feat(types)!: retire ChartDataSeries.data, correct categories' prose (#6896) - #7111
Merged
huangyiirene merged 1 commit intoSep 1, 2026
Merged
Conversation
…6896) `ChartDataSeries.data` was REQUIRED on every authored series and read by nothing. `normalizeChartSchema`'s `normalizeSeries` (`@object-ui/plugin-charts`) reads `dataKey`/`name`, `label`, `chartType`/`type`, `variant`, `opacity`, `dashArray`, `stack`, `yAxis` and `color` — `data` is not among them. Rows come from the chart-level `data`, a key `ChartSchema` never declared at all, which survives only because `BaseSchema` carries an index signature and so suppresses excess-property checking on chart literals. Retired through the ADR-0049 kit rather than deleted: `?: never` on the interface, `retirementTombstone()` on the Zod mirror. `ChartDataSeriesSchema` is a non-strict `z.object`, so a deletion would strip an authored value in silence — one silent no-op traded for another. Immediate tombstone, no dual-reading window, per the maintainer's ruling of 2026-08-31. `categories` is NOT retired. It keeps its behaviour — an alternative series list consulted only when `series` is absent, each entry normalized as `{ dataKey }` — and its prose was corrected to say so, on the docblock, the `ChartDataSeries` header and the Zod `.describe()`. Prose follows machine. The census was re-measured on this branch's merge-base with a control that had to hit in the same query (4 sites in `report-schema-authoring-face.test.ts`). It corrects the record: one non-test authoring site sits inside `packages/` (`packages/types/examples/data-display-examples.json`) and four more in `content/docs/`. All five are unreferenced documentation authored against the documented-but-unimplemented model, not consumers; their migration is filed separately. Pinned in `chart-inline-data-retired.test.ts`, including a counter-probe that builds the deletion this retirement did not choose and measures the contrast in the same run, and in `normalizeChartSchema.test.ts`, where the `categories` read now has behaviour coverage it never had. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012wwHa4aaFybxXrfmfHioDM
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 1, 2026
Merged
Collaborator
契约复审:PASS —— head
|
huangyiirene
marked this pull request as ready for review
September 1, 2026 13:45
huangyiirene
deleted the
claude/issue-6896-retire-static-chart-inline-data
branch
September 1, 2026 13:59
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
3 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 #6896
⛔ Draft, and it stops here. Clause ② applies: this narrows the accept set of a published
@object-ui/typessurface, so it waits on contract review. Not ready, not enqueued, noauto-merge — and the dispatch tier for this build was opus under the maintainer's standing
quota exemption (2026-08-13) after fable returned HTTP 429, so the merge-queue prohibition is
unchanged.
needs:contract-reviewattached with this PR, not before it.Implements the maintainer's ruling of 2026-08-31 (决裁批 #20 ③, 「其他同意」). Both halves land;
the ruling named half (a) alone as not a valid landing.
(a)
ChartDataSeries.data— RETIRED, announceddata: number[]was required on every authored series and read by nothing.normalizeSeries(packages/plugin-charts/src/normalizeChartSchema.ts:232) readsdataKey/name,label,chartType/type,variant,opacity,dashArray,stack,yAxis,color.datais not among them. Rows come from the chart-leveldata— a keyChartSchemanever declared at all, surviving only becauseBaseSchemacarries an indexsignature, which is why no gate ever saw this.
data: number[]→data?: neverplusretirementTombstone()on the Zod mirror. Deletingthe member was the option not taken:
ChartDataSeriesSchemais a non-strictz.object, so adeletion strips an authored value in silence — one silent no-op traded for another. Immediate
tombstone, no dual-reading window.
The accept set moves in both directions, both deliberate:
{ name: 'Revenue', data: [1,2,3] }{ name: 'Revenue' }datarequired)(b)
categories— live, with corrected proseNot retired. It is read as an alternative series list, consulted only when
seriesisabsent, each entry normalized as
{ dataKey }— so the strings name columns to plot. Thecategory axis comes from
xAxisKey/xAxis(:290-292). The docblock said "X-axislabels/categories", so an author following the documentation got a different chart from the
documented one. Prose follows machine: the
ChartSchema.categoriesdocblock, theChartDataSeriesheader (which promised numbers "positionally aligned with the chart'scategories", a model that never existed) and the Zod.describe()now state the read.Behaviour is byte-identical;
categoriesstays writable.The census, re-measured — with a correction to the record
Re-measured at merge-base
2c3cd1b75, with a control that had to hit in the same query. Theinstrument was not blind: it scores 4 authoring sites in
packages/types/src/__tests__/report-schema-authoring-face.test.ts.series[].dataoutside tests acrosspackages//apps//examples/. That is off by one site inside those roots:packages/types/examples/data-display-examples.json— 2 series (insidepackages/)content/docs/api/schema-reference.md— 3 series (outside the scanned roots)content/docs/core/report-schema.mdx— 1 series (outside the scanned roots)The ruling's conclusion is unaffected, and these strengthen it. None is a consumer: nothing
imports, type-checks, parses or renders any of them (
packages/typesships onlydist,README.md,CHANGELOG.md,LICENSE, soexamples/is not published; the mdx site sits in aplaintextfence, so no doc gate covers it). Each authors the documented model — month namesin
categoriesbeside inlineseries[].data— which renders an empty chart today, becausedatais dropped andcategoriesis ignored wheneverseriesis present. They are instancesof the divergence this PR closes, not users of a working inline-data model.
They are out of scope here by file surface and filed separately; until that lands, an author
copying them trips the tombstone and reads the remedy instead of being dropped in silence.
Mechanism note for the reviewer
The dispatch expected an "ADR-0087 retirement registry and its generated baselines". objectui has
no such files: neither ADR-0049 nor ADR-0087 exists under
docs/adr/here — both areupstream
@objectstackADRs referenced by convention, and the retirement "registry" in this repois the declaration pair (
?: never+retirementTombstone()) plus its pin test. That is theshape #6949 landed (changeset, pin test, interface, Zod mirror — no registry file, no generated
baselines), and this PR follows it. Nothing was skipped; the artifacts do not exist.
Verification
Run on the exact commit under review,
9d53bee5d. Exit codes captured before any pipe.vitest(4 affected files)Test Files 4 passed (4) · Tests 70 passed (70)vitestfulltypes+plugin-chartsTest Files 117 passed (117) · Tests 1266 passed (1266)@object-ui/typestype-checkDone—tsc --noEmit && tsconfig.examples.json && tsconfig.test.json@object-ui/plugin-chartstype-checklint(both packages)0 errors(260 + 267 pre-existing warnings)check:changeset-no-majorNo changeset declares a major bump.check:changeset-presence5 source file(s) of 2 released package(s) changed, and this change declares 1 changeset(s)check:doc-snippets271 of 271 block(s) judged, 0 failed(sentinel controls fired)check:doc-typesEvery documented component type is registered.check:spec-symbol-derivation1329 files scanned … 0 untriaged collisionscheck:readme-exports386 self-imports judged (386 real, 0 wrong-path, 0 fabricated)check:control-bytesOK (scanned 5883 tracked text file(s))check:phantom-deps/self-import/vi-mock-specifiers/esm-specifiersTwo gates first returned NOT MEASURED, and neither is reported as a pass:
check:doc-snippetsanswered
PRECONDITION NOT MET (exit 2)andcheck:readme-exportsexit 1 withtype entry not on disk— both the unbuilt dependency closure. Both were re-run to exit 0 afterbuilding what they name.
plugin-chartstype-checkfailed the same way (TS2307 on@object-ui/core/components/react) before its closure was built.@ts-expect-errorenforcement is real here and was verified rather than assumed:tsc -p tsconfig.test.json --listFilesconfirms both the new pin and the edited pin are inside theproject, so a re-widened declaration fails the build on the unused directive.
The counter-probe, and the ablation that proves it fires
chart-inline-data-retired.test.tsbuilds the deletion this retirement did not choose — thesame series schema with
datasimply absent — and measures the contrast in the same run: thedeletion accepts an authored
dataand strips it silently, the shipped schema refuses and sayswhy. Announcement is pinned separately (
datastays in.shape; the description names both theremedy and
xAxisKey), so "gone" cannot pass as "retired".Ablated to prove it is not vacuous. The tombstone was removed from the Zod mirror; the mutation
was confirmed on disk —
data: retirementTombstone(occurrences dropped to 0 and the blobhash moved from
71a1b768— and the pin then went red: 5 failed / 10 passed, the failuresbeing exactly the refusal, the whole-node refusal, the shape/announcement pin, the counter-probe
and the
categories-contrast pin. No rebuild leg applies: the pin imports the mirror by relativesource path (
../zod/data-display.zod), not throughdist, so nothing stale can answer for it.Restore was proven the same way rather than by exit code — blob hash back to
71a1b768,git diff HEADempty, tombstone present again — and the pin re-run green (70/70) on that restoredtree, which is the commit under review. The mutation ran under a
trap ... EXIT INT TERMwithabsolute paths.
Generated by Claude Code