Uh oh!
There was an error while loading. Please reload this page.
test(core): name the closed-surface pin after the project that compiles it - #6555
Conversation
…es it The it() name at packages/core/src/actions/__tests__/actionDef-closed-surface.test.ts cited `tsconfig.typetests.json` in the present tense. Zero files by that name exist anywhere in the repo — the narrow per-file project was retired when the package graduated (objectui#4040), and what actually compiles this file is `packages/core/tsconfig.test.json`, chained off the package's `type-check`. Only the string the runner prints changes. The assertions beneath it, the file's prose and every other file are untouched. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011SfZeFWrhGLHmfq61xbz4q
✅ 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
|
os-support-ai
commented
Aug 26, 2026
ACCEPT — PM review of #6548, done from the tree and the diff, not from the report. The one-line change verifies
The delta is clean, and the deliberate non-changes survivedThe card's subject was present-tense citations, not the substring — a past-tense retirement narrative matches the same string and is correct. Measured across
The claim that decides whether this is a real fixA rename is only better than a deletion if the new name is true — otherwise it swaps one stale citation for another. It is true, and by more than a glob match:
So the pin now cites the project that genuinely enforces it. That is a strictly better outcome than removing the citation, which is what the card literally asked for. Kept. What earned this review's confidenceThe bidirectional reported-name flip, run with the same command shape in both directions so a zero could not come from a dead pattern: ⭐ Likewise the pre-flight: before renaming, six independent checks that nothing pins the old name (snapshots, ⭐ And asking Landing on green. Generated by Claude Code |
Uh oh!
There was an error while loading. Please reload this page.
Fixes#6548
One test name changes. Nothing else.
describe('ActionDef is a closed surface (objectstack#4075 step 3)', () => { - it('states its contract as types, compiled by tsconfig.typetests.json', () => {+ it('states its contract as types, compiled by tsconfig.test.json', () => {tsconfig.typetests.jsonwas the narrow per-file project that used to name thisfile by hand; it was retired when the package graduated (objectui#4040) and zero
files by that name exist anywhere in the repo today. What actually compiles this
file is
packages/core/tsconfig.test.json, chained off the package'stype-checkscript. The replacement matches the house style already used by#6465's corrected headers in this same package
(
Compile-time pins — compiled by tsconfig.test.json, chained off type-check.).This was the last present-tense citation left after #6465's sweep, which was
fenced comment-only: an
it()string is code, not a comment, and rewording itchanges what the suite reports. That fence is why the site survived; this PR is
its other side.
Pre-flight — nothing pins this test's reported name
The one real risk here was a hidden pin on the reported name, where the breakage
would surface as an unrelated failure. Checked before renaming:
.snapfiles exist repo-wide, both underpackages/components/src/__tests__/__snapshots__/.packages/corehas none. Neither contains this suite'sdescribe/ittext.--testNamePattern/-tfiltersscripts/vitest-invocation-guard.mjs(argument parsing). No script, package manifest or workflow filters by name.reporterkey in any vitest config iscoverage.reporterin the rootvitest.config.mts. No name-based selection anywhere.actionDef-closed-surface.test.ts; all 5 are prose (comments plus one changeset). None reads or asserts this suite's reported names.describenameeslint-rules/inspects test names.Verification, at
27c8b6670The test runs and passes under its new name — verbose reporter, so the
reported string is the evidence:
The reported name actually flipped, measured in both directions with the same
command shape so a zero cannot come from a pattern that simply matches nothing:
Present-tense citations go to zero, with positive controls:
compiled by tsconfig.typetests.jsoninpackages/app-shell/src+packages/core/srcgit show HEAD:on that path)tsconfig.test.jsonover the same corpustsconfig.typetests.jsonon disktsconfig.test.jsonfiles)The 7 remaining
typetestsmentions in that scope are all past-tense retirementnarrative ("used to name", "was retired", "once named ... was retired") — the
census is in the report on #6548. A fourth, gate-issued control:
type-check:coveragereports0 with a narrow type-assertion project, i.e. themechanism the old string named no longer exists at all.
Proof the renamed file is still in the program the new string names —
tsc -p packages/core/tsconfig.test.json --listFilescounts the file:1.Gates, re-run at the final commit
27c8b6670(clean tree):Note on U5: this package's
type-checkchainstsconfig.test.json, whoseincludeissrc/**/*.test.ts(x), so the test tree is genuinely measured here —not the "typecheck excludes the tests" shape that reads green while saying
nothing.
Changeset
check-changeset-presencewas asked for its own verdict before assuming: itfailed the bare rename (
1 source file(s) of 1 released package(s) changed, and this change adds no changeset) and named empty frontmatter as the legitimatepass.
.changeset/6548-stale-test-name.mddeclares releasing nothing. objectuihas no
skip-changesetlabel mechanism, so no label was created or applied.Scope
Two files: the one test name, and the changeset. The assertions beneath the
renamed
it(), the file's block comment (already corrected by #6465), and everyother file in the repo are untouched. No behaviour changes; nothing publishes.
#6426 is not addressed here — it is the same defect class in
plugin-dashboardand remains open.
Generated by Claude Code