Uh oh!
There was an error while loading. Please reload this page.
docs(vscode-extension): remove SUMMARY.md, which claimed a CodeQL scan this repo never ran - #6276
Conversation
… file Adds the changeset and re-anchors the objectui#4938 floors in check-doc-links.test.ts to the population left after the SUMMARY.md removal.
✅ 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
|
yinlianghui-tw
commented
Aug 25, 2026
PM review — ACCEPT, and ⭐ the override of my lean is UPHELD on its evidenceReviewed by the Delete over correct — you were right, and I checked every leg myselfMy dispatch leaned "correct, not delete", gated on the reference/packaging evidence. You gathered it and it points the other way. Independently re-measured before this review:
That last row is what decides it. The card saw one false sentence; the file is a frozen session status report whose every checkbox is an unverifiable point-in-time claim. Correcting it has no stable endpoint — a maintainer would have to re-measure every tick forever, which is the exact rot mechanism #6274 just pinned on The floor edits — correct, and correctly minimalThe deletion broke two #4938population floors in
Evidence qualityNarrowing declared and measured with the universe read from vitest's own config ( #6275 — good catch, right scopeA Retiring the package — noted, not mine to schedule@os-zhuang's 「此软件包应该作废」 is consistent with everything measured here (private, zero tests, unreferenced), but retiring a package is a maintainer/triage decision. Escalated in session; this PR neither advances nor blocks it. ⛔ Not armed yetWaiting for every check to carry a conclusion, then mark ready → arm — sequenced against #6026's landing per the collision note above. Generated by Claude Code |
Uh oh!
There was an error while loading. Please reload this page.
Fixes#5965
Deletes
packages/vscode-extension/SUMMARY.md. Verified onorigin/mainef2a3bd8dbefore editing:grep -rni codeql .github/exits 1 (no matches) and there is nocodeql.ymlamong the 21 workflow files, so the premise holds — the file asserted a run outcome for a scan that does not exist.The repo's canonical fact already says the opposite.
CONTRIBUTING.md:393:Remove rather than correct — the evidence both ways
The card left this open, and the dispatch leaned correct. I went the other way; here is the evidence in both directions.
Nothing references it.
git grep -n 'SUMMARY\.md'across all tracked files returns exactly one hit, and it is a different file (SECURITY_FIX_SUMMARY.md, filed separately below). Zero inbound links to this file.It never shipped. The package is
"private": true(never published to npm), and.vscodeignorereads*.md/!README.md— SUMMARY.md was excluded from the VSIX. It reached no user through either channel.The false claim was not one line. Correcting "the security section" would have left three more instances of the same assertion:
✅ **安全扫描**/- CodeQL扫描通过/- 无安全漏洞- [x] 安全扫描通过(release checklist)✅ **高质量代码** - 通过审查和安全扫描And the same class of unverifiable assertion runs through the rest of it —
- [x] 单元测试通过in a package with zero test files (find -name '*.test.ts' -o -name '*.spec.ts'→ 0), plus✅ 代码完成度: 100%,✅ 文档完成度: 100%, and "扩展已准备好发布到VSCode Marketplace". This was a frozen one-off session status report, not living documentation; there is no state it could be corrected into that stays true without a maintainer re-measuring every tick.Deleting loses no unique content. Its durable reference material — command list, snippet prefixes, configuration keys, project structure — is already carried by
README.md(232 lines),DESIGN.md(380), andPUBLISHING.md(270); the file's own "文档资源" section says as much.Precedent and maintainer signal.
SECURITY_FIX_SUMMARY.md, an artifact of exactly this class, was already deleted from this repo inea72f1886. And on the issue, @os-zhuang:vscode不是开发方向，此软件包应该作废. Scope here stays SUMMARY.md only — retiring the package is not this card.Had either reference-or-shipping test come back positive, the correct move would have been to correct the section to name the gates that actually run; both came back negative, so the file goes.
The floors this moved
scripts/__tests__/check-doc-links.test.tscarries twoobjectui#4938population floors measured "the day the row landed" (15 files = 12 underpackages/*+ 3 underapps/*). Removing one markdown file underpackages/*took them to 11 and 14, so both failed:Lowered each by exactly this one deliberate deletion, with the reason recorded inline. They are floors against the scanner silently stopping, not pins on today's file list, so they still fail if the walk breaks. The companion
decidable >= 4assertion was unaffected and still passes — SUMMARY.md contributed no decidable links.Changeset
Added
.changeset/5965-vscode-summary-codeql-claim.mdwith empty frontmatter — no package bump.check-changeset-presence.mjsdoes not decide this (it exits 0 either way); its printed verdict classifies the diff:3 file(s) changed, 0 of them published source of a package the release covers. Reasoning independently:object-uiis in the 39-packagefixedgroup andprivatePackages.versionistrue, so it is version-managed — but it isprivate: truewithtag: false, so nothing publishes, and the deleted.mdwas excluded from the VSIX anyway. The other two files are a test and the changeset itself. Nothing user-visible changes, so no bump is owed; the empty-frontmatter form is this repo's established way to say that out loud (101 of 354 current changesets use it, including4938-timeline-dead-example-link.md— the very change that set the floors above).Verification
All run on the final commit
ff19626b7, clean tree.lint:rootwas run unnarrowed; all 28 warnings are pre-existingno-explicit-anyin files this PR does not touch.Exit codes were captured by redirect before any pipe, and each line above quotes the gate's own printed verdict.
Vitest — declared narrowing. The root suite (
pnpm exec vitest run) could not complete in the foreground: it was SIGTERM'd at the container's ~10-minute cap (exit 143) while sprayingECONNREFUSED 127.0.0.1:3000from live/e2e specs unrelated to this diff. Narrowed instead, and declaring it:pnpm exec vitest list --filesOnly→ 1995 test files.vscode-extension(6) plus the doc-link/changeset/doc-type gate tests: 9 files, 294 tests, all passing. Re-run on final headff19626b7for the ratchet family: 5 files, 180 tests passing.SUMMARY.md(grep hits: 0), and no source module is touched, so no test outside the ran set can change verdict through this diff.CI runs the full farm regardless.
Out of scope — filed, not fixed
packages/core/src/validation/validators/object-validation-engine.ts:221has@see …/blob/main/SECURITY_FIX_SUMMARY.md, deleted inea72f1886; the link 404s and ships in the published@object-ui/core.#5408is not addressed here and remains open — it already fixedREADME.md.Generated by Claude Code
Generated by Claude Code