refactor(spec)!: finish #5775's SDUI props count — 4 keys declared, page:tabs.type renamed to tabStyle (#6776) - #6868
Merged
Conversation
…abs.type → tabStyle (#6776) #5775 recorded that "the rest of the keys the renderers honour are declared". Re-counting against objectui origin/main (b3439f4) found five more: for each, objectui's published manifest called the key legal, validateComponentProps (#5068) called it undeclared, and the renderer honoured it anyway — #5435's shape with the wrong half on the spec side. Declared (no behaviour change, the contract catching up): - page:header.recordChrome / .showStar / .showCopyId - page:accordion.variant Renamed (tombstone + ADR-0087 D2 conversion + RETIRED_KEYS_BY_MAJOR[17]): - page:tabs.type → tabStyle. A props key named `type` collides with the page component's own dispatch key: SchemaRenderer refuses to hoist properties.type, sdui-parser's BASE_PROPS skips it before any validation, and in a flat/JSX carrier `type` is the tag name. Unauthorable in every carrier but the nested properties object, and unvalidated even there. Route A per the tier-3 ruling; all three objectui premises re-verified at origin/main before implementing. Also: mapPageComponents now walks pages[].slots.<slot>, not only regions[].components[]. PageSchema.slots is a closed map of seven declared PageComponent slots and packages/lint's walkPageComponents has always visited both. page:tabs IS one of those slots and all four in-repo sites are slots.tabs, so region-only reach would have made the tombstone's `os migrate meta` prescription a false promise. Coupled factual correction (comment-only) in packages/lint: the falsified "rest of the keys are declared" sentence in authoring-rules.ts, and this card added to validateComponentProps' error-upgrade inventory alongside #5728. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011sGk4SKHqGRgmmqUok1P8M
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
Contributor
📓 Docs Drift CheckThis PR changes 3 package(s): 113 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:
|
github-actions
Bot
added
documentation
…-to-tab-style (#6776) The Check Changeset gate (#6148) requires a breaking changeset to answer the ledger question in writing. This one's answer is 'registered': the conversion entry, its D3 chain step and the RETIRED_KEYS_BY_MAJOR[17] key all land in this PR. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011sGk4SKHqGRgmmqUok1P8M
…ponent-props-declarations # Conflicts: # packages/spec/src/migrations/registry.ts
os-zhuang
marked this pull request as ready for review
August 9, 2026 04:07
This was referenced Aug 9, 2026
os-zhuang
pushed a commit
that referenced
this pull request
Aug 9, 2026
`main` gained #6868 (`page-tabs-type-to-tab-style` + `ui/PageTabsProps:type`) after this branch's last CI head. The merge itself was clean — git placed the three siblings' entries at different offsets — but the generated artifacts are driver-deferred and were regenerated from the merged tree rather than text-merged. All three protocol-17 siblings verified present after regeneration: `notification-list-cursor-retired` (this PR), `action-descriptor-is-async-retired` (#6862) and `page-tabs-type-to-tab-style` (#6868, a D2 conversion). `authorable-surface/api.json` and `authorable-defaults/api.json` pick up `api/ApiRoutes:email` and `api/MetadataEndpointsConfig:maskObjectFields` from other PRs merged in the same window — this branch's copies were simply behind. check:generated 10/10. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011sGk4SKHqGRgmmqUok1P8M
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 #6776
#5775 reconciled
ComponentPropsMapwith the renderers that serve it, then recorded that "the rest of the keys the renderers honour are declared". That sentence did not hold. Five more author-facing props are read by objectui's renderers and were undeclared here, so for each of them three platform authorities disagreed at once: objectui's published manifest and generatedsdui-intrinsics.d.tstold an author (very often an AI author, ADR-0033) the key was legal,validateComponentProps(#5068) reported it undeclared, and the renderer honoured it anyway. That is #5435's shape, with the wrong half on the spec side this time — and these five are what blockvalidateComponentProps's error upgrade alongside #5728.Per-prop itemization
page:header.recordChromePageHeaderProps(component.zod.ts)validateComponentPropsreportscomponent-props-unknown-keyz.boolean().default(true)—falsedrops the record chip for the bare heading a non-record page wantspage:header.showStarPageHeaderPropsz.boolean().default(true)— the follow (favourite) star beside the record titlepage:header.showCopyIdPageHeaderPropsz.boolean().default(true)— the copy-record-id button beside the record titlepage:accordion.variantPageAccordionPropsitems/allowMultiple/aria)z.enum(['flush','card']).default('flush')—flushdraws the divider,cardleaves the border to each panel's contentpage:tabs.tabStylePageTabsPropstype, which no author can write in a flat or JSX carriertabStyle;typebecomes aretiredKey()tombstone + ADR-0087 D2 conversionpage-tabs-type-to-tab-styleRoute A for
page:tabs, and whyPer the tier-3 ruling on the card. The concept was declared all along; the spelling was unauthorable, because a props key named
typecollides with the page component's own dispatch key. Three consequences, each re-measured on objectuiorigin/mainb3439f4 before implementing (the card's baseline was c4c0ac8 — all three survived the drift):origin/mainb3439f4containers.tsx:381reads both spellings, spec's firstconst type: 'line' | 'card' | 'pill' = schema?.properties?.type || schema?.tabStyle || 'line';SchemaRenderer.tsx:251-270refuses to hoistproperties.type:253comment "never let innerproperties.type/properties.idshadow the outer component descriptor", naming tab visual style as the case;:264if (k === 'type' || k === 'id') continue;sdui-parserBASE_PROPSincludes'type'validate.ts:20-30set contains'type';:68if (BASE_PROPS.has(key)) continue;runs before the unknown/typed checksSo
typeis unauthorable in the flat and JSX carriers (where the node reads{ type: 'page:tabs', ... }andtypeis the tag name), and unvalidated even in the nestedpropertiesobject.tabStyleis what objectui's registry publishes as the designer input (containers.tsx:657) and what the renderer reads in every carrier. Converging on the spelling that works, rather than the one that declares well, is #5775'sdisplayFieldtolabelFieldagain; one spelling rather than two is Prime Directive #12. Option B (declaretabStyleas an alias) was refused for the reason the card gives: the dialect that would survive is the one that silently fails to validate.Corroboration the card did not have: objectui's registry
inputsfor all three components already publish exactly the five names this PR declares (tabStyle,variant,recordChrome,showStar,showCopyId). After this lands, spec and objectui's manifest agree on every one.The conversion's reach — one shared-walk fix, and why it was not optional
mapPageComponents(the conversion layer's page walk) now visitspages[].slots.SLOTas well aspages[].regions[].components[]. Its own comment used to call region level "the whole surface", reasoning that everything else lives inside a free-formpropertiesbag.PageSchema.slotsis the counter-example: a closed map of seven named slots, each declaredz.union([PageComponentSchema, z.array(PageComponentSchema)])— as typed as any region component — andpackages/lint'swalkPageComponentshas always visited both, so every conversion reached strictly less than the lint rule that judges its result.#6776 is where that gap cost something.
page:tabsis one of those slots, and all four in-repo authoring sites areslots.tabsonkind: 'slotted'record pages:packages/platform-objects/src/pages/sys-organization.page.ts:51packages/platform-objects/src/pages/sys-user.page.ts:139packages/platform-objects/src/pages/sys-position.page.ts:49examples/app-showcase/src/ui/pages/project-detail.page.ts:38A region-only rewrite would have left
os migrate meta --from 16unable to touch the only shape this key is written in, while the tombstone's prescription promised it could — a false prescription is the ADR-0078 shape the tombstone exists to avoid. Every other page-component conversion gains the same reach, in the direction it already declares; their fixtures are region-based, so nothing else moves. All four sites are migrated totabStylein this PR.One consequence worth stating plainly: because
PageComponent.propertiesis an open bag,tscdoes not catchproperties.typeat a slot authoring site. The tombstone's compile channel reaches a site only wherePageTabsPropsis the declared type; for slotted pages the audible channel is the parse (andvalidateComponentProps, pinned below).Coupled factual correction in
packages/lint(declared in the claim)Comment-only, no rule behaviour changed, no other lint edits:
authoring-rules.ts— the "the rest of the keys the renderers honour are declared" sentence this card falsifies, rewritten to what actually happened.validate-component-props.ts— the same claim in the rule's own header, plus this card added to the error-upgrade precondition inventory alongside component.zod.ts 的 I18nLabelSchema 键声明为纯字符串,但 3 个平台页在 31 处授权内联多语言 map #5728. That comment IS the inventory the upgrade is measured against, so a count reporting itself complete while it is not is exactly how the second half went unnoticed for a release.Tests
component.test.ts— defaults and value refusals for the four new keys; the console preview sample (recordChrome: false) parsed verbatim;tabStyleaccepted for all three values;typerefused by name with the prescription (/`type`.*removed.*`tabStyle`/s), not a bare.toThrow()— an undeclared key on this non-strict schema is stripped silently, and a throw-only assertion cannot tell the two apart.conversions.test.ts— the discriminator (the entry keys oncomponent.typewhile rewritingproperties.type, so "whichtype" is the whole correctness question, pinned by asserting the node's owntypeis untouched); slot reach for both the single-component and array forms with their path spellings;#4923both-keys, in both the agreeing (twin dropped) and disagreeing (both kept, no notice) directions; idempotence; and the acceptance face in both directions, withPageSchemagreen on both spellings because the open bag never judged the key at all.validate-component-props.test.ts— the five shapes that used to warn now report nothing, and the retiredtypereportscomponent-props-invalidcarryingtabStylerather thancomponent-props-unknown-keycarrying nothing.Reverse verification
Deleting the
retiredKey()tombstone while keeping the rename turnspnpm --filter @objectstack/spec check:authorable-surfacered, naming the key:Restoring the tombstone returns it to green (
check:authorable-surface PASS).Worth stating because the expected direction was predicted wrong before the run: the gate that fires is the disappeared-key check (a) — the silent-strip trip wire — not the
RETIRED_KEYS_BY_MAJORregistration check (b2) that this PR's registry entry feeds. (b2) rejects an entry naming a key that is still LIVE; here the key is neither live nor tombstoned, so (a) catches it first and (b2) is never reached. The registry entry is still load-bearing in the other direction — it is what lets the[RETIRED]line exist inauthorable-surface/ui.jsonat all — but it is not what makes the tombstone's removal audible.Generated artifacts regenerated
authorable-surface/ui.json(four new keys,ui/PageTabsProps:tabStyle, andui/PageTabsProps:type [RETIRED]),authorable-defaults/ui.json,spec-changes.json,docs/protocol-upgrade-guide.md,content/docs/references/ui/component.mdx.api-surface/is unchanged, as expected for key-level narrowing on a def that still exists.Follow-up for the objectui shard
Nothing to change in objectui source — its registry already publishes all five names. On the spec pin bump, the parity gate's exemption entries for these five keys expire and must be deleted (the gate asserts expiry), alongside the three
element:record_pickerexemptions #5775 already discharged.Generated by Claude Code