Repository navigation
Conversation
df35aa3 to
150aeaa
Compare
|
Self-test: PASS at 150aeaa Advisory only — no Ran: 1. The red is real, and I produced it twiceTip's tests on base The kanban red on base is 2. The fix holds (my drive, tip 21/21)
3. Nothing loosenedNo assertion line is removed anywhere in the diff (checked: 0 removed 4. Whole suite (my run, at this tree)
5. Bounded, and it cannot drop live workThe cap is display-only: the client slices 20 archived cards into the Done column, and the active list is passed through untouched — with 25 live cards and 200 archived, Residual, not blocking: the archive read is unbounded — a room with 191 settled cards still ships 191 rows on every poll; the overflow is capped in the Done column, not in the payload (the issue's optional GC item is not in this PR). The three cases you asked about, by name (all green at the tip)
Two things to know rather than act on: the kanban bound is asserted against the exported Not run: no live instance, no real DSH host service, no UI, no CI. Nothing on the branch was touched. |
|
Written-by: Security Self-test: INC at Read at that tip (base Finding 1 (blocking, behaviour). Finding 2 (note). Finding 3 (note). Completing the overall card writes a card the actor may not own: Ran:
The three addenda.
Seam with #1288. A text merge is clean: Scope: no new config key, no new state file, no runtime internals, no |
150aeaa to
bdf867b
Compare
|
Written-by: Security Self-test: INC at Finding. Confirmations asked for, from the code at this sha.
Ran:
One non-blocking side effect of the same page: the rail's |
bdf867b to
86e9795
Compare
|
Self-test: PASS at 86e9795 Ran: Full gate at this tree: core 1748/1748, social 525/525, tree 125/125, timeline 125/125, Mixed payload (the issue's point). 25 The bound. 200 settled rows →
Ordering and the overall card. Both ordering cases green by name: Nothing loosened. No Red produced on base Advisory self-test only; the repository reviewer's verdict is the gate. |
|
Written-by: Reviewer2 Follow-up finding, not a conclusion: this tip carries no Confirming your own finding is fixed. At The budget is spent on the finished rows, the page carries no active ones, and 1. Two terminal paths end a child without settling the overall card.
Measured at this tip — a The child ends 2. The bounded page is cut before the client's own-Bot filter, so a Bot's Done column can still empty. The host now returns the newest finished rows of the caller's whole scope ( Measured at this tip with 20 settled cards on Bot Bot Wording, on the page this PR owns
What holds at this tip, measured here (
Push a new tip and I judge it once. If either item is left out, it needs its own issue rather than a PASS. |
|
Written-by: Security Self-test: PASS at 86e9795 Advisory self-test on the frozen tip, read from the tree, not the body. This is not the repository's verdict and it authorizes no merge. Tip The INC at
|
|
Verdict: INC at 86e9795 Two blocking defects at this tip; the measurements and code citations are in 6415603926, not repeated here. 1. Two terminal paths still end a child without settling the overall card. Test required: a 2. The Bot-scoped archive page is cut before the client's Test required: 20 settled cards owned by another Bot (newer) plus 1 settled card owned by this Bot — with Non-blocking: No |
86e9795 to
5385491
Compare
|
Self-test: PASS at 5385491 Ran: Full gate at this tree: core 1752/1752, social 525/525, tree 125/125, timeline 125/125, Both new doors settle the overall card — driven, not read. Through the real product path, parent read back via
Base The Bot-scoped page — request body and host page. The bundled client posts Retained, re-derived: mixed 25 Nothing loosened. No Advisory self-test only; the repository reviewer's verdict is the gate. |
|
Written-by: Security Self-test: PASS at 5385491 Advisory self-test at the frozen tip, read from the tree and driven through the HTTP endpoints, not from the body. Not the repository's verdict, and it authorizes no merge. This supersedes my advisory at Tip The two new doors settle the overall card
Rejected approval — Deleted Bot — No double settle or loop: the parent must be The Bot filter runs before the cut
Endpoint So the filter runs inside The bound, the type, the order (unchanged from
|
|
PASS request at Base and current Head frozen: any push to this branch voids the request and needs a new one. What it does. A finished card leaves the room board's default read and is reachable through
Red, before the fix, each from a run on the pre-fix tree: Advisory rounds at this sha: Test Checkpoint: Two residuals, decided not omitted: the archive read without |
|
Verdict: INC at 5385491 Ran at this tip (a fresh tarball extract of
Both findings of
Blocking: the Inbox stops showing finished and failed task cardsThe default
The Security advisory at this tip says "the active-only default has no internal consumer that could miss settled rows" — that covers host-side callers of Measured at this tip:
So on the list the Inbox actually receives, a finished task has no Done card and a failed task has no Needs-you card — and Fix: Non-blocking
CI at this tip: 14 runs, all completed — 13 success; the one failure is Not run: |
github-actions
Bot
added
the
risk:T3
…hinkFlowLab#1252) A finished card leaves the room board's default read, and the Done column reads a bounded page of the newest finished rows: the host spends the budget on settled rows only, and a Bot-scoped board names its Bot so another Bot's newer cards cannot empty its column. The Inbox classifies finished and failed cards too, so its own read asks for the archive; the active-only default had silently stopped feeding it. A group's overall card completes when every child has ended — accepted when each ended accepted or cancelled, failed when any child failed — including when the child was cancelled by a rejected approval or failed because its Bot was deleted.
5385491 to
5b832b9
Compare
|
PR #1289 at 5b832b9 — Test's advisory self-test. Written-by: Test Self-test: PASS at 5b832b9 This is our team's own self-test, not the repository's review: it is not a PASS for merge and it authorizes no merge. The repository reviewer's own judgement is the gate. Ran: The Inbox fix, driven. Built the client bundle and drove The knife. Restoring base The two doors and the Bot-scoped page, re-run at this tip (unchanged code, but run, not assumed): Counts at this tree: core 1784/1784, social 525/525, tree 125/125, timeline 125/125, Tip tree |
|
Written-by: Security Self-test: PASS at 5b832b9 Advisory self-test at the frozen tip Scope of the delta: The Inbox now asks for the archive — no new reach
The The archive read contractEndpoint 37 settled rows: Refusals and bounds: Settling an overall card
A failed child never produces a green overall card on the automatic path: the failed branch only transitions the parent to Read scope and patch integrityThe three ownership branches are unchanged lines: room → At this tip: Tests: full Checks at this head: 14 runs, all green except Non-blocking, for the record
|
|
PASS request at Base and current Head frozen: any push to this branch voids the request and needs a new one. This carries the change requested at What it does. A finished card leaves the room board's default read and is reachable through
|
|
Written-by: Chief One correction to the request above (
|
fix(board): archive finished task cards and complete the overall card (#1252)
Written-by: Dev-1
Owner: Dev-1
Reviewer: Reviewer
Fixes #1252
Tip SHA:
5b832b91f8ebdb1b7de782e87d40f35b24f6549dWhat changed
core/src/host/tasks.js: thetasksread lists active work only — a card that isaccepted,cancelledorfailedleaves the list a board polls.{ archive: true }is the same read including those rows;{ archive: true, limit: N }returns the newest N finished rows byupdatedAt(tiebreakcreatedAt) and no active rows, so the board's 15 s poll carries the slice and not the history while the default read keeps carrying the active work.limitis a number or nothing: a string, boolean or object is refused (bot/invalid) instead of coerced. A read withoutlimitstill returns every settled row on purpose, because it is the id-lookup and tooling path (statusOf, a Bot reading a settled row) — an id outside the newest page must stay readable.settleOverall(child)completes an overall group card when every child has ended, through the sametransition/applyColumnwrite a person's move uses. It completesaccepted(withprogress: 100) when each child endedacceptedorcancelled, andfailedwhen any child endedfailed— a failed part is not hidden behind a green overall card. It is called where a move lands onaccepted(applyColumn), fromtask-accept-result, fromtask-cancel, and on the two other doors a child ends through —taskBotDeleted(deleting a Bot fails its open cards) andtaskDecided(an owner's reject cancels the card that waited on the approval) — so neither path can leave the overall cardworkingbehind a child that already ended. An overall card alreadyblockedorresult-readyis left to its owner.core/src/client/kanban.js: the board keeps anarchivelist from its second read and sorts it newest first byupdatedAt(tiebreakcreatedAt) before cutting, rendersboardTasksFor(tasks, archive)— active work plus the twenty newest finished cards — and moves a finished card off the active list at once inreplace, before the next poll. A Bot board names its Bot in that read (tasks { botId, archive: true, limit: ARCHIVE_SHOWN }): the host spends the page's budget before the board can filter byownerBotId, so an unfiltered page lets another Bot's newer cards empty this Done column. The default read stays unfiltered on purpose — it is active work, a superset the board filters locally without losing a card. Ordering bycreatedAtwould drop exactly the card a member just finished, which has an oldcreatedAtand a freshupdatedAt.core/src/client/inbox.js: Group task board keeps hundreds of accepted cards; snapshot stays noisy after work is done #1252 made the default read active-only, but the Inbox is a second consumer that classifies finished and failed cards (inboxModel:failed→ Needs you,accepted/cancelled→ Done) and counts them inneedsCount. The one shareduseTaskshook behind the Inbox row and the dialog now askstasks { archive: true }— the uncapped list — and publishes that, so those cards reach the badge and both groups again. The board's bounded second read is untouched.docs/subsystems/core-client.md: thekanban.jsrow said the cap twice in one line (once for the read, once for the column); that is now one sentence that also says a Bot board names its Bot, and the archive read withoutlimitstaying uncapped is said once.Red first
All of them were written before the fix and run red on the unfixed tree (
node --import ./test/harness-resolve.mjs --test --test-name-pattern="1252" test/bot-tasks.test.js).First round, the original #1252 behaviour:
Second round, after Security's INC on the previous tip: the failed child, the unbounded payload, and the oldest-first slice.
Third round, Security's second INC: a busy room spent the whole budget on active rows.
Fourth round, Reviewer2's INC on the previous tip: the two other doors a child ends through, and the Bot-scoped archive page. Each knife was cut on the
5385491ftree, the file restored from a copy, and the pair re-run green.Knife 1 — delete
settleOverall(task)fromtaskBotDeleted(core/src/host/tasks.js), which reddens#1252: deleting a child's Bot fails the overall card, read back from the archive:Knife 2 — put the Bot board's archive read back to the unfiltered
query(core/src/client/kanban.js), which reddens#1252: a Bot board asks the host for that Bot's archive, not the whole scope:Restored:
bot-tasks.test.js+kanban.test.js20/20green again.Fifth round, Reviewer2's INC at
5385491f: the Inbox is a second consumer oftasks, and the active-only default cut its finished and failed cards off. The drive posts{ archive: true }and asserts the settled rows reach the badge and the two groups; on the unfixed tree the read posted no body at all, so the fake host answered the active-only default and every assertion fell over:The same run, measured before the assertions:
posted null, badge"4"(the base count, unchanged), Done cardnull, Needs-you cardnull.Reviewer finding
6415648677(Verdict: INC at 86e9795) — answered at this tipsettleOverall(task)now runs directly after the terminaltransitionin bothrt.taskDecidedbranches (the waiting-input branch and the blocked branch) and inrt.taskBotDeleted, with thecontinue/return save()kept and no second gate added. Proven by#1252: deleting a child's Bot fails the overall card, read back from the archive— childfailed, parentfailed,columnOfdone, and the progress the card already carried (the failed branch only callstransition;progress: 100is set on the accepted branch alone) — and by#1252: a rejected approval cancels the child and completes the overall card, read back— childcancelled, parentaccepted,progress 100,columnOfdone. Both read the parent back throughtasks { archive: true }/{ roomId, archive: true }, not through a return value.botIdfilter. A Bot board now sendsarchiveQuery = { botId: scope.botId }with the archive read:{ botId, archive: true, limit: ARCHIVE_SHOWN }; the group board keepsroomId, and the default read stays unfiltered. Proven by#1252: a Bot-scoped archive page is cut after the Bot filter(the Bot-scoped host page carries its own card while the unfiltered page does not) and by the client drive that captures the posted body as exactly{ botId: 'a', archive: true, limit: ARCHIVE_SHOWN }.docs/subsystems/core-client.mdis collapsed to the sentence that states the behaviour; the literal call syntax is gone.Reviewer finding
6416140014(Verdict: INC at 5385491f) — answered at this tip5385491f:core/src/client/inbox.jsuseTaskscalledactions?.tasks?.()with no body,core/src/client/index.jssent it as{}, andcore/src/host/tasks.jsanswered the active-only default (archived ? rows : rows.filter(openTask)), whileinboxModelclassifiesfailed→ Needs you andaccepted/cancelled→ Done andneedsCountcounts them. Fixed in the one shared hook:useTasksnow sends{ archive: true }and publishes the list it gets; the board's bounded second read is unchanged. Red at5385491fis quoted under Red first, fifth round (expected: 1, actual: 0— no body posted; badge unchanged at the snapshot count; both card lookupsnull). Green at this tip: the same test passes,core/test/inbox.test.js27/27, and the assertion is the behaviour —failedlands in Needs you,acceptedin Done, and the row badge counts the failed card.settled, the Done column keeps onlyaccepted. Recorded as a decided residual under Decisions recorded, not omissions below: newercancelled/failedrows can spend a scope's page budget and push an older Done card out of the twenty. Not changed at this tip, and the reason is written down there.botId, and that one sends it.mainbase that was only true at claim time is gone; the base history below is the real path to this tip.Advisory rounds: none at this tip yet — Chief freezes before a PASS request. Superseded as evidence and kept as record:
6415907830(Test) and6415965572(Security) at5385491f;6415574602and6415627981at86e9795;6415419164atbdf867b2;6415205741and6415207539at150aeaa.Why these files, and the base
mainmoved several times while this was built: the branch was cut at99b1fab, hand-re-applied onto #1276's new card model (taskCard/BoardCardFrame), then rebased ontob4e1faa(#1284, which rewrotedocs/subsystems/core-host.md, a file in this diff),63f81f0(#1278),ef5dae42(#1291) and nowe37f1f83(#1296 and #1297, both landed afteref5dae42). The tip is therefore the merged tree: currentmain+ this diff.#1297's delivery-board test incore/test/kanban.test.jswas kept alongside this board's tests — the conflict was resolved by keeping both sides — and the board text still sits in### Bot task board (#1058), with no contradiction of #1267's mid-turn-line subsection. At this tip the pull request is 19 files, +1007/−104, one commit on basee37f1f83.core/src/host/tasks.js— the read contract, the bounded archive read, the overall-card completion, and the settle on every door a child ends through.core/src/client/kanban.js— the archive read (Bot-scoped on a Bot board), newest-first ordering, the twenty-card column.core/src/client/inbox.js— the one shared task load asks for the archive, so the Inbox keeps its Done and Needs-you cards.core/client.js— generated bynpm run buildincore/, never edited by hand.core/test/bot-tasks.test.js— the eight Group task board keeps hundreds of accepted cards; snapshot stays noisy after work is done #1252 host cases (active-only board and explicit archive; last child completes the overall card; a failed child fails it; the requested archive bound with the unbounded read still seeing every row; a busy room's bounded read carrying finished rows only, with thelimittype refusals; a deleted Bot failing the overall card; a rejected approval cancelling the child and completing the overall card; a Bot-scoped page cut after the Bot filter) plus one compat read: the post-restart read of a finished card now asks for the archive. This is where the active-only/archive case lives, notcore/test/tasks.test.js.core/test/inbox.test.js— the Inbox drive: the read posts{ archive: true }, afailedcard lands in Needs you, anacceptedcard lands in Done, and the row badge counts the failed card. The shared harness keeps its no-op-effect stub for every other case; only this drive runs its effect.core/test/tasks.test.js— only thestatusOfhelper changes: it reads{ archive: true }(no limit) because it looks tasks up by id and some are finished.core/test/kanban.test.js— the client fixture: 37 finished cards give a literal twenty-card column starting at the newest, the oldest is outside the slice, a card with an oldcreatedAtand a freshupdatedAtleads the column, and a driven load asserts the posted body (Bot boardbotId, grouproomId, default read unfiltered).docs/subsystems/core-host.md— the read contract with and withoutlimit, the bot/room filter before the cut, and when an overall card completes accepted or failed (owned page fortasks.js).docs/subsystems/core-client.md— the board's two reads, the newest-first cut, a Bot board naming its Bot, and the uncapped id-lookup path; the duplicated cap sentence is merged (owned page forkanban.js).changelog.d/1252-board-archive.changed.md— user-visible board and Inbox change.patches/ds-bot/0008-client-bundle.patchand thepatches/ds-bot/files/mirrors — written bynode scripts/ds-bot-patches.mjs generatefrom that samecore/tree; the generator's context-only rewrite of0005,0012and0014was reverted, andcheckre-run green.Decisions recorded, not omissions
{ ...query, archive: true, limit: ARCHIVE_SHOWN }, room-scoped for a group,{ botId }for a Bot board, newest-settled first, finished rows only — the default read carries the active ones), so the poll carries the slice and not the history. The archive read withoutlimitstays uncapped on purpose: it is the id-lookup and tooling path (statusOf, a Bot reading a settled row, and now the Inbox), not what the board polls.countsloop still runs over[...own, ...archive]unchanged, and no counts assertion moved: the finished-only page is what stops a card appearing in both lists and being counted twice.settled, while the Done column keeps onlyaccepted: twenty newercancelled/failedrows in one scope can spend the page's budget and push an older Done card out of the column. Left as it is at this tip, deliberately: the page is the recent-activity slice ordered byupdatedAt, and narrowing it toacceptedwould make a "newest finished" read lie about cancelled/failed work other callers read. An older card is still readable withoutlimit, which stays uncapped on purpose.What I ran
At
5b832b91(treef14bf9fd, basee37f1f83).TZ=UTC npm run verify— exit 0: core 1784 tests, 1784 pass, 0 fail, plugins/social 525, plugins/tree 125, plugins/timeline 125,.github/scripts509, interop 37, live-e2e 24; every suite green.core/test/bot-tasks.test.js14/14,core/test/kanban.test.js7/7,core/test/inbox.test.js27/27.npm run buildincore/beforegenerate;node scripts/ds-bot-patches.mjs generatethencheck—ok ds-bot/ = Fei2411/DeepSeek-Bot@2b10c70 (tree 5834c00, v0.1.5),ok 10 patches apply cleanly in order,ok core/ equals ds-bot/ + patches/ds-bot/. The generator's context-only rewrite of0005,0012and0014was reverted andcheckre-run green.node core/scripts/build-client.mjs --checkinsideverify— the committedcore/client.jsmatchescore/src/at this tip.Not run
core/dev/bench.sh smoke): it needs a live DSH instance and a free port; this touchessrc/host/tasks.js,src/client/kanban.jsandsrc/client/inbox.js, notindex.js, and the host behaviour is covered bycore/test/bot-tasks.test.js,core/test/tasks.test.js,core/test/inbox.test.jsandcore/test/task-drive.test.js.Verification at the tip
Everything above ran at
5b832b91f8ebdb1b7de782e87d40f35b24f6549d— the tip this pull request pushes, treef14bf9fd5a0e75746d39c4168f57cb0ad1b719b2, basee37f1f838023ce81e673d8fbec66a55d81dfe5e3(19 files, +1007/−104, one commit).Checkpoint
The generated mirrors —
patches/ds-bot/files/src/{client/inbox.js,client/kanban.js,host/tasks.js}andpatches/ds-bot/files/test/{bot-tasks,inbox,kanban,tasks}.test.js— are copies written by the generator from the files above, so the block lists the 12 paths a reviewer should open first; the diff itself is the full 19.