Uh oh!
There was an error while loading. Please reload this page.
fix(types,plugin-markdown,plugin-kanban): one authority for MarkdownSchema and the in-package Kanban pair - #6971
Merged
Conversation
…Schema` and the in-package Kanban pair `MarkdownSchema` converges onto `@object-ui/types`: the plugin's second declaration differed on one member (`content`, optional there and required in types), measured as drift rather than a semantic difference — the plugin's own registration marks the input `required: true` (test-pinned), its renderer props type is non-optional, the Zod mirror is `z.string()`, and every authored `type: 'markdown'` node supplies it. plugin-markdown now re-exports the one authority. `KanbanCard` / `KanbanColumn`: the three in-package declarations converge to one. An AST comparison found `KanbanImpl.tsx` and `KanbanEnhanced.tsx` strict subset copies of `./types` with nothing typed differently, so their extra members (`cardSubtitle`, `cardFieldCells`, `coverImage`, `collapsed` — all optional) move onto the canonical declaration and both files re-point at it, keeping their export surface via re-export. Baseline `one-authority-per-exported-name-6273`: `MarkdownSchema` row deleted; the `KanbanCard` / `KanbanColumn` rows shrink from four sites to the two cross-package ones, which are escalated rather than guessed. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013hfmP9hoMd3dJwTh85J4yB
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 Aug 31, 2026
… scope the Kanban authority note C1: the "Schema API" block in packages/plugin-markdown/README.md still spelled `content` as optional — the exact outlier spelling this branch removes, in a doc that ships in the npm tarball. The canonical declaration is non-optional: `content: z.string()` at packages/types/src/zod/data-display.zod.ts:276 and `content: string` at packages/plugin-markdown/src/MarkdownImpl.tsx:31. No gate catches this; check-readme-exports judges export existence, not shape prose. C2: KanbanEnhanced.tsx's authority comment now carries the "in this package" qualifier its KanbanImpl twin already spells. The cross-package Kanban pair is deliberately open, so the unqualified wording overstated what this branch settles. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013hfmP9hoMd3dJwTh85J4yB
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
|
ContributorAuthor
C1 and C2 are discharged in 892625f. Still draft, nothing else touched. C1 (was blocking) — - content?: string, // Markdown content (supports GitHub Flavored Markdown)+ content: string, // Markdown content (supports GitHub Flavored Markdown)Re-derived against the declaration rather than taken on the review's word: C2 — -// One authority for these two names, in `./types` (objectui#6172 / #6155).+// One authority for these two names in this package: `./types` (objectui#6172 / #6155).matching Verification at 892625f, every command from the repo root:
The shrink-only baseline Generated by Claude Code Generated by Claude Code |
os-sam
marked this pull request as ready for review
August 31, 2026 07:57
Uh oh!
There was an error while loading. Please reload this page.
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 freeto 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.
Refs #6172 — two of the four names converge here.
MarkdownSchemais fullyconverged;
KanbanCard/KanbanColumnconverge inside@object-ui/plugin-kanban(three declarations to one).
KanbanSchema, and the remaining cross-packageKanbanCard/KanbanColumnpair, are escalated rather than guessed — see"What is NOT resolved here". A closing keyword is deliberately not used.
⛔ Draft by dispatch order. Do not mark ready, enqueue, or enable auto-merge.
Verified on
10340f225, clean tree.MarkdownSchema— converged, and the fork branch was tested and rejectedThe ruling allowed forking the name "if it turns out to be a real semantic
difference, not a drift." It is drift. Five independent statements in this
repository already say
contentis required; the plugin's own interface was theonly one saying otherwise:
plugin-markdown/src/index.tsxregistration —{ name: 'content', required: true }plugin-markdown/src/index.test.ts:44— pinscontentInput?.required === trueMarkdownImpl.tsx—MarkdownImplProps.content: string(the renderer that consumes it)types/src/zod/data-display.zod.ts:276—content: z.string(), not.optional(), parity-testedtype: 'markdown'node in the treeplugin-markdown/src/types.ts—content?: stringCorpus measurement, because this narrows the accepted set (clause ②, the
irreversible direction). 34
type: 'markdown'object literals in the tree, brace-matched rather than line-grepped. Every one that omits
contentis not a markdownnode: they are rich-text FIELD metadata (
MarkdownFieldMetadata,packages/types/src/field-types.ts:198— a different type withmax_length), plusone source comment and two doc restatements of the declaration itself. Zero
authored markdown nodes omit
content, so the narrowing changes nothing in-tree.Direction is forced, not chosen:
@object-ui/typesis the zero-workspace-dependencybottom layer (
dependencies=@objectstack/spec,zod), so it cannot re-point at aplugin without a cycle.
plugin-markdownalready depends on it, so no new edge.classNameis not lost — it is declared byBaseSchema, which both copiesextended, so it was always inherited rather than added by the plugin. The plugin's
face does gain
sanitizeandcomponents; both are optional, and neither is readby any renderer — sanitization is unconditional (
rehypeSanitizeis a fixed link inthe rehype chain). That deadness is recorded as a finding, not fixed here.
KanbanCard/KanbanColumn— the in-package ×3 converge to ×1Remedy picked by a TypeScript-AST probe (heritage clause, member set, and each
shared member's declared type and optionality), not by reading:
types.tsvsKanbanImpl.tsx—KanbanCardtypes.tsvsKanbanEnhanced.tsx—KanbanCardtypes.tsvsKanbanImpl.tsx—KanbanColumntypes.tsvsKanbanEnhanced.tsx—KanbanColumnNothing typed differently anywhere ⇒ stale copies, so the re-point branch, exactly
as #6936 treated the Breadcrumb pair. The extras (
cardSubtitle,cardFieldCells,coverImage,collapsed) move onto the one declaration in./types; all four areoptional, so every value that type-checked before still does. Both
.tsxmodules keeptheir previous export surface through
export type { … } from './types', so no importpath changes.
badgesas "typeddifferently" between
types.tsand both.tsxcopies. That was an artifact — JSDocprose inside the type literal, and a trailing member separator. After normalising
comments and separators the difference is zero. Comparing raw
getText()wouldhave produced the wrong verdict (rename instead of re-point).
Why direction (a) was not used
The gate counts
export type X =/export interface Xas an authority anddeliberately does not count
export type { X } from …. A derived downstreamdeclaration is still a declaration, so it cannot deliver the baseline deletion. The
live proof is on the tree:
ComponentMetawas converged the (a) way by PR #6297 andits
KNOWN_COLLISIONSrow is still there today.The shrink-only baseline
Re-derived on this branch's merge-base rather than inherited: 39 rows at
main = 8f1d995ef447…— the card body's table says 46 and is seven rows stale.MarkdownSchema— row deleted (39 → 38). One authority remains.KanbanCard/KanbanColumn— rows shrunk from four sites to two.reconcilecompares sites, so leaving the four-site rows would have failed the
staledirection. The two survivors are the cross-package pair.
KanbanSchema— unchanged.Ablation — direction predicted in writing before each run
Both legs mutate the fact, never the assertion, under
trap … EXIT INT TERMwithabsolute paths from
git rev-parse --show-toplevel, restoring withgit checkout HEAD -- ABSOLUTE_PATH.Leg A — predicted: restoring the deleted
MarkdownSchemarow while the collisionis gone reds in the
staledirection. Observed exactly that:MarkdownSchema — no longer collides at:, exit 1, 1 failed / 10 passed.Leg B — predicted: reverting the source convergence while the row stays deleted
reds in the
freshdirection. Observed exactly that:MarkdownSchema — a NEW colliding name, exit 1, 1 failed / 10 passed.Mutation proved on disk both times (anchored count moved 0→1 / re-export 1→0, and
git hash-objectmoved off theHEADblob); restore proved both times (hash back tothe
HEADblob andgit diff HEADempty). Together the two legs show the gate isnon-vacuous in both of its directions for this change.
What is NOT resolved here, and why it is not a guess
The cross-package
KanbanCard/KanbanColumn/KanbanSchemacollision between@object-ui/typesand@object-ui/plugin-kanbanstays open.These are two genuinely different dialects, not a stale copy —
itemswhere the pluginsays
cards,labelswhere it saysbadges;KanbanSchemashares 3 keys of 7 and 19with
columnsrequired on one side and optional on the other. That is theActionSchemasignature, whose remedy is rename — and because the bottom-layerpackage cannot re-point at a plugin, a rename here means removing a published name
from one of the two packages.
The 2026-08-25 ruling named a cross-package authority for
FormField("the@objectstack/speccopy is the contract-side authority") but its Kanban clause isabout in-package multi-declarations. It does not say which package owns the bare
name here, so that authority call is escalated on the card rather than made in a PR.
Measurement is in hand for whoever rules: the plugin's dialect is the live one —
authored corpus writes
cards13× againstitems1× (and that singleitemssite is the types copy's own demo file); every registered renderer (
kanban,kanban-ui,kanban-enhanced,object-kanban) consumes the plugin shape; and thetypes copy's distinguishing keys
draggable/onColumnAdd/onCardAddare read bynothing. Against that, renaming the
@object-ui/typestrio touches its barrel,registry.ts'sComponentSchemaMap, theComplexSchemaunion, the Zod mirror and itsparity test, plus two docs pages — a published-surface removal well outside this
dispatch's declared file surface.
#6155 is only partially addressed and stays open. It measured
KanbanCard/KanbanColumnas declared four times; after this diff they are declared twice. Theremaining two are the cross-package pair above.
Docs truth kept in step
content/docs/plugins/plugin-markdown.mdxasserted "Declared by@object-ui/plugin-markdown… the plugin ships its own copy of this name." This diffmakes that false, so the page is corrected in the same PR: it now names the declaring
package and file path (a checkable existence claim), records that
contentis requiredand why, and documents
sanitize/componentsas declared-but-unread rather thansilently listing them as behaviour.
Gates — exit codes captured by redirect BEFORE any pipe
scripts/__tests__/one-authority…+packages/{types,plugin-kanban,plugin-markdown}/10340f225type-check, the 3 affected packagesScope: 3 of 47 workspace projects(not a zero-match)check:control-bytescheck:doc-snippetscheck:doc-fencescheck:doc-typesdocs:check-linkscheck:self-importcheck:phantom-depscheck:esm-specifierscheck-changeset-presence(direct)check-changeset-no-major/-fixed/-overwriteCoverage measured, not assumed.
check:doc-snippetswas first run against anunbuilt tree and returned exit 2 —
PRECONDITION NOT MET, which the gate itselflabels "I could not run, NOT I ran and found errors." It was re-run after building the
32 packages its own
--build-filternames; only that second run is quoted. And theedited page is genuinely in the covered set:
content/docs/plugins/plugin-markdown.mdxappears 0 times in the script's declared 43-page ungated list.
Declared narrowing.
check:readme-exportsexits 1 on this tree, and it isnot attributable to this change: every one of its findings is
type entry ./dist/index.d.ts is not on disk — run pnpm build firstfor packages thisbranch never touches. After building the two edited plugins, judged self-imports rose
85 → 91 with 0 wrong-path and 0 fabricated, and no failure line names either
edited README. The repo-wide
pnpm lintwas likewise not run here; CI runs the fullfarm regardless.
Generated by Claude Code