Repository navigation
refactor(types,collaboration,data-objectstack,plugin-designer,providers): one authority each for ConnectionState, VersionEntry and ThemeProviderProps (objectui#6349, batch 10) - #12083
Merged
objectstack-fleet[bot] merged 3 commits intoOct 10, 2026
Conversation
…try and ThemeProviderProps (objectui#6349, batch 10) Two of batch 10's three ledger rows, each two meanings behind one name: - VersionEntry: collaboration's is a version useConflictResolution records (id plus a per-field changes diff); the designer's is a row the VersionHistory timeline displays (description, isCurrent). The designer's entry never exported the name, so its declaration now spells VersionHistoryEntry. - ThemeProviderProps: react's is its theme-system ThemeProvider (Theme documents, a theme-name defaultTheme, defaultMode); providers' is a mode-only provider whose defaultTheme is the mode. providers' declaration now spells ThemePreferenceProviderProps and its root publishes that name instead; react keeps the general name. Both rows leave KNOWN_COLLISIONS. ConnectionState stays: the two sites are one identical union, and neither package depends on the other. Claude-Session: https://claude.ai/code/session_01CGZy1BGCjdN5cXqL9cnvB8 Co-authored-by: Claude <noreply@anthropic.com>
…ally gets (objectui#6349, batch 10) Measured against the rebuilt providers entry: the old import fails with TS2724, whose suggestion names the ThemeProvider component rather than the replacement, and following that suggestion fails with TS2749. The changeset said TS2305. Claude-Session: https://claude.ai/code/session_01CGZy1BGCjdN5cXqL9cnvB8 Co-authored-by: Claude <noreply@anthropic.com>
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
|
…nectionState (objectui#6349, batch 10) Round 2, on the seat's amendment to the batch 10 claim (option A). collaboration's and data-objectstack's ConnectionState were the same five-state union, type-equal on both built entries, and neither package depends on the other. Both depend on @object-ui/types, so the one declaration moves down into types' data.ts beside DataSource, the types root publishes it, and both former sites re-export it. Both package roots keep the name and the shape. The row leaves KNOWN_COLLISIONS. Claude-Session: https://claude.ai/code/session_01CGZy1BGCjdN5cXqL9cnvB8 Co-authored-by: Claude <noreply@anthropic.com>
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
|
Contributor
Author
Contract reviewServed-tier: Isolated, read-only review of PR objectui#12083 (card objectui#6349, batch 10): base ① Derived judgmentsCheck-runs on Each accept-set and public-surface change the diff implies, judged:
② Semver level
③ Boundary flags
Implemented-by: VERDICT: PASS |
objectstack-fleet
Bot
changed the title
objectstack-fleet
Bot
deleted the
claude/issue-6349-batch10-collab-providers
branch
October 10, 2026 04:34
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.
Refs #6349 (batch 10)
Clause-②: yes
Batch 10 of the one-authority burn-down, on the claim
6093179763. It covers threeKNOWN_COLLISIONSrows from the collaboration and providers packages, none with an@object-ui/typesside. Two rows converge here. The third,ConnectionState, stopped with measured options in round 1 and converges in round 2 (above), because the rules this card has applied do not settle it (see "The stop" below).(Round 1.) Implemented in session
https://claude.ai/code/session_01CGZy1BGCjdN5cXqL9cnvB8(the seat's session; this was a subagent dispatch).Round 2 (head
65da7a995): all three rows convergeBrought up to round 2 by the seat (
domain:uiseat 3). The round-1 text below is kept and marked where round 2 supersedes it.ConnectionState: ruled A by the seat's claim amendment6093405166on objectui#6349, under batch 3's rule and batch 4'sUndoableOperationprecedent, with the maintainer's veto window open.@object-ui/types(data.ts, besideDataSource) and is published from the types barrel.useRealtimeSubscription.tsand data-objectstack'sindex.tsare plain re-exports of it.KNOWN_COLLISIONS: 10 → 7 rows.ConnectionStatejoins). collaboration 39 = 39 and data-objectstack 104 = 104, with each copy type-equal to the types declaration. plugin-designer, providers and react are unchanged by round 2..jsfile differs.@object-ui/typesminor(a new name), and@object-ui/collaborationand@object-ui/data-objectstackpatch. ⛔ Nomajor.65da7a995:type-check:scriptsexit 0;check:doc-snippets: 785 of 785 blocks judged, 0 failed.check:readme-exportsis OK;What converged:
KNOWN_COLLISIONSgoes from 10 to 8 rows (round 1; 7 after round 2)VersionEntry: two meanings. The designer's side becomesVersionHistoryEntryversion,timestamp,userIdanduserName;userNameis optional in collaboration and required in the designer. Only collaboration's hasid,changes(a per-field before/after diff thatrevertToVersionandcompareVersionsreplay) andmessage. Only the designer's hasdescriptionandisCurrent. Collaboration's is a version itsuseConflictResolutionhook records. The designer's is a row itsVersionHistorytimeline displays.i18nandtypes, and plugin-designer's dependency list has no collaboration.VersionEntry. The plugin-designer root does not. Its entry exports theVersionHistorycomponent, but neitherVersionEntrynorVersionHistoryProps, and thecomponents/index.tsbarrel that re-exported it has no importer. So the rename removes no published name: the plugin-designer root has 38 names before and 38 after. Batch 6'sDesignerConfirmDialogStaterow had the same shape.@objectstack/spec17.7.0 (control:RealtimeConfigandThemeModehit there).ThemeProviderProps: two providers. providers' side becomesThemePreferenceProviderPropsdefaultTheme?: ThemePreference,storageKey?andchildren. react's hasthemes?,defaultTheme?: string,defaultMode?: ThemeMode,persist?,storageKey?,target?andchildren. The shared keydefaultThememeans two things: a MODE on providers' side ('dark') and a theme NAME on react's ('my-theme'). Neither provider wraps the other. providers' provider stores one mode and toggles thelight/darkclass. react's registersThemedocuments and writes CSS variables.types; react depends oncore,data-objectstack,i18n,typesand the spec.ThemePreferenceProviderPropsinstead, with identical members (23 names before and 23 after). The react root is unchanged (300 = 300). react's theme system is the general one:@object-ui/types' theme module names react'sThemeProvideras the one that applies aTheme. So react keeps the general name.@objectstack/spec17.7.0. This is therefore the zero-reader rename of a published name that the card's rules already settle (batch 7'sGroupAggregationConfig, batch 8'sListViewSwitcherProps), and it is not a stop.ThemeProvidercomponent, not the replacement. Following the suggestion fails too (TS2749, a value used as a type). Nothing compiles silently wrong, and the changeset says this.ThemeProvidercomponents (anduseTheme) keep their names. They are outside the gate's type-level bound, in the same way batch 9 kept the twoAppShellcomponents.The stop:
ConnectionState(round 1; ruled A in6093405166and implemented in round 2)ConnectionStateare type-equal: a mutual-equality probe exits 0, the same assertion flipped exits 2 (TS2322), and a control that drops one member is not equal.i18nandtypes; data-objectstack hascore,types,clientand the spec. Both already depend on@object-ui/types.AppContent,AppHeader,ConnectionStatusandConsoleLayout, and the console'sdataSource.tsre-export. collaboration's has 0 named readers, but it is the published state type ofuseRealtimeSubscription, which app-shell'sObjectViewmounts. 0 hits in@objectstack/spec17.7.0.@object-ui/types, which both packages already depend on. collaboration and data-objectstack each re-export it.typessource file (besideDataSourceindata.tsis natural) and thetypesbarreltypesminor (one new name); collaboration and data-objectstack patch (shape and name sets unchanged)RealtimeConnectionState, published at its root in place of the old name (0 readers)Why A, along the four axes:
UndoableOperation, objectui#8261'sTranslateFnand objectui#7664's Kanban trio.typesand removes a duplicate. No capability is added.Changeset grades
@object-ui/providersminor with a breaking note: a name leaves the root, and the note names the migration and the measured TS2724 / TS2749.@object-ui/plugin-designerminor with a nothing-breaks note: only the type name in the published declaration ofVersionHistory'sversionsprop changes. This is the same grading batch 6 gaveuseConfirmDialog'sDesignerConfirmDialogState.@object-ui/typesminor(ConnectionStatejoins), and@object-ui/collaborationand@object-ui/data-objectstackpatch(each site becomes a re-export).Verification (head
48285b125, base1b2d0160)ThemeProviderPropsleaves,ThemePreferenceProviderPropsjoins) and react 300 = 300..d.ts/.d.ts.mapfiles differ, so no first-load byte moves.one-authority-per-exported-name-6273) passes 11 of 11.VersionHistorylocale test and the gate ran at the final head: 5 files and 26 tests passed. The plugin-designer package suite ran atef2e5845b, whose code is identical to the head (the later commit changes one changeset only): 46 files and 449 tests passed.type-check:scriptsall exit 0. app-shell's first attempt exited 2 with TS2307 on unbuilt dependencies. That was an unmet prerequisite, not a measurement, and it exited 0 once its dependency closure was built.check:control-bytesOK,check:test-path-rootsOK,check:changeset-claimsOK,check:pending-changeset-literalsOK,check:readme-exportsOK (582 of 582 self-imports judged, after building the 7 packages its first run reported unbuilt),check:doc-snippets(785 of 785 blocks judged, 0 failed),check:new-line-citations(0 new), changeset presence OK and no-major OK. eslint on the touched files reports 0 errors and 3 warnings, all on lines this diff does not touch.ablation-replace, with predictions written first. Control: 11 passed. All four legs went red as predicted, each with 1 failed and 10 passed in "declares no name twice outside the named, shrinking baseline":VersionEntryrow restored): stale, both files listed;VersionEntryagain): fresh, "a NEW colliding name";ThemeProviderPropsrow restored): stale, both files;ThemeProviderPropsagain): fresh.Acceptance notes (observations, not filed)
ThemeProvider: providers', react's and app-shell's chrome one, which the app-shell root exports. This is outside the gate's type-level bound, like theAppShellpair.CollaborationProviderspells a fourth, unnamed connection-state union inline: four members, noreconnecting. It is not a collision under the gate. If option A lands, it has a name to point at.components/index.tsbarrel has no importer, like thehooks/index.tsbarrel batch 6 noted.Generated by Claude Code