finding(types): series chartType is undeclared and SILENTLY STRIPPED - the normalizer reads it first, measured not live, alias-refusal owed #7694

Activity

1. claude

   added

   package: types

   pm:queue

   finding

   priority:p3

   domain:uiobjectui ui stream: fix lands on the published library or apps — objectui execution seatobjectui ui stream: fix lands on the published library or apps — objectui execution seat

   pm:dispatched

   needs:contract-review

   and removed

   pm:queue

   on Sep 5, 2026
2. claude commented on Sep 5, 2026

   

   claudeBot

   on Sep 5, 2026 – with Claude

   ContributorAuthor

   More actions

   Claim: PM loop round 25 — domain:ui execution seat
   Session: session_01KbJQ1y1J12nZxYzFWhP8Q3
   Branch: claude/issue-7694-chart-type-alias-refusal
   Worktree: objectui-issue-7694
   Domain: domain:ui
   File surface: packages/types/src/data-display.ts, packages/types/src/zod/data-display.zod.ts, the affected pin tests, content/docs/plugins/plugin-charts.mdx (stop on breach; explain in the report)
   Container & model: M, mode:subagent, model: fable — node scripts/pm/dispatch-gates.mjs --tier this round returns no path-derived mandate; the fable tier is the clause-② CONTENT mandate. This card is a narrowing of a published accept surface, which is the clearest clause-② shape there is.
   Clause-②: yes
   Serial constraints cleared: None contended. All four in-flight packages/types PRs were read file-by-file this round: #7731 (parity test only), #7732 (complex.ts / complex.zod.ts / chatbot docs), #7730 (zod/base.zod.ts only), #7726 (base.ts, both barrels, base.zod.ts, form.zod.ts, parity test, many docs) — not one touches data-display.ts or data-display.zod.ts. #7684 merged 8fe8e5c16, so the six series keys this card narrows against are on main. Read-coupling declared: #7698 (p2) and #7682 cover the renderer side of the same series; ⛔ this card must not touch a reader.

   ---

   Generated by Claude Code
3. claude commented on Sep 5, 2026

   

   claudeBot

   on Sep 5, 2026 – with Claude

   ContributorAuthor

   More actions

   os-dev-report

   {
     "issue": 7694,
     "status": "done",
     "branch": "claude/issue-7694-chart-type-alias-refusal",
     "pr": "https://github.com/objectstack-ai/objectui/pull/7737",
     "premise_still_valid": true,
     "summary": "Implemented the ruled option A: `ChartDataSeriesSchema.chartType` is a named ALIAS REFUSAL arm pointing at `type`, and `ChartDataSeries.chartType` is a `?: never` tombstone. Exact refusal shipped (one string on both channels, code `invalid_type`, path `['chartType']`): \"Unrecognized key(s) on this chart series: `chartType`. Did you mean `chartType` → `type`? `chartType` is the renderer's INTERNAL spelling of the declared `type` (objectui#7694): `@objectstack/spec`'s `ChartSeriesSchema` lists it as an alias of `type` and refuses it the same way, and nothing on this authoring face writes it. Write `type` (`bar` | `line` | `area`) for a per-series family override. Until this refusal an authored `chartType` was STRIPPED in silence by this non-strict object, so the series drew in the chart's own family — precisely what the author was overriding.\" Both-written is refused at `chartType` alone (not folded, no precedence minted); no reader touched. Convention: the package's named-refusal family in packages/types/src/zod/tombstone.zod.ts — `retirementTombstone` (line 65, z.never + optional + describe) and `handlerKeyRefusal` (line 121, z.custom) — fits an alias by NAME on neither side, so I added a third sibling `aliasKeyRefusal(alias, canonical, surface, detail)` reusing retirementTombstone's z.never PRIMITIVE (no new shape) — chosen over z.custom because, measured on the base with tsx, `z.toJSONSchema` throws on a z.custom arm and represents a z.never arm as `{ not: {} }`; `z.toJSONSchema(ChartDataSeriesSchema)` worked before and still does (11 → 12 properties). Re-measured census (series-level `chartType`, comment-aware `series` array-literal bracket matcher, controls dataKey/name/type/color in the same query) vs the card: docs 0 vs 0 (controls 10/11/2/2), fixtures 0 vs 0 (3/2/0/2), designer inputs 0 vs 0 (the chart registration's `series` is one `code` input, packages/plugin-charts/src/index.tsx:114), src literals 0 vs 0 (13/3/1/0), tests 9 vs 9 (70/48/11/8; comboMarkClick:114 x2, comboFromSeries:59,71,101,116,134 x6, ChartRenderer.specSeries:97) — all five agree with the card. Spec posture measured live on the installed @objectstack/spec 17.2.0: ChartSeriesSchema refuses `{ name, chartType }` with `unrecognized_keys` and the same `Did you mean` sentence; `type` control green. Files touched: packages/types/src/zod/data-display.zod.ts (arm + import; uncontended), packages/types/src/data-display.ts (`chartType?: never` + docblock; uncontended), packages/types/src/zod/tombstone.zod.ts (aliasKeyRefusal + header line), packages/types/src/__tests__/chart-series-chart-type-alias-refusal-7694.test.ts (new pin, blocks a–f), packages/types/src/__tests__/chart-series-keys-7546.test.ts (block (d) restated as the handoff per objectui#7070, header prose), .changeset/7694-chart-series-chart-type-alias-refusal.md (`@object-ui/types: minor`, narrowing stated in release-reader words, own grade), content/docs/plugins/plugin-charts.mdx and content/docs/api/schema-reference.md (one sentence / one clause). Contended files from the brief — zod-mirror-parity.test.ts, zod/index.zod.ts, index.ts — NOT touched; no parity-ledger entry needed (arm z.input is `undefined`, twin is `?: never`, parity suite passed unchanged). PR #7737 is draft, body carries `Fixes #7694` and a Clause-② section (5 lines), label `needs:contract-review` added and read back; PR body read back intact (0 angle-bracket-shaped tokens, footer survived in session form). Assignee os-sam was set by the PM; not touched.",
     "tests": "All on HEAD de8fda691 (branch pushed), exit codes captured before any pipe, under scripts/pm/os-verify-lock.sh (shared box). LIMB ABLATION on the untouched base a00db9ef9 (before any edit), population 304 files / 5817 tests (plugin-charts, plugin-dashboard, types, plugin-report, schema-catalog + the 8 outside test files importing plugin-charts/normalizeChartSchema; vitest alias resolves @object-ui/plugin-charts to src so the mutated SOURCE is what tests resolved — no dist rebuild on the path; each leg: anchor count 1 before, `grep -c` old anchor 0 / new line 1 after, `git diff HEAD --stat` 1 file +1/-1, mutated blob hash printed; restore by `git checkout HEAD -- ABSOLUTE_PATH` in a trap EXIT INT TERM, proven by `git hash-object` == HEAD blob 05a66dfb and empty `git diff HEAD --stat`): leg A delete `str(raw.chartType) ?? ` → `Test Files 304 passed (304) / Tests 5817 passed (5817)`, vitest exit 0 (GREEN — key still not live); leg B delete ` ?? str(raw.type)` → `Test Files 2 failed | 302 passed (304) / Tests 2 failed | 5815 passed (5817)`, exit 1 (RED: normalizeChartSchema.test.ts 'carries stack / yAxis / color on a series', ChartRenderer.specSeries.test.tsx 'honours a spec series[].type override end to end' — the card's same two). VERIFY UNION on de8fda691: packages/types `tsc --noEmit` exit 0; `tsc -p tsconfig.test.json --listFiles` exit 0 with both edited test files listed (2); `tsc -p tsconfig.examples.json` exit 0; `eslint .` exit 0 ('0 errors, 268 warnings', pre-existing no-explicit-any, none on touched lines); `pnpm exec vitest run --maxWorkers=2 packages/types/` → 'Test Files 116 passed (116) / Tests 1991 passed (1991)' exit 0 (incl. zod-mirror-parity); `pnpm --filter @object-ui/types build` → '✓ dist completeness: 1 package(s) complete (120 emitted files verified)', `chartType?: never` present in dist/data-display.d.ts. REVERSE VERIFICATION (fix committed first; data-display.zod.ts + data-display.ts reverted to BASE, helper kept so the instrument loads; predicted direction RED): pin files → 'Tests 15 failed | 37 passed (52)' exit 1; `tsc -p tsconfig.test.json` exit 2 with exactly the 3 predicted errors (TS2339 z.input line, TS2339 ChartDataSeries['chartType'] line, TS2578 unused directive on the non-fresh assignment); restored, both hashes == HEAD blobs, `git diff HEAD --stat` empty. Observed direction: red as predicted. GATES on the committed tree: check:control-bytes '✅ OK (scanned 6302 tracked text file(s))'; check:doc-fences ✅; check:doc-types '✅ Every documented component type is registered'; check:spec-symbols ✅ (3 lines); check:unreferenced-sources 'OK Every shipped source file … reachable'; check:phantom-deps ✅; docs:check-links 'Links are valid across 17 scan roots'; check-changeset-presence '✅ 5 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s)'; check-changeset-no-major ✅; check-changeset-fixed ✅; check-changeset-overwrite '✅ No pre-existing changeset was modified or deleted' (1 added, 0 modified, 0 deleted). DECLARED NARROWINGS (CI owns the farm): check:doc-snippets NOT MEASURED locally — 'PRECONDITION NOT MET (exit 2)', needs 26 packages built; cannot turn red on this diff because the docs edits are prose only (no fence) and the sole type change (`chartType?: never` on a series) can only bite a snippet writing series-level `chartType`, which the docs census reads 0 with lit controls — doc-snippet-types.yml runs it in CI. Downstream consumer type-check (prefix filter '...@object-ui/types' = consumers) not run in full (15-package build closure vs the 10-min cap): zero packages outside packages/types import ChartDataSeries or ChartSchema from @object-ui/types (grep), src-literal census 0 with lit controls; CI type-check job runs the farm. Repo-wide `pnpm lint` / `pnpm test` left to CI. CI status at report time: in_progress.",
     "mcp_calls": "0 — issue read, PR create, label add, read-backs and this report comment all went over REST (curl / urllib with GITHUB_TOKEN); no MCP GitHub call in the whole run",
     "open_questions": [
       {
         "question": "The reader's first limb: `normalizeSeries` still reads `chartType` first on the internal-shape arrays its producers hand it, while the authoring face now refuses the same key. Fence-correct here (no reader touched). Is that limb owed a docblock or a card of its own stating it is an internal-shape read only?",
         "options": [
           "A — leave the reader as is; the internal-shape contract (ChartRenderer's dataKey-shaped arm) legitimately carries chartType",
           "B — file a follow-up card to docblock the limb as internal-only",
           "C — retire the limb at the reader (would change what internal producers render)"
         ],
         "recommendation": "A for now, B if the contract reviewer wants the split written down at the read site; C is a reader decision outside this card"
       },
       {
         "question": "`aliasKeyRefusal` as the third member of tombstone.zod.ts reuses retirementTombstone's z.never primitive (code invalid_type) rather than handlerKeyRefusal's z.custom (code custom), because z.toJSONSchema throws on z.custom and ChartDataSeriesSchema was JSON-Schema-representable before this change. Should an alias arm instead share the `custom` code and give up the JSON-Schema surface?",
         "options": [
           "A — keep z.never: JSON-Schema surface preserved, told apart from a retirement by wording (pinned)",
           "B — switch to z.custom for code parity with handlerKeyRefusal, accepting that z.toJSONSchema(ChartDataSeriesSchema) starts throwing (a Clause-② change of its own)"
         ],
         "recommendation": "A — block (f) of the pin measures the trade; B would narrow a surface that worked"
       },
       {
         "question": "The arm's lead sentence is byte-identical to the spec's today (`Unrecognized key(s) on this chart series: \\`chartType\\`. Did you mean \\`chartType\\` → \\`type\\`?`). The pin asserts the lead and the `Did you mean` fragment, deliberately not the spec's trailing `history` clause, so a spec reword does not turn an objectui pin red. Should the full spec sentence be pinned instead?",
         "options": [
           "A — pin lead + remedy fragment only (as shipped)",
           "B — pin the spec's whole message byte-for-byte (couples objectui's pin to spec rewording)"
         ],
         "recommendation": "A — the remedy fragment is the contract; B trades a spec-bump PR's time for no author-facing gain"
       }
     ],
     "out_of_scope_findings": []
   }

   ---

   Generated by Claude Code
4. claude commented on Sep 5, 2026

   

   claudeBot

   on Sep 5, 2026 – with Claude

   ContributorAuthor

   More actions

   os-dev-report

   {
     "issue": 7694,
     "status": "done",
     "branch": "claude/issue-7694-chart-type-alias-refusal",
     "pr": "https://github.com/objectstack-ai/objectui/pull/7737",
     "premise_still_valid": true,
     "summary": "Land-prep only on the reviewed PR #7737; the PASS stands and nothing behavioural moved. (1) MERGE, never rebase: origin/main had moved seven commits (a00db9ef9..81a2eb1fb — #7621, #7738, #7686, #7732, #7731, #7730, #7726); merge commit 6bcd58ff2, 'Merge made by the ort strategy', 0 conflicts, and `git diff de8fda691..HEAD` restricted to this PR's own files is EMPTY, so the merge moved nothing the PR owns. Final head 3929d1036, pushed fast-forward (de8fda691..3929d1036), no force. (2) tombstone.zod.ts docblock: the falsified census claim is gone, replaced by the re-measured MenuItemSchema.type / #6523 precedent and the real differentiator (composed spec sentence: a vocabulary, not a shape). (3) Pin block (e) now derives the expected lead from the installed spec's LIVE message. (4) Clause-2 items 1 and 2 and the changeset now name the embedding parents and the io-mode nuance. (5) The PR body now cites `tsc -p packages/types/tsconfig.test.json` as the parity instrument, not vitest — and it also carried the same falsified census claim, which I corrected there too (naming objectui#7678). PR stays DRAFT; `needs:contract-review` verified still present on BOTH carriers (PR 7737 and issue 7694); I stripped nothing. TWO PM-SIDE NOTES. (a) The brief's premise 'the branch tip is unchanged at de8fda691, so there is no half-finished work' was half false: the worktree /home/user/objectui-7694-land carried TWO UNPUSHED local commits from the killed prior run (merge f7f706f25 + land-prep c816881be). The remote tip was indeed de8fda691. I saved that work as a patch (scratchpad issue-7694/prior-landprep-c816881be.patch), reset the branch to the remote tip, and redid everything from my own measurements — nothing was reused. (b) The 92-byte lead: measured here through the `?` the shared cut is 91 bytes, not 92; the difference is the single space that follows it. Recorded in the PR body rather than silently adopting either number.",
     "tests": "All on final head 3929d1036 (git rev-parse --short HEAD from the same runs), exit codes captured by redirect BEFORE any pipe, heavy runs through scripts/pm/os-verify-lock.sh (shared box). GATES: `tsc -p packages/types/tsconfig.test.json --noEmit` exit 0 — and NOT the 'ran but measured nothing' trap: `--listFiles` reports 580 files and all three edited paths are listed (the pin test, zod/tombstone.zod.ts, data-display.ts). `pnpm --filter @object-ui/types build` exit 0, 'dist completeness: 1 package(s) complete (124 emitted files verified)'. `pnpm exec eslint .` REPO-WIDE exit 0, '11939 problems (0 errors, 11939 warnings)' — all pre-existing no-explicit-any. `pnpm exec vitest run packages/types/` exit 0, 'Test Files 119 passed (119) · Tests 2054 passed (2054)'. Four changeset gates all exit 0: presence ('5 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s)'), no-major, fixed, overwrite ('1 changeset(s) added, 0 modified, 0 deleted'). Also check:control-bytes exit 0 (6317 tracked text files), check:doc-fences exit 0, check-governed-queue-guard --test over the PR's 8 paths: 'NOT GOVERNED — 8 path(s) checked against 5 governed surface(s); none matched'. PARITY RATCHET LIT ON THE MERGED HEAD: control `tsc -p packages/types/tsconfig.test.json --noEmit` exit 0 on the untouched tree; mutate the TS twin chartType?: never to chartType?: string — mutation CONFIRMED ON DISK before reading anything (old anchor grep -c 1 to 0, injected line 1, blob 9aa36e06212459f6ebf4a5329b1c2f0156c8cd0e differs from HEAD blob 9fb073280cfe57d94446b0d150217dd3ec1e39b7, git diff HEAD --stat = 1 file / 1 insertion / 1 deletion) — same command exit 2 with, among 5 errors, `packages/types/src/__tests__/zod-mirror-parity.test.ts(1522,14): error TS2322: Type 'LedgerMismatch' is not assignable to type 'never'.` naming the pair \"data-display.zod.ts#ChartDataSeriesSchema\". Restored inside `trap ... EXIT INT TERM` against an ABSOLUTE path, proven by git hash-object == 9fb073280cfe57d94446b0d150217dd3ec1e39b7 and an EMPTY whole-tree git diff HEAD --stat. BLOCK (e) CAN STILL FAIL, two directions: (i) OUR side diverges (real tree mutation) — Unrecognized to Unrecognised in the composer, mutation confirmed on disk (old anchor 1 to 0, injected 1, blob 71d581582e7043f9bf9995f4d9a5b5c546236ac8 differs from HEAD d431a1f9b263a76997921426cd4ebc3f86a02d76) — block goes 'Tests 1 failed | 24 skipped (25)', vitest exit 1, 'AssertionError: expected false to be true'; restored under trap, hash-object == d431a1f9b263a76997921426cd4ebc3f86a02d76, git diff HEAD --stat empty. (ii) THE SPEC's side rewords its lead (scratch fixture, no tree mutation): the OLD shape stays GREEN on BOTH halves — exactly the defect the review named — while the NEW shape goes RED; unreworded live control GREEN. Pin file green after the rewrite: 'Test Files 1 passed (1) · Tests 25 passed (25)'. IO-MODE MEASUREMENT (changeset / Clause-2 item 2), against a base stand-in validated byte-for-byte against the real emission in both modes: io input — HEAD 12 properties, additionalProperties absent; base 11 properties, additionalProperties absent (so the accept set genuinely narrows). io output (the default) — HEAD 12 / additionalProperties false; base 11 / additionalProperties FALSE ALREADY (so the key was already outside the accept set; what output mode gains is the named refusal). EMBEDDING PARENTS re-measured on the merged head: ChartSchema.series at zod/data-display.zod.ts:622, ReportSectionSchema.chart at zod/reports.zod.ts:105, safeValidateSchema at zod/index.zod.ts:434, objectui check / validate at packages/cli/src/cli.ts:211 and :223.",
     "mcp_calls": "0 — the whole run went through authenticated REST (repo read, PR read, PR body PATCH, issue read, comment post); zero MCP GitHub tool calls",
     "open_questions": [],
     "out_of_scope_findings": []
   }

   ---

   Generated by Claude Code
5. claude

   removed

   needs:contract-review

   on Sep 5, 2026