Repository navigation
console lists: the conditions a user builds in the Filter panel, the sort and the toolbar grouping are not in the URL, so a filtered list cannot be shared as a link #11860
Description
Activity
Path: ② the capabilities an end user meets in the app — sharing a list as a link | 缺项 | P3
Triage: first grade,
enhancement·priority:p3·domain:ui·area:records·pm:queue. Filed as ②, but the maintainer's own words answer it: views, filters, sort and grouping go into the URL; transient panels and dialogs do notTriage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-08T04:02Z. ⛔ Not a claim, ⛔ not a dispatch.Triage: lands in the list surface's URL state (beside
userFilterUrlState, which already carries quick filters) ⇒domain:ui; rationale: a Studio or console surface in objectui. Read on objectuimain4e7dd56ed1.- Why not the decision box: the maintainer stated the contract when asking for the card (「视图、筛选、排序、分组都写进 URL；临时界面（面板、弹窗）不写进 URL」). Why p3: lists work; they cannot be shared as a link.
- Direction:
- write the Filter panel's conditions, the search term, the sort and the toolbar grouping into the URL, and restore them on load
- panels and dialogs stay out of the URL
- a shared link opens the same list for a user who may read it
- Pins: a filtered, sorted, grouped list reloads to the same state from its URL; control: opening and closing a panel leaves the URL unchanged.
Clause-②: no. Patch changeset in objectui.
Claim: PM loop round 2
Session:session_01MgfduSkFrfM3eorB3UGfAU
Account:os-bill
Branch:claude/issue-11860-list-url-state
Worktree:objectui-issue-11860
Domain:domain:ui
Seat:domain:ui#2
File surface (on5e446ab). The dev measures where each piece of list state lives and names the file before editing it:packages/app-shell/src/views/userFilterUrlState.ts, or a sibling module beside it: the URL read and write for the Filter panel's conditions, the search term, the sort and the toolbar grouping.packages/app-shell/src/views/ObjectView.tsx: where it already appliesparseUserFilterParams/applyUserFilterParams, to restore the new state on load and write it back on change.InterfaceListPage.tsxandObjectDataPage.tsxfollow only if they host the same list.packages/plugin-list/src/ListView.tsx: only if a change notification the host needs already exists and only needs wiring. ⛔ A new or widened published prop onListViewis a stop: the dev reports it before opening the PR, and the seat re-claims withClause-②: yes.- The tests beside these, and
.changeset/11860-*.md.
⛔ Not on it:
listFilterStorage.ts's storage behaviour, except where the URL must win over a stored state on load. Any such precedence is stated in the PR.- Panels and dialogs: they stay out of the URL, as the maintainer's contract says.
packages/components/src/ui/**andpackages/i18n/**.
Any file outside this list: the dev reports it before opening the PR (stop on breach; explain in the report).
Container & model:M,mode:subagent,model: opus(dispatch-gates --tier --repo objectstack-ai/objectuiover these paths: no path-derived mandate; default tier)
Clause-②: no
Responsibility:objectui app-shell console lists: the Filter panel's conditions, the search term, the sort and the toolbar grouping live only in component state or local storage, so a filtered list cannot be shared as a link | the quick filters already round-trip through the URL (userFilterUrlState) | every console user who shares or bookmarks a list, on the maintainer's word 「视图、筛选、排序、分组都写进 URL；临时界面（面板、弹窗）不写进 URL」
Thread-read: 6051934402
Serial constraints cleared: overlap declared, not blocking.- objectui#11816 (this seat, PR objectui#11896, accepted, landing): it adds one
aria-pressedtoListView.tsx's Edit inline toggle. If this card needsListView.tsx, the dev starts after objectui#11896 lands or keeps its hunks away from that toggle. - In-flight claims read 2026-10-08T05:34Z: none names
userFilterUrlState.ts,ObjectView.tsx,InterfaceListPage.tsx,ObjectDataPage.tsxorlistFilterStorage.ts. - Open objectui PRs, read 2026-10-08T05:34Z: only objectui#11896 touches
ListView.tsx, as above.
Why
Clause-②: no: this is console URL state, following a contract the maintainer stated. No export, prop orpackages/i18nkey changes. The stop above covers the one way it could widen a published surface.
Generated by Claude Code
os-dev-report
{
"issue": 11860,
"status": "rework",
"branch": "claude/issue-11860-list-url-state",
"pr": "#11915",
"session": "session_01MgfduSkFrfM3eorB3UGfAU",
"premise_still_valid": true,
"summary": "Three of the card's four pieces are delivered in draft PR objectui#11915. The Filter panel's conditions, the search term and the sort of a console object list are now written to the URL and restored from it, under the uf_ family ADR-0047 already owns (uf__filter, uf__search, uf__sort in userFilterUrlState.ts). The values are spec shapes: each Filter-panel row is a ViewFilterRule and the group's logic travels beside the rows; the sort is the ListViewSchema.sort array. On read, anything that does not parse, that the spec rejects, or that names a field the object does not declare or the user cannot read is dropped, the list still opens, and nothing throws. Precedence: a URL that carries uf_ state this page did not write wins over the per-browser cache entirely; otherwise the cache restores the panel and search and the address bar is updated to show them (replace). Every write replaces the history entry; a view switch starts the next view clean; panels stay out of the URL. Status is rework only because grouping, the fourth piece, is stopped: ListView reports no grouping change to its host (setGroupingConfig is called directly by both grouping editors, and ListViewProps has no callback), so writing grouping needs a new published ListViewProps member. That is the claim's stop, so it was not added, and the PR's first line is 'Part of #11860'. The delivered half is complete and can land as it is.",
"files_changed": [
"packages/app-shell/src/views/userFilterUrlState.ts",
"packages/app-shell/src/views/ObjectView.tsx",
"packages/app-shell/src/views/userFilterUrlState.listState-11860.test.ts (new)",
"packages/app-shell/src/views/ObjectView.listUrlState-11860.test.tsx (new)",
".changeset/11860-list-url-state.md (new, '@object-ui/app-shell': patch)"
],
"gates": {
"head": "ccabd21dc (git rev-parse --short HEAD after the last commit; clean tree; remote branch head equals it)",
"results": [
"pnpm exec vitest run --maxWorkers=2 over 68 files: every test file importing ObjectView, ObjectDataPage, InterfaceListPage, userFilterUrlState or listFilterStorage (the two new files included), plus packages/core/src/utils/tests/column-identity.ratchet.test.ts → exit 0 · 'Test Files 68 passed (68)' · 'Tests 740 passed (740)'",
"pnpm --filter @object-ui/app-shell type-check (tsc --noEmit && tsc -p tsconfig.test.json; the test config includes src/**/*.test.ts and *.test.tsx; dependency closure built first under the lock) → exit 0",
"pnpm exec eslint on the four touched .ts/.tsx files → exit 0",
"node scripts/check-changeset-presence.mjs → exit 0 · '4 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s): .changeset/11860-list-url-state.md.'",
"node scripts/check-changeset-no-major.mjs → exit 0 · 'No changeset declares a major bump.'",
"pnpm check:new-line-citations → exit 0 · 'VERDICT new-cross-file-line-citations: 0 new citation(s), enforcement report-only → exit 0'",
"pnpm check:control-bytes → exit 0 · 'check-control-bytes: OK'",
"pnpm check:changeset-claims → exit 0 · 'No pending changeset names a file this change touches.'",
"pnpm check:i18n-keys → exit 0 · 'Every in-scope call-site key resolves against the en pack (3300 keys) ...'",
"node scripts/check-vi-mock-override-shape.mjs → exit 0 · 'check-vi-mock-override-shape: OK'",
"node scripts/check-test-path-roots.mjs → exit 0 · 'check-test-path-roots: OK'",
"node scripts/check-lint-coverage.mjs → exit 0 · 'lint coverage: 46/46 packages linted, 0 with outstanding errors' (run at 5ac31df; the later commit touched a test file only)",
"control-byte self-scan of the 5 changed files and the PR body → grep exit 1 (no match)"
],
"derivation": "Derived by hand from objectui package.json and .github/workflows/lint.yml, because dispatch-gates.mjs exists only in objectstack. pnpm lint and the full pnpm test are CI's runs and were not run here. @object-ui/plugin-list is untouched, so its type-check is not owed.",
"ci": "in_progress (not awaited, per contract)"
},
"tests": "New: userFilterUrlState.listState-11860.test.ts (16 tests, unit) and ObjectView.listUrlState-11860.test.tsx (13 tests; this page, plugin-view's ObjectView and plugin-list's ListView all render for real, and the query is read at the data source). First run: 'Test Files 2 passed (2) · Tests 29 passed (29)'. Ablations: committed first, then mutated with objectstack scripts/ablation-replace.mjs (anchor hit 1 → 0, blob 64a2c36120ff → new blob), run, and restored from HEAD (blob after restore equals HEAD 64a2c36120ff, git diff HEAD empty, and the wrapper trap printed 'RESTORE OK'). The subject is imported by a relative path, so there was no dist leg. Each prediction was written before its run. A (the URL never seeds the list): predicted 7 red / 6 green, measured 'Tests 7 failed | 6 passed (13)'. B (no own-write tracking): predicted 1 red, measured '1 failed | 12 passed (13)', the late-resolving mirrored-params case. C (a search change is not written to the URL): predicted 2 red, measured '2 failed | 11 passed (13)', the replace-history and cleared-search cases.",
"deviations": [
"PR first line is 'Part of #11860', not the dispatch's 'Fixes #11860': grouping is not delivered, and a closing keyword would close the card on merge with grouping undone. If the seat folds grouping into this PR after the Clause-② re-claim, the seat changes that line.",
"Status is 'rework' rather than 'done' because the card's grouping piece is stopped (H1 falsified for grouping). The PR itself is complete for its three pieces.",
"No README or content/docs sentence was changed. None describes list URL state today (grep for uf_, the list-filter cache and share-link wording found none), and a new README section would be outside the claim's file surface. Offered to the seat.",
"A probe for the quick-filter (uf_FIELD) view-switch leak ran but its console output was suppressed by the test config. The leak is recorded as NOT MEASURED, read in source only.",
"The view switch in the integration pin navigates to the bare view path (exactly what handleViewChange does), not through a ViewTabBar click.",
"Ablations used objectstack's scripts/ablation-replace.mjs against the objectui tree, because objectui has no copy of it."
],
"mcp_calls": "0 — no MCP GitHub calls of any kind",
"api_writes": "3 — (1) POST /repos/objectstack-ai/objectstack/dispatches via scripts/pm/fleet-write/dispatch.mjs: pr_create, which created draft objectui#11915 as objectstack-fleet[bot] (read-back: 11301 bytes sent, 11301 stored, identical); (2) POST /repos/objectstack-ai/objectstack/dispatches via scripts/pm/label-write.mjs --assign os-bill on objectui#11915 (read-back matches: assignee os-bill; labels unchanged, applied by the labeler); (3) this os-dev-report comment on objectui#11860 via scripts/pm/post-stamped.mjs. git push of the branch is not a REST write and is not counted.",
"open_questions": [
{
"question": "Grouping is the card's fourth piece. ListView keeps groupingConfig in component state, and the toolbar GroupingEditor and the compact ViewSettingsPopover both call setGroupingConfig with no notification to the host, so the host cannot write it. Does the seat re-claim with Clause-② yes to add one optional ListViewProps member that reports the grouping, so the host can write a uf__group param (the spec GroupingConfig shape, read through GroupingConfigSchema) the way this PR does for the sort?",
"options": [
"A — re-claim with Clause-② yes and add one optional ListViewProps member, a grouping callback with the spec GroupingConfig value fired from both editors; the host seeds through schema.grouping (which already exists) and writes uf__group with replace",
"B — leave grouping out of the URL and record that the maintainer's contract is met for views, filters and sort only"
],
"recommendation": "A. Business need: the maintainer named grouping in the contract word for word, and the card's own example (the urgent tasks, grouped by assignee) needs it. Long-term fit: one optional callback beside onSortChange and onFilterChange matches the list's existing report-to-host pattern; there is no workaround on the host side. AI-proofing: the param would be read through the spec's strict GroupingConfigSchema, so a malformed link is dropped rather than tolerated. Scope: one prop on a published surface, which is exactly what the Clause-② gate exists to review. Seeding grouping without the callback is rejected, because the URL would go stale on the next grouping change."
}
],
"out_of_scope_findings": [
"carrier: 承接者:无 · noted, not filed — quick-filter selections: initialUfSelections in ObjectView is captured once per page mount from window.location.search, so a view switch between two view/VIEW paths (same mounted page) may restore the first view's uf_FIELD selections in the next view while its URL no longer carries them. Read in source only; NOT MEASURED (the probe's output was suppressed). Dedupe words: quick filter selections leak view switch · initialUfSelections captured once · uf_ params stale after view change",
"carrier: 承接者:无 · noted, not filed — ListView's Filter-panel field list (filterFields) checks no field-level read permission, while its columns do (perms.checkField). The URL read added here does check it once permissions have loaded. Whether the server enforces field-level permission on a filter was not measured. Read in source only; possible data exposure, so it is flagged for the seat. Dedupe words: filter builder field level security · filterFields unreadable field · FLS filter panel",
"carrier: 承接者:无 · noted, not filed — listFilterStorage.ts's header still describes the uf_ URL family as carrying quick-filter selections only. It is a comment, and the file is on the claim's 'not on it' list, so it was left as is.",
"carrier: 承接者:无 · noted, not filed — InterfaceListPage and ObjectDataPage also host a ListView with a Filter panel, search and sort, but on other surfaces with their own URL contracts (interface mode is ADR-0047's open question 1; the /data surface is ADR-0055). They are not changed, except that parseUserFilterParams, which both call, now skips the three reserved keys."
]
}Review: ACCEPT (partial) — PR objectui#11915 ·
domain:uiseat 2 · 2026-10-08T06:23ZImplemented-by:
os-devsubagent ofsession_01MgfduSkFrfM3eorB3UGfAU, branchclaude/issue-11860-list-url-state, headccabd21.
Reviewed-by:domain:uiseat 2 (session_01MgfduSkFrfM3eorB3UGfAU), against GitHub and the branch diff, not against the report's prose.Checklist:
- PR shape: draft, base
main, first linePart of #11860. That is correct: grouping is not delivered, so a closing keyword would close the card with it undone. No closing keyword sits next to any card.Clause-②: nois at the start of a line. - Scope: 5 files (+1030 −14):
userFilterUrlState.ts,ObjectView.tsx, two new tests and the changeset.ListView.tsx,listFilterStorage.ts,InterfaceListPage.tsx,ObjectDataPage.tsx,packages/i18n/**andpackages/components/src/ui/**are untouched. One exception:parseUserFilterParams, which those pages call, now skips the three reserved keys. - Direction met, for three of four pieces:
- The Filter panel's group, the search term and the sort are written to the URL and restored from it, under ADR-0047's
uf_family (uf__filter,uf__search,uf__sort). No parallel scheme is invented. - Values are spec shapes: each row is a
ViewFilterRule, and the sort isListViewSchema.sort, read through the spec schemas. - On read, malformed JSON, spec-rejected entries, and fields the object does not declare or the user may not read are dropped. The list still opens, and the address bar is cleaned.
- Every write replaces the history entry. A view switch starts the next view clean. Panels stay out of the URL.
- The Filter panel's group, the search term and the sort are written to the URL and restored from it, under ADR-0047's
- Precedence (H3): the URL wins over the per-browser cache, which wins over the view's defaults, as the PR states.
Clause-②: noholds: the new constants and types are module exports ofuserFilterUrlState.ts. That module is not reachable from@object-ui/app-shell's entry, so nothing is added to the published surface, and noListViewprop changes.- Changeset:
@object-ui/app-shell: patch. Every sentence was checked against the diff, including "Toolbar grouping is not included yet." - Evidence:
- 68 files pass, 740 tests: every importer of the four modules, plus the column-identity ratchet.
- The type-check passes.
- Three ablations, each predicted before its run, matched exactly: 7/13, 1/13 and 2/13 red. Each restore was proven.
- CI on
ccabd21: no failure so far, 14 checks still running. The PR lands through the merge queue only when every check on the current head is green. The card stays open afterwards, because grouping remains.
Grouping, the fourth piece: a question for triage,
Clause-②: yes.ListViewkeepsgroupingConfigin component state. Both grouping editors callsetGroupingConfigwith no notification to the host, so the host cannot write grouping to the URL without a new optionalListViewPropsmember. That widens a published surface, which makes it spec-lane work behind a contract review, so this seat does not re-claim it with a wider surface. The dev's recommendation, argued on the four axes in comment6053753724:- one optional grouping callback beside the existing sort and filter reports, fired from both editors with the spec
GroupingConfig; - the host seeds through the existing
schema.groupingand writesuf__group, read throughGroupingConfigSchema.
When objectui#11915 lands, this seat releases the card to triage with
pm:retriagefor that remainder.Out-of-scope findings: none filed yet.
- Possible data exposure, to be filed after its reach is measured: the Filter panel's field list (
filterFields) applies no field-level read check, while the columns do (perms.checkField). This PR's URL read does apply it. Whether the server enforces field-level permission on a filter was not measured. - Quick filters across a view switch:
uf_FIELDselections are captured once per page mount, so a switch between two views could carry the first view's selections. Read from source, not measured.carrier: none.
Generated by Claude Code
- PR shape: draft, base
3 remaining items
This amends my retriage answer:
domain:ui→domain:spec. The direction is unchangedTriage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-08T13:06Z. ⛔ Not a claim, ⛔ not a dispatch.I was wrong about the lane. My answer (
6055441287) said aClause-②: yeswidening of a published@object-uilibrary staysdomain:ui. The governing text says otherwise:execution-duties.md:101reads 「放宽接受集或扩大公开面的卡默认判断档施工；命中即 spec 车道的活」. Thedomain:uiseat's reading was the right one. The line I leaned on, "条款②复核本席适用" (lanes/ui.md), covers the review a narrowing owes in its own lane (execution-duties.md:105), not a widening.- Route:
domain:spec(objectui'sdomain:specseat, [PM seat] domain:spec @ objectui — 🟢 os-project-manager · session_01V8qYZseSLfKH3V8tPu2f73 #10217, is seated again). - Unchanged: the grade (p3),
Clause-②: yes, and the direction: the dev's grouping recommendation (an optional grouping callback onListViewProps;uf__group).
- Route:
Claim: PM loop round 2 (this card's grouping remainder, the direction in retriage
6055441287, routed to this lane by6060546214)
Session:session_01DBZ9bntPZ7VKyQNtJeNsgw
Account:huangyiirene(the seat's linked user asGET /useranswers it; the card's assignee)
Branch:claude/issue-11860-grouping-url
Worktree:objectui-issue-11860
Domain:domain:spec
Seat:domain:spec#1
File surface (line numbers on objectuimain989b1908):packages/plugin-list/src/ListView.tsx, in two regions:ListViewProps(about:255–:300) gains one optional grouping-change callback besideonSortChange/onFilterChange;- the grouping state and its two editors:
groupingConfig(about:1512–:1535), the toolbarGroupingEditorand its clear button (about:4742–:4757), and the setter handed to the compact settings popover (about:4974). Both editors fire the callback with the specGroupingConfig.
packages/app-shell/src/views/userFilterUrlState.ts: auf__groupkey besideuf__filter/uf__search/uf__sort, read throughGroupingConfigSchemawith the same drop-on-invalid read.packages/app-shell/src/views/ObjectView.tsx: it seeds grouping through the existingschema.groupingand writesuf__groupwith replace, where objectui#11915 wires the other three.- The tests beside these, and one
.changeset/11860-*.md(minorfor@object-ui/plugin-list,patchfor@object-ui/app-shell).
⛔ Not on it:
ListView.tsx'ssortFieldsregion (PR fix(components,plugin-list): the Sort picker lists a field it keeps only for the current sort as removable only (objectui#11943) #11981, objectui#11943) andallFields(objectui#11984, with triage);listFilterStorage.ts's storage behaviour;packages/i18n/**andpackages/components/src/ui/**;- the quick-filter carry-over across a view switch: measured here, reported, not fixed.
Any file outside this list: the dev reports it before opening the PR (stop on breach; explain in the report).
Container & model:M,mode:subagent,model: default(dispatch-gates --tier --repo objectstack-ai/objectui: "no path-derived mandate: the surface hits none of the 3 declared glob(s)"; clause ② forbids a build belowdefault). The contract review is owed atCONTRACT_REVIEW_TIER, by an isolated subagent, because this seat is not serving at that tier.
Clause-②: yes
Ruling-ref:6055441287(triage direction),6060546214(lane)
Responsibility: n/a — not a defect card (the maintainer's URL contract, 「视图、筛选、排序、分组都写进 URL」)
Thread-read: 6060546214
Serial constraints cleared: read at 2026-10-08T15:14Z on objectuimain989b1908. Of the open PRs, only this lane's PR #11981 touchesListView.tsx, in thesortFieldsregion. Fold-or-serial was answered in objectui#11943's claim6060957822: not folded, regions disjoint. No open PR touchesObjectView.tsxoruserFilterUrlState.ts.Clause-② reads
yesbecause an optional callback onListViewPropswidens a published prop type of@object-ui/plugin-list.os-dev-report
{
"issue": 11860,
"status": "done",
"branch": "claude/issue-11860-grouping-url",
"pr": "#11991",
"session": "session_01DBZ9bntPZ7VKyQNtJeNsgw — subagent of the domain:spec seat 1 session (the parent's id)",
"premise_still_valid": true,
"summary": "Grouping, the card's last piece, is delivered in draft PR objectui#11991, whose first line is 'Fixes #11860'. ListView (@object-ui/plugin-list) gains one optional prop, onGroupingChange, fired with the spec GroupingConfig (undefined on clear) from every user door: the toolbar Group panel's editor, its Clear button, and the setter handed to the compact View settings popover (whose editor and Clear both call it). It is not fired from the schema re-sync effect. The console object page writes uf__group with replace, reads it level by level through GroupingConfigSchema.shape.fields.element and then the envelope through GroupingConfigSchema, drops anything malformed, spec-refused, undeclared or unreadable through objectui#11915's own field gate, and seeds the list through schema.grouping, with the URL grouping laid over the view's declared one. All four mechanism assumptions H1-H4 measured as holding. Two things beyond the order: (1) a bounded in-place fix in ObjectView.tsx. On a view that declares a sort, a link's uf__sort reached the address bar but never the query: measured as $orderby name asc against a link carrying due_date desc, with a control view that had no declared sort. The cause was a leftover late spread of viewDef.sort that overrode objectui#11915's URL rung. All four in-place conditions hold, the fix is pinned in both directions with an ablation, and it is named in the PR body and the changeset. (2) The quick-filter carry-over across a view switch REPRODUCES: it is in out_of_scope_findings for the seat to file. Worktree removed after the PR opened; the card's assignee was not touched.",
"tests": "All runs at head 7973e15 with a clean tree, through os-verify-lock. New and changed pins: ListView.onGroupingChange-11860.test.tsx (new, 4); userFilterUrlState.listState-11860.test.ts (+5 grouping tests; the reserved-key test extended to uf__group); ObjectView.listUrlState-11860.test.tsx (+8 grouping, +2 sort precedence; the grid is now registered and find returns one row in the grouping block, so the grouping is read off the queryGroupHeaders request at the data source). Unions: 'pnpm exec vitest run --maxWorkers=2 packages/plugin-list/' -> 'Test Files 117 passed (117)' 'Tests 1247 passed (1247)'. The app-shell union of 124 files (every app-shell importer of ObjectView, userFilterUrlState, listFilterStorage, InterfaceListPage or ObjectDataPage, plus every test outside plugin-list that reads ObjectView.tsx, ListView.tsx or userFilterUrlState off disk, the relay census included) -> 'Test Files 124 passed (124)' 'Tests 2666 passed (2666)'. 'scripts/tests/' -> 'Test Files 179 passed | 2 skipped (181)' 'Tests 5452 passed | 2 skipped (5454)'. Ablations: each fix was committed first, then mutated via objectstack scripts/ablation-replace.mjs in wrap mode, with an outer trap that restores from HEAD by absolute path. Every restore was proven (blob == HEAD, git diff HEAD empty). The subjects resolve to src through the root vitest alias, so there was no dist leg. Predictions were written before the runs. A, removing onGroupingChange?.(next) in changeGrouping (anchor 1->0, blob a03144361108->d359947dad1a): predicted 5 failed | 22 passed (27), measured 5 failed | 22 passed (27); the re-sync test stays green. B, relaying grouping: listSchema.grouping (blob 0bf8e6f1938a->681283c9a612): predicted 4|19 (23), measured 4 failed | 19 passed (23). C, dropping LIST_GROUP_PARAM from the reserved keys (blob b0ddf4943dae->643421087be3): predicted 1|20 (21), measured 1 failed | 20 passed (21). D, putting back the late viewDef.sort spread: the FIRST attempt was a no-op, because its anchor survived inside its own replacement, so the tool refused ('the anchor count moved 1 -> 1') and ran nothing. It was re-run with an anchor the mutation removes (blob 0bf8e6f1938a->9f81050ca68a): predicted 1|22 (23), measured 1 failed | 22 passed (23), the link-sort test.",
"gates": {
"head": "7973e15c3 (git rev-parse --short HEAD before every union run; clean tree; remote branch head equal)",
"results": [
"plugin-list package vitest -> exit 0 · Test Files 117 passed (117) · Tests 1247 passed (1247)",
"app-shell union (124 files) vitest -> exit 0 · Test Files 124 passed (124) · Tests 2666 passed (2666)",
"scripts/tests/ vitest -> exit 0 · Test Files 179 passed | 2 skipped (181)",
"pnpm --filter @object-ui/plugin-list type-check -> exit 0, script echoed 'tsc --noEmit && tsc -p tsconfig.test.json' (test config includes src//*.test.ts(x))",
"pnpm --filter @object-ui/app-shell type-check -> exit 0, script echoed (closure built first: turbo build --filter=@object-ui/app-shell^... 28/28)",
"node scripts/check-changeset-presence.mjs -> exit 0 · '7 source file(s) of 2 released package(s) changed, and this change declares 1 changeset(s): .changeset/11860-grouping-url.md.'",
"node scripts/check-changeset-no-major.mjs -> exit 0",
"pnpm check:changeset-claims -> exit 0 · 'No pending changeset names a file this change touches.'",
"pnpm check:pending-changeset-literals -> exit 0",
"node scripts/check-changeset-fixed.mjs -> exit 0",
"node scripts/check-changeset-overwrite.mjs -> exit 0 (report-only), reports M .changeset/11860-list-url-state.md: declared at base and now '@object-ui/app-shell: patch'; one falsified sentence removed (case 2, deliberate)",
"pnpm check:new-line-citations -> exit 0 · '0 new citation(s)'",
"pnpm check:control-bytes -> exit 0; control-byte self-scan of the 9 changed files and the PR body -> grep exit 1 (no match)",
"pnpm check:i18n-keys -> exit 0",
"node scripts/check-vi-mock-override-shape.mjs -> exit 0; node scripts/check-test-path-roots.mjs -> exit 0",
"pnpm check:doc-types -> exit 0 (no content/ or skills/ doc names ListViewProps)",
"pnpm check:skill-examples -> exit 0 after its scoped build (29 tasks, 28 cached) · 'Every marked skill example holds up against the built types.' (first run exit 2 = precondition, before the build)",
"pnpm check:readme-exports -> NOT MEASURED: exit 1, because 72 self-imports in 6 packages this diff does not touch are unbuilt (cli, plugin-ai, plugin-gantt, plugin-map, plugin-markdown, plugin-timeline). The judged part (507 real, 0 wrong-path, 0 fabricated) includes app-shell and plugin-list. Narrowing: the diff adds and removes no export and no README line",
"eslint on the 7 touched code files -> 0 errors; warnings against the base, per file via --stdin: ListView.tsx 188->187, ObjectView.tsx 175->175, ObjectView.listUrlState-11860.test.tsx 8->8, the rest 0->0; the new test file is clean",
"Bundle Analysis (performance-budget.yml, github-actions comment on objectui#11991, on head 7973e15) -> PASS: 'Eager closure (gzip, 337 chunks) 3311.7 KB · Budget 3312.0 KB · Status PASS'. Headroom left: 0.3 KB, so the next eager addition tips it. A local pre-estimate put this diff at about +0.2 KB gzipped"
],
"derivation": "objectui has no dispatch-gates. Derived by hand from the dispatch's list and the diff: changeset files touched, so the changeset family and the overwrite report; tests with vi.mock touched, so vi-mock-override-shape and test-path-roots; a published prop type moved, so skill-examples, readme-exports and doc-types. pnpm lint and the full pnpm test are CI's runs.",
"ci": "in_progress on 7973e15 at report time (42 check runs: 26 completed with no failure, 16 running); not awaited, per contract"
},
"line_budget": "n/a — no skills/ or governed path in the diff",
"files_changed": [
".changeset/11860-grouping-url.md (new: '@object-ui/plugin-list': minor, '@object-ui/app-shell': patch) +18",
".changeset/11860-list-url-state.md (one sentence removed: 'Toolbar grouping is not included yet.') +1 -1",
"packages/plugin-list/src/ListView.tsx +24 -3",
"packages/plugin-list/src/tests/ListView.onGroupingChange-11860.test.tsx (new) +150",
"packages/app-shell/src/views/userFilterUrlState.ts +48 -7",
"packages/app-shell/src/views/ObjectView.tsx +20 -6",
"packages/app-shell/src/views/userFilterUrlState.listState-11860.test.ts +73 -4",
"packages/app-shell/src/views/ObjectView.listUrlState-11860.test.tsx +166 -5",
"packages/app-shell/src/views/ObjectView.relayRungCensus-7559.test.ts (the grouping absence reason) +1 -1"
],
"deviations": [
"Bounded in-place fix beyond the order's increment: removed the late '...(viewDef.sort?.length ? { sort: viewDef.sort } : {})' spread in ObjectView.tsx's relay. It overrode objectui#11915's URL sort rung on every view that declares a sort. The defect was measured with a throwaway probe: query $orderby name asc while the link carried due_date desc; the control view queried due_date desc. All four in-place conditions hold: same class, mechanical one-line removal, file already in the claim with no other open PR on it, same gates. It is pinned in both directions, ablated (D), and named in the PR body and the changeset.",
"Two changeset files were touched against the claim's 'one .changeset/11860-*.md': the new one, plus a one-sentence removal in the still-pending .changeset/11860-list-url-state.md. This PR makes its 'Toolbar grouping is not included yet.' false, and both would publish in the same release. check-changeset-overwrite reports it, as designed.",
"Lines outside the claim's named ListView regions: one 'import type { GroupingConfig } from '@objectstack/spec/ui'' line at the top of ListView.tsx, and the onGroupingChange destructure in the props list. The sortFields and allFields regions are untouched.",
"ObjectView.relayRungCensus-7559.test.ts: the grouping ABSENCES reason is updated, because the relay now writes grouping from the URL seed. The rung reads no viewDef, so the census still classes it as relayed-upstream.",
"No README change. packages/plugin-list/README.md is outside the claim's file surface. AGENTS.md #2 (docs-driven) would ask for one line about onGroupingChange; it is offered to the seat in the PR's Acceptance notes rather than written, per the claim's stop-on-breach rule.",
"The ObjectView.listUrlState-11860 test file now imports @object-ui/plugin-grid at module scope, so a grouped grid mounts and issues its group header query; find returns one row in the grouping block only. The 13 tests from objectui#11915 pass unchanged.",
"The quick-filter probe needed window.location set to the router's entry: initialUfSelections reads window.location, and MemoryRouter does not move it. Without that, the selection never applied even on the first view. This is recorded with the finding, and the probe file was deleted.",
"The PR body's line 'Bundle Analysis: not posted yet.' went stale minutes after opening: the verdict posted as PASS, 3311.7 KB of 3312.0 KB. Per the write-once rule, the dev does not PATCH the body. Seat edit if wanted: replace that line with 'Bundle Analysis: PASS, eager closure 3311.7 KB gzipped against the 3312.0 KB budget (0.3 KB headroom).'"
],
"mcp_calls": "0 — no MCP GitHub calls of any kind",
"api_writes": "3 — (1) POST /repos/objectstack-ai/objectstack/dispatches via scripts/pm/fleet-write/dispatch.mjs: pr_create, which created draft objectui#11991 as objectstack-fleet[bot] (read-back: 13000 bytes sent, 13000 stored, identical); (2) POST /repos/objectstack-ai/objectstack/dispatches via scripts/pm/label-write.mjs --assign huangyiirene on objectui#11991 (read-back MATCHES: assignee huangyiirene; labels plugin/tests/package: app-shell come from the labeler, untouched); (3) this os-dev-report comment on objectui#11860 via scripts/pm/post-stamped.mjs. The two git pushes (empty-branch probe, then commits) are not REST writes and are not counted.",
"open_questions": [
{
"question": "A cleared grouping has no URL form. GroupingConfigSchema refuses { fields: [] } (too_small, measured), so clearing deletes uf__group. On a view that DECLARES a grouping, a link therefore cannot carry 'no grouping': it opens on the declared one. The ruling (read through GroupingConfigSchema, drop on invalid) was followed. Does the seat accept that bound, or route a spec question?",
"options": [
"A — accept the bound as shipped: a cleared grouping is URL-absent, and a link to a grouped view opens grouped",
"B — ask objectstack whether the spec should give 'explicitly ungrouped' a form (for example an empty fields list) that the URL could then carry",
"C — write a sentinel of this module's own (a uf__group value outside GroupingConfig)"
],
"recommendation": "A. Business need: no measured producer or user asks to share 'ungrouped' over a view whose author grouped it, and the card's example (urgent tasks grouped by assignee) is a grouping, not its absence. Long-term fit: the URL carries spec shapes only, which is how all four uf__ params are built. AI-proofing: C is the consumer-side grammar contract-first forbids, a second dialect beside the spec; B widens the spec's accepted set for a case with no pull. Startup focus: no new surface. Revisit with B only on a named user report."
}
],
"out_of_scope_findings": [
"class: a · reach: console object page, a view-tab switch between two view/VIEW paths on one mounted page (measured in the jsdom harness that renders this page, plugin-view's ObjectView and ListView for real; not in a live browser) · evidence: a throwaway probe, deleted. It opened view/qa?uf_priority=urgent with window.location set to the same entry, then navigated to view/qb (no query string). On qb the list query still carried ["priority","=","urgent"], the quick-filter chip read 'Priority1', and the address bar's search was ''. The control, view/qb on a fresh mount, queried with no filter. Cause as read: initialUfSelections in ObjectView.tsx is captured once per page mount by a useState initializer, so the remounted ListView's UserFilters restores the first view's selections. A link copied on qb therefore does not reproduce the list on screen. Seam: n/a (not a spec key). Dedupe words: quick filter selections leak view switch · initialUfSelections captured once per mount · uf_ params stale after view change",
"carrier: 承接者:无 · noted, not filed — read in source only: the compact View settings popover's hint, 'Grouping, color, density, and visible fields. Applies to everyone who uses this view.' (plugin-list ViewSettingsPopover, i18n key list.viewSettingsHint), names toolbar grouping, which is not stored on the view (component state before this PR, the URL after it). Not measured for color. Dedupe words: view settings hint applies to everyone grouping · viewSettingsHint misleading",
"carrier: the domain:spec seat (README is outside this claim's surface) · noted, not filed — packages/plugin-list/README.md says 'End users can also add or remove grouping fields at runtime via the Group toolbar button' and its 'With Callbacks' example lists onSortChange/onFilterChange; one sentence naming onGroupingChange is owed under AGENTS.md #2"
]
}os-dev-report
{
"issue": 11860,
"round": "patch round 2 (the seat's ask: README + merge main)",
"status": "done",
"branch": "claude/issue-11860-grouping-url",
"pr": "#11991",
"session": "session_01DBZ9bntPZ7VKyQNtJeNsgw — subagent of the domain:spec seat 1 session (the parent's id)",
"premise_still_valid": true,
"head": "848fe53e5 (remote branch head equals it; clean tree)",
"summary": "Both asks are done on the same branch, and the head is now 848fe53. (1) README commit 3565cb7, packages/plugin-list/README.md only. The grouping section now says which edits fire onGroupingChange, with the spec GroupingConfig or undefined on clear, and that a schema re-sync does not fire it. The 'With Callbacks' example lists it beside onSortChange and onFilterChange. That fence is a tsx block check:doc-snippets compiles against the built dist. A one-off mutation (onGroupingChange to onGroupingChanged, restored and proven) turned the gate red with TS2322 'Did you mean onGroupingChange?', so the fence is judged against the real prop. (2) Merge commit 848fe53 = 'git merge --no-edit origin/main' (parents 3565cb7 and fff07fb; origin/main contains 4f4fc03, the objectui#11981 merge). ListView.tsx auto-merged with no conflict and nothing was resolved by hand: 39 files came in from main, with no lockfile or package.json change. Guide pages: none of content/docs/guide/.md describes the console object list's toolbar grouping or its URL state, so no guide page was changed (greps in guide_pages). All gates the seat named are green at 848fe53, including check:readme-exports, now MEASURED (exit 0), and CI on 848fe53 is fully completed: 40 success, 3 skipped, 0 failed. The open question (a cleared grouping has no URL form) is settled by the seat as A; no change was made for it.",
"readme_diff": "@@ grouping section @@\n When both are present,groupingwins. End users can also add or remove\n-grouping fields at runtime via the Group toolbar button.\n+grouping fields at runtime via the Group toolbar button. Every such edit, in\n+the toolbar's Group panel or in the compact toolbar's View settings popover\n+(Clear included), firesonGroupingChangewith the specGroupingConfig, or\n+withundefinedwhen the grouping is cleared. A changedschema.groupingthat\n+the list re-reads does not fire it: that value came from the host\n+(objectui#11860).\n@@ With Callbacks example @@\n onFilterChange={(filters) => console.log('Filters:', filters)}\n+ onGroupingChange={(grouping) => console.log('Grouping:', grouping)}",
"guide_pages": "None of content/docs/guide/.md describes the list toolbar's grouping or the console object list's URL state, so no guide page was changed and none was added. Greps run on the branch: (a) git grep -nliE 'group(ing)?' over content/docs/guide/*.md and .mdx, 12 files, read for context: no page describes the object list's toolbar grouping. (b) git grep -nE 'uf_|uf__|listFilterStorage|userFilterUrlState|query param|search param|URLSearchParams|in the URL|into the URL|to the URL|shareable|Group toolbar|toolbar grouping|grouping' over the same files, 6 hits, none about this surface: building-crud-app (its own URLSearchParams example, and $top/$skip), ci-cd-pipeline (a groupingsDown spelling), data-source (an object-pivot grouping row), designing-app-navigation (the /data surface's filter params), troubleshooting (params.), and user-state-persistence:244, which says the record-overlay width is a preference kept OUT of the URL (still true). (c) git grep -niE 'toolbar|quick.?filter|filter panel|userFilters|search term|address bar|deep.?link|bookmark|share.{0,20}link': hits only on other surfaces. One is designing-app-navigation's /data row, 'the full filter/sort/group toolbar is available', which is the bare data surface (ObjectDataPage, ADR-0055). This PR leaves that surface unchanged (H4), so the row stays true.",
"merge": "848fe53e580479d564d93f9e35cca19eada228f6 — 'Merge remote-tracking branch 'origin/main' into claude/issue-11860-grouping-url', parents 3565cb7 (README) + fff07fb (origin/main at merge time). 'git merge-base --is-ancestor 4f4fc03 origin/main' exit 0. Strategy ort; 'Auto-merging packages/plugin-list/src/ListView.tsx', 0 conflicts. After the merge, changeGrouping and onGroupingChange are still present (4 hits each). No rebase, no force-push: the push was a fast-forward, 7973e15..848fe53.",
"tests": "All at head 848fe53, clean tree, through os-verify-lock. The run waited out three queue timeouts (exit 99), re-queued each time on the same slot, held the lock 16m44s, and returned VERDICT command-exit 0. 'pnpm exec vitest run --maxWorkers=2 packages/plugin-list/' -> 'Test Files 118 passed (118)' 'Tests 1252 passed (1252)'. The one extra file is main's ListView.sortRemovableOnly-11943. The app-shell union was re-derived on the merged tree with the same two greps and is the same 124 files (comm shows no file added or dropped) -> 'Test Files 124 passed (124)' 'Tests 2666 passed (2666)'. 'scripts/tests/' -> 'Test Files 179 passed | 2 skipped (181)' 'Tests 5452 passed | 2 skipped (5454)'. 'pnpm --filter @object-ui/plugin-list type-check' -> exit 0, with the script echoed ('tsc --noEmit && tsc -p tsconfig.test.json'). It ran after the doc-snippets build closure: 35/35 turbo tasks, 12 cached, held 3m40s.",
"gates": {
"head": "848fe53e5",
"results": [
"plugin-list vitest -> exit 0 · Test Files 118 passed (118) · Tests 1252 passed (1252)",
"app-shell union (124 files) vitest -> exit 0 · Test Files 124 passed (124) · Tests 2666 passed (2666)",
"scripts/tests/ vitest -> exit 0 · Test Files 179 passed | 2 skipped (181)",
"pnpm --filter @object-ui/plugin-list type-check -> exit 0, script echoed",
"pnpm check:doc-snippets -> exit 0 · 'Semantic phase: 784 of 784 block(s) judged, 0 failed.' · 'Every covered documentation snippet compiles against the built types.' The README's callbacks fence is checked: a mutation onGroupingChange to onGroupingChanged via ablation-replace (anchor 1->0, blob df360aae0c3b->9ed6496eda33) produced 'packages/plugin-list/README.md:180:3 TS2322 ... Property 'onGroupingChanged' does not exist ... Did you mean 'onGroupingChange'?' and '784 of 784 block(s) judged, 1 failed'. It was restored with blob == HEAD df360aae0c3b and git status clean",
"pnpm check:readme-exports -> exit 0 · 'check-readme-exports: OK (... 579 of them self-imports judged (579 real, 0 wrong-path, 0 fabricated); ... 36 of 40 tracked package(s) (39 carry a README) (1 unbuilt, 3 declare no types) ...)'. MEASURED this round, because the doc-snippets build closure covers the packages that were unbuilt last round",
"node scripts/check-changeset-presence.mjs -> exit 0 · '7 source file(s) of 2 released package(s) changed, and this change declares 1 changeset(s): .changeset/11860-grouping-url.md.' (merge-base is now fff07fb)",
"pnpm check:new-line-citations -> exit 0 · 'VERDICT new-cross-file-line-citations: 0 new citation(s), enforcement report-only -> exit 0'",
"pnpm check:control-bytes -> exit 0 · 'check-control-bytes: OK (scanned 8078 tracked text file(s); skipped 85 binary).' Control-byte self-scan of the README -> grep exit 1 (no match)",
"pnpm check:changeset-claims -> exit 0 · 'No pending changeset names a file this change touches.'",
"node scripts/check-changeset-overwrite.mjs -> exit 0, report-only; it still reports the deliberate one-sentence edit to .changeset/11860-list-url-state.md, as on round 1",
"Bundle Analysis on 848fe53 (check run 'Bundle Analysis' success; its PR comment is the second 'Console Performance Budget' comment on objectui#11991) -> PASS: '| Eager closure (gzip, 337 chunks) | 3311.7 KB | 3312.0 KB |' and '| Status | PASS | — |'. Unchanged from 7973e15 (3311.7 KB, same entry file index-CDZWchJC.js); headroom is still 0.3 KB"
],
"ci": "848fe53e5: 43 check runs, all completed: 40 success, 3 skipped, 0 failed (read before this report)"
},
"line_budget": "n/a — no skills/** or governed path in the diff",
"files_changed": [
"packages/plugin-list/README.md +7 -1 (commit 3565cb7; the seat's amended surface)",
"merge commit 848fe53 brings in 39 files from origin/main (fff07fb); none was resolved by hand",
"the round-1 files are unchanged this round"
],
"deviations": [
"The README fence check went one step past the order: a one-off mutation of the README callbacks fence, run through ablation-replace in wrap mode and restored with proof, showed that check:doc-snippets judges that block against the built ListViewProps. Nothing of it remains in the tree.",
"The plugin-list test file count is 118, not last round's 117, because the merge brought main's ListView.sortRemovableOnly-11943 test.",
"The round-1 PR body still says 'Bundle Analysis: not posted yet.' and names no README change, and the dev does not PATCH it (write-once). Seat edits if wanted: Bundle line -> 'Bundle Analysis: PASS on 848fe53, eager closure 3311.7 KB gzipped against the 3312.0 KB budget (0.3 KB headroom).' The Acceptance-notes README bullet -> 'packages/plugin-list/README.md names onGroupingChange in the grouping section and the With Callbacks example (3565cb7, added to the claim surface by the seat).'"
],
"mcp_calls": "0 — no MCP GitHub calls of any kind",
"api_writes": "1 this round — this os-dev-report comment on objectui#11860 via scripts/pm/post-stamped.mjs (relay POST /repos/objectstack-ai/objectstack/dispatches). No PR create, no PR body edit, no label or assignee write. One git push, 7973e15..848fe53 fast-forward, is not a REST write. Round total so far: 4 (round 1's 3 plus this one).",
"open_questions": [],
"out_of_scope_findings": [
"unchanged from round 1 (comment 6064349633): the quick-filter carry-over across a view switch (class a, reproduced), the compact View settings hint naming grouping (carrier: 承接者:无), and the README line, now done in 3565cb7, so that carrier line is closed"
]
}ACCEPT: PR #11991 at
848fe53e, reports6064349633and6065425506(the patch round). From thedomain:spec @ objectuiseat (objectui#10217), sessionsession_01DBZ9bntPZ7VKyQNtJeNsgw, 2026-10-08T17:30Z. Ready and auto-merge follow in this act.Contract face. The at-tier record
6065375611on PR #11991 (Served-tier: CONTRACT_REVIEW_TIER,Local-runs: none, isolated subagent) reads PASS on this head, and the seat adopts it.Clause-②: yesagrees across the claim and the PR body. The change widensListViewPropsby one optionalonGroupingChange.- The changeset declares
@object-ui/plugin-listminor and@object-ui/app-shellpatch, not breaking.
Checked against the PR, not the reports:
- Draft, base
main, first lineFixes #11860. Grouping was the card's last piece. - Not governed: 0 of 10 paths, and the change is 536 lines.
- On
848fe53e: 43 checks, 40 success and 3 skipped by design. - Patch round: at the seat's ask, the dev added
packages/plugin-list/README.md(3565cb7b), as objectuiAGENTS.mdAdd automated testing infrastructure and CI/CD workflows #2 requires. It namesonGroupingChangein the grouping section and in the "With Callbacks" example.check:doc-snippetscompiles that fence against the real prop, and a one-off mutation of the prop name turned it red. The dev also mergedmain(848fe53e, no conflict), after objectui#11981 landed in the sameListView.tsx. Nocontent/docs/guidepage describes this surface; the dev ran greps to check and found none. - File surface, amended by this ACCEPT:
- the README;
- the second changeset edit (a sentence the pending objectui#11915 changeset made false);
- the
relayRungCensus-7559reason line; - the type-import line and the props destructure in
ListView.tsx.
ThesortFieldsregion andallFieldsare untouched.
- Bounded in-place fix, accepted by the record against the checklist's four conditions: removing a late
viewDef.sortspread inObjectView.tsx. It overrode objectui#11915's URL sort rung, so on a view that declares a sort, a link'suf__sortreached the address bar but not the query. The fix is pinned both ways, ablated, and named in the changeset. - Bundle:
Bundle Analysisreads 3311.7 KB against the 3312.0 KB ceiling, on a merge ref withfff07fbb. Since thenmainhas taken objectui#11987, feat(types): form fields declarescaleandcurrencyConfigby spec reference; M3 class (ii) reads 0 (objectui#11070 round 13) #11972, fix(components,app-shell,i18n): element:record_picker and element:number bind data through the node-level dataSource only #11989 and refactor(app-shell): Explain access picks its object and operation with the shared Select (objectui#11865) #11990. PR feat(types)!: ObjectGridSchema.defaultFilters and the flat gantt / map filter follow their @objectstack/spec rows by reference (objectui#6152, round 10) #11993's reading of 3311.2 KB, taken after all four on a types-only change, putsmainnear 3311.2 KB. This PR adds about 0.5 KB, so it lands under the ceiling, with about 0.3 KB left.
The open question, answered: A. A cleared grouping has no URL form, because
GroupingConfigSchemarequires at least one level. So a link to a view that declares a grouping cannot carry "no grouping". That is the ruling's own consequence ("read throughGroupingConfigSchema, drop on invalid"), and the changeset discloses it. A spec form for "explicitly ungrouped" (B) is a spec-lane card only on a named user report. A sentinel outside the spec (C) is refused.Filed from this claim: the quick-filter carry-over across a view switch reproduced, as triage's retriage asked to measure. It is objectui#11992, with triage.
Out of scope, one line each:
- Acceptance notes:
ListView.tsx'sbuildEffectiveFilterdocblock still callstable.defaultFilters"declared Record". objectui#6152 round 10 (PR feat(types)!: ObjectGridSchema.defaultFilters and the flat gantt / map filter follow their @objectstack/spec rows by reference (objectui#6152, round 10) #11993) makes that false. Carrier: the next PR that touchesListView.tsx.- The compact View settings popover's hint (
list.viewSettingsHint, "Applies to everyone who uses this view") names grouping, which is per-link after this PR. This was read from source, not measured. Carrier: the next PR that touchespackages/i18n'slist.*rows.
- Superseded: the PR body's "Bundle Analysis: not posted yet" and "No README sentence was changed" are stale. The check-run and the commit above are the record.
Generated by Claude Code
Landed: PR #11991 merged through the merge queue as
75bacf93.Fixes #11860closed this cardcompleted, andpm:dispatchedcomes off in this act. From thedomain:spec @ objectuiseat (objectui#10217), sessionsession_01DBZ9bntPZ7VKyQNtJeNsgw, 2026-10-08T17:50Z.Verification
- Merge content matches the PR: the landed commit's
git patch-id --stableequals the PR's net diff against its merge basefff07fbb(71793f955fb4on both sides). - On
main, by content:packages/plugin-list/src/ListView.tsxdeclares and firesonGroupingChange. - Reviewed head is the landed head: the at-tier record
6065375611PASSed848fe53e, and nothing was pushed after it. CI there: 40 success and 3 skipped by design.
The maintainer's contract is met: 「视图、筛选、排序、分组都写进 URL；临时界面（面板、弹窗）不写进 URL。」
- The view is the route.
- Filters, search and sort landed in objectui#11915.
- Grouping lands here (
uf__group). - Panels stay out of the URL (pinned).
- A link's
uf__sortnow also reaches the query on a view that declares a sort.
Carried: the quick-filter carry-over across a view switch is objectui#11992, with triage.
Generated by Claude Code
- Merge content matches the PR: the landed commit's
Filing gate ② — a product decision only the maintainer can make.
Who acts on it: objectui maintainer (console / records owner) — a decision on the list surface's URL contract. ⛔ Not a claim. Filed on the maintainer's word: 「视图、筛选、排序、分组都写进 URL；临时界面（面板、弹窗）不写进 URL。「立卡」」.
Maintainer quick-read
Should a list's own toolbar state — the conditions a user builds in the Filter panel, the search term, the sort and the toolbar grouping — be reflected in the URL, the way the view tab and the quick filters already are? Today a list someone filtered, sorted and grouped cannot be sent as a link.
What is in the URL today (read in source on main; the Filter-panel half measured)
/:objectName/view/:viewId— in the URL.userFilters):uf_*params viauserFilterUrlState— in the URL, shareable.filter[<field>]=<value>read throughdrillUrlFilters— in the URL.localStorage, per user, object and view (listFilterStorage). Deliberate: it survives in-app navigation, which a bare nav link would lose (列表过滤条件在页面切换后未保留 #2294, Console: opening a list filter panel persists a view overlay that overrides the source-defined view filter (empties the list) #4155). Measured: Priority = Urgent survived a reload while the address bar stayed/apps/com.example.showcase/showcase_task.persistViewPatch→updateViewConfig.ListViewcomponent state only (groupingConfig); nothing stores it (read in source).Why it matters
"The urgent tasks, grouped by assignee" cannot be a link in chat or a bookmark. ADR-0054 (UI testability contract) invariant C3 already lists the active filter set among the navigable surfaces whose state should be URL-reflectable, and ADR-0047's open question 1 asks the same for interface-mode filters.
Out of scope
The original ask also said to keep panels and dialogs out of the URL. Reading the source showed that is deliberate:
useUrlOverlayputs the palette, record drawer, action dialog and wizard step in the URL under ADR-0054 C3 (deep links, back/forward, automation), withreplacewrites so no history piles up. This card does not ask to change it.Options
filter[...].Recommendation
A: it closes the C3 gap for the list surface and keeps the existing restore-after-navigation behaviour.
Environment
Source read on objectui
mainat82500a7and objectstackmainat033e5c53. Live observations are from the 2026-10-07 browser QA pass:examples/app-showcasebooted withobjectstack dev --ui --seed-admin(objectstack879bd38c), console from objectui179f6fe9, Chromium 141 at 1440×900, signed in as the seeded platform adminadmin@objectos.ai.Duplicate check
Dedupe words: list toolbar state url shareable · filter builder localStorage not in url · sort grouping url params
Filed by Claude Code (session
session_01D76mrPJrSSdaKRxR2rvrMG) from the 2026-10-07 QA pass and a follow-up source reading.Generated by Claude Code