docs(liveness): designer previews count as consumers — re-grade four docs-shaped rows and write the principle into the ledger methodology (#7131) - #7425
Merged
Conversation
…docs-shaped rows (#7131) The ledger recorded "no runtime consumer" for job.label, job.description, translation.label and translation.name. objectui's metadata-admin previews had been rendering all four to a human the whole time. Per the maintainer ruling of 2026-08-10, a designer preview that renders a key to a human is a runtime consumer. The four rows re-grade dead -> live with realm-marked, commit-pinned objectui evidence (@aeb8424b) and a `producer` naming the registerMetadataPreview call plus the surface that resolves it — a preview no registry hands a draft to is a read point that never runs. Nothing about enforce-or-remove moves: all four remain docs-shaped, deliberately KEPT under ADR-0033, and still not authorWarn'd. The README gains the methodology section the ruling asked for, dividing against the existing "an authoring/preview renderer is NOT a runtime consumer" section on what the property CLAIMS rather than on what the surface is. The 2026-07 sweep's ten corrections are not reopened. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016R9de1FqP7NvwKvqXi92Gh
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
github-actions
Bot
added
size/m
documentation
Contributor
📓 Docs Drift CheckThis PR changes 1 package(s): 106 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:
⛔ 7 release-owned page(s) also reference the affected code. These are read-only:
|
This was referenced Aug 10, 2026
os-help
marked this pull request as ready for review
August 10, 2026 11:11
This was referenced Aug 10, 2026
This was referenced Aug 10, 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 #7131
Executes the maintainer ruling of 2026-08-10 (quoted verbatim below and in the ledger README): previews count as consumers.
What was wrong
packages/spec/liveness/job.jsonandtranslation.jsonrecorded "no runtime consumer" for four docs-shaped keys. objectui's metadata-admin previews had been rendering all four to a human the whole time — measured at objectuiorigin/main@aeb8424b:job.labelJobPreview.tsx:257(d.label, falling back to the jobname):313, the preview card's titlejob.descriptionJobPreview.tsx:258(d.description):316, beneath the title when non-emptytranslation.labelTranslationPreview.tsx:67, first choice of the display chain:100, the item's titletranslation.nameTranslationPreview.tsx:67, the fallback whenlabelis unset:100, same titleIn both files
dis the metadata body of the exact type the row covers, and the previews are reachable rather than merely present:previews/index.ts:62/:50register them against thejob/translationtype names, andResourceEditPage.tsx:949resolves the registration and hands the component the draft being edited.The ruling
What changed
Four rows re-graded
dead→live, each with a realm-marked, commit-pinnedevidencestring,evidenceScope: "cross-repo",verifiedAt: "2026-08-10", and aproducernaming theregisterMetadataPreviewcall plus the surface that resolves it — a preview no registry ever hands a draft to is a read point that never runs.Each note records what the re-grade supersedes and what it does not:
job.label— the old wording ("no runtime consumer (sys_job stores name/schedule only)") was true about the scheduler and false as a whole-system claim; for a docs-shaped key the display is the runtime effect, sosys_jobwas never the surface that could falsify it. The note says explicitly thatlivehere does not mean the scheduler acquired a use for it.translation.label— the superseded wording hedged, "no runtime consumer in this repo", and that hedge was never false. What changed is that the cross-repo look was finally taken, which is exactly the blind spotevidenceScope(audit: #4667 liveness 判定的跨仓覆盖核查——两个方向各有一个实锤反例；顺带更正 homePageId 墓碑文案 #4895) exists to expose.translation.name— the re-grade supersedes one clause ("dead as a BODY key — the honest reading of a copy nobody reads"). The row's substantive door/row-column analysis is preserved verbatim and called out as the substance: thenamecolumn is still the live one on the sync path, and the body copy is still not whatauthored-translation-syncreads. The preview reads it; the sync does not.Nothing about enforce-or-remove moves. All four keys remain docs-shaped, deliberately KEPT under the ADR-0033 exemption, and still not
authorWarn'd.Methodology note — a new README section,
Designer previews count as consumers, quoting the ruling verbatim and giving the sweep a mechanical step: enumerate a type's registered preview read points before writing "no runtime consumer", and record their absence when there are none. Two commands make it a lookup rather than a search.It divides against the pre-existing
An authoring/preview renderer is NOT a runtime consumersection on what the property claims, not on what the surface is — for a display key the render is the whole of the declared effect; for a behavioural key a panel echoing the value back still proves nothing. The 2026-07 sweep's ten corrections are explicitly not reopened, and that section gains a short scope pointer so a reader landing there first is not misled. Its heading is unchanged, because several ledger notes cite it by name asREADME §preview-renderer.Verification
pnpm --filter @objectstack/spec check:liveness— green;job13 live / 2 dead → 15 live / 0 dead,translation17 live / 2 dead → 19 live / 0 dead. Repo-local evidence paths stayed at 353/353 resolved and the foreign bucket moved 131 → 135, i.e. all four new citations landed in the cross-repo bucket and none leaked into the local one.Reverse verification — dropped the
objectuirealm marker fromjob.label's evidence and re-ran the gate: exactly one MISSING, namedjob/label, with the local count rising 353 → 354 and the foreign count falling 135 → 134. Restored, green again. The realm marker is load-bearing, as expected.Also green:
packages/specliveness script tests (9 files / 166 tests),pnpm --filter @objectstack/spec typecheck, andnode scripts/check-nul-bytes.mjs.Out of scope, deliberately
The README "Current state" table's
jobandtranslationrows now carry stale counts (13/2 and 17/2) and Notes prose that enumerates the old dead sets. That table's count columns are the subject of unassigned #7377, whose stated method — "for each drifted row read the Note beside it and reconcile the prose with the new numbers" — is exactly what these two rows need. Left untouched to avoid colliding with that lane; commented on #7377 with the delta instead of filing a twin. Nothing fails:readme-table.mtsdeliberately checks the row set, never the count columns.Two ledger notes still record the superseded principle as their ground —
datasource.json's file-level_noteandpermission.json'srowLevelSecurity.labelrow. Flagged in the report on #7131 rather than fixed here; they need re-measurement under the new principle, not a text edit.Generated by Claude Code