though they differ in liveness. They do not.page.tsx:463is the only first-party reader ofmeta.inputs, and it feedsmanifestFromConfigs, which copiesname,type,required,enum,binding,descriptionand dropsdefaultValue. Both rows are unconsumed by identical reasoning, so there is no honest "retire one row, keep the other" — they stand or fall together.The disposition, argued
Retire the membership half. Keep the coupling half. Four measurements, not a preference:
iconmeta; the pin guarded 1, and the gate is adjudicated not to cover the other 44. A membership check over 1/45 of its own population does not manage that risk — it signals that it is managed, and that signal is what sent Locate the consumer of a registration'siconmeta before extending the icon gate to it — zero read points in objectui AND objectstack, so two suspect spellings stay undetermined #5936 looking for a consumer the pin implied must exist.describeIconLookup/resolveIconand tokenises on/[-_\s]+/; the copy still saidsplit('-'). Both halves of its own stated correctness condition went stale inside one round with nothing going red. That is what a pin over a surface with no consumer does.iconmetas are live on both lucide surfaces today. Controls fired both ways (circle-checktrue /no-such-glyph-xyzzyfalse, on record and dynamic; retiredsmilereads recordfalse, dynamictrue).FaceSlightlySmilingtrue, LITCircleChecktrue, DARKXyzzyfalse in each), the gate reads the installed lucide at runtime so it re-judges the authored-node population across any bump, and the 44 unpinned registrations were never covered either way.The gate's own suite pinned the claim being retired — declared, not slipped in
scripts/__tests__/check-lucide-icon-record-names.test.tsasserted the pin "is kept, and says why" by matching the literal string'NOT retired by objectui#5633'. Retiring the membership half forces that assertion red mechanically — this PR cannot land without it, so it is part of the change rather than a drive-by. Verified free of competing claims: no open PR touchesscripts/orpackages/components/src/__tests__/(7 open PRs listed), and #5935 is closedcompletedby PR #7491, MERGED.It now pins what must not go silently instead: the gate header carries the three readings, the adjudication, and the not-a-proof-of-absence caveat, and the pin file survives without re-importing the
iconsrecord. Matched against a whitespace-normalised header, because these are assertions about prose and prose re-wraps — the first draft matched raw text and failed on a line wrap, which would have taught the next reader to reflow the sentence rather than keep the meaning.The two retired spellings: omitted, because they already landed
icon: 'Layout'andicon: 'AlertTriangle'are not in this diff, and the card's body is stale on this point: PR #7590 landed both ase30ed155con 2026-09-04, before this dispatch. Nothing to include or omit. Confirmed from git, not from the card. ⇒ The dispatch's "state which and why" is answered by "neither — they are already onmain".Verification
Exit codes captured by redirect-then-read, never through a pipe; every verdict below is the gate's own line.
Gate union re-run AFTER the final commit, at
c3372c22, tree clean —check-lucide-icon-record-namesexit=0OK lucide icon names: 182 authored/declared names reaching 1 record-reading resolver are live 'icons' keys (record 1767 keys; dynamic surface 4 sites, 2025 names, not judged here).|check-control-bytesexit=0 (6228 files) |check-changeset-presenceexit=0 |check-changeset-fixedexit=0 |check-changeset-no-majorexit=0 |check-changeset-overwriteexit=0 |check-vi-mock-specifiersexit=0 |check-vi-mock-inheritexit=0.Tests, from the repo root (the package-scoped form is refused by this repo's guard):
pnpm exec vitest run packages/components/ scripts/__tests__/— 331 files / 5099 tests passed, 0 failures. Dependency closure built first (pnpm --filter '@object-ui/components^...' build, 8 projects) so nothing read a stale dist.Type-check:
pnpm --filter @object-ui/components type-checkexit=0, script line echoed (@object-ui/components@17.6.0 type-check) so the filter matched — not the zero-match false green. It runstsc --noEmit && tsc -p tsconfig.test.json, and the edited pin file is in that program (--listFiles, 1 hit) — not the excluded-tests NOT-MEASURED case.tsconfig.scripts.jsoncovers both editedscripts/files (1 hit each, 0 errors).Three ablations, each under
trap RESTORE EXIT INT TERMwith absolute paths, restoring viagit checkout HEAD -- ABSOLUTE_PATH(never the bare form, which reads the index). Implementation committed first so the restore leg's reference contains it. Every mutation proven on disk before any result was read, every restore proven by state:ui:icon's registration meta so it disagrees with the input default. On disk: anchored counts 1 to 0 and 0 to 1, blobd9b8ad55fto01062e928. Result: red,AssertionError: expected 'circle-check' to be 'face-slightly-smiling', 1 failed / 1 passed — the precondition correctly still passed. Restored: blob back tod9b8ad55f,git diff HEAD --statempty,git status --porcelainempty.'NOT a proof of'1 to 0, marker 0 to 1, blob830c2a8bbto714860707. Result: red,AssertionError: the boundary reads as a proof of absence, 1 failed / 41 passed. Restored and proven by state.24e027e93(proven on disk: gate blob830c2a8bbtoe1600563f, boundary text count 0) and re-ran the gate. Base and head output are byte-identical, both exit 0 — 182 judged either way. Restored and proven by state.Declared narrowing (one): lint. eslint ran on the changed files rather than repo-wide. Three pieces of evidence make that a measurement rather than a skipped run: (1) the receiving population is read off eslint's own config —
files: ['**/*.{ts,tsx}']— and its membership answer was exercised in both directions, 117 rules resolved for a changed.tsfile versusundefinedfor a path outside the population; the.mjsgate script resolves 0 rules, i.e. eslint does not lint it at all; (2) the count is eslint's own, from--format json: 3 paths, 0 errors, 0 warnings, exit 0; (3) type-aware linting is not enabled — norecommendedTypeChecked, noprojectService, noproject:ineslint.config.js— so every rule is per-file and this diff cannot move the verdict of any untouched file. A repo-wide run could not have caught anything this one did not. Nothing else was narrowed; CI runs the full farm regardless.Not run, with reason:
check:sdui-registration-pins— its population derives from thesideEffectsarray and the registered keys; this diff moves no module and no key.🤖 Generated with Claude Code
Generated by Claude Code
Generated by Claude Code