Repository navigation
Conversation
Decision takenQuestion. #916's renewal line marks the ending Inbox item and an observed Options.
Taken: A. No throwaway code, and nothing is decided that #914 or #915 might change. Folded into #916's Depends on, marked pending the maintainer. To revert: drop the two lines from Depends on and run the build on this branch as it is. Written by an agent. |
Review ledgerRound 1: f49b192 R1 reopened in round 2 (matched renewals still rescheduled the polls), fixed at the fix-check, verified by coordinator. {"findings":[
{"id":"R1","role":"code","class":"every renewal reschedules the polls it patches","outcome":"fixed","path":"ui/src/state/queries.ts","round":"1","fix":"0e82b04e6a557b01b38c420bf9c0a1a131172bd1"},
{"id":"R2","role":"code","class":"gateway caller-chosen lease id gets wrong key","outcome":"fixed","path":"ui/src/model/renewal.ts","round":"1","fix":"03ec8e21f9b432bf9dd97a98f03cff537d3f7159"}
]}Written by an agent. |
Review notesNot blocking, not verified. Each is one reviewer's claim.
Written by an agent. |
…ion's menu waits for the frozen clock to show the answer it needs (#916)
|
CI fix after review: Written by an agent. |
…quests a minute (#918) (#940) Closes #918 ### Status Implement: done (1/1 green) Review: round 1 passed, no blocking findings Hardware: n/a Gate: waiting for the stack - [x] The console browser lane passes with `big-fleet.spec.ts`: audit ran it with `long-lists.spec.ts` on Chromium, 24 green; CI console lane. - [x] Setting the intervals in `ui/src/state/queries.ts` back to 1 s makes it fail: audit set all six to 1 s, every page failed the ceiling (Overview 295, Queue 237, Activity 239, Leases 182, Inbox 120, Fleet 118, Usage 120 in 60 s). #905 Completion conditions: - [x] Idle tab ≤ 40 `/v1` requests in 60 s on each of the seven pages, 50 workers / 500 devices: `big-fleet.spec.ts` (this PR). - [x] A lease granted through the gateway shows on Leases within about a second: event map and burst window, #926 (#914) `ui/src/state/queries.test.ts`. - [x] Leaving a page stops its reads; coming back within a minute shows kept data: #926 (#914) polling and drop tests. - [x] A second read of an unchanged list answers 304: #920 (#919) `src/http/app.test.ts`; the console sends the tag, #927 (#916). - [x] Sidebar count reads only `GET /v1/inbox`; Inbox and Needs you read it plus `/v1/workers`: #937 (#915) `e2e/console/shell.spec.ts` and `ui/src/model/overview.test.ts`. ## Assumptions - Assumption: the stub builder is typed by `WorkerView`, derived in `big-fleet.spec.ts` as `z.infer<typeof workerViewSchema>` from the contract's exported schema; `src/contract/` is unchanged. - Assumption: `leases` and `pad` builders moved to `stub-answers.ts` with `answer`, so both specs share them. ## Review Spec review: 0 blocking, 0 fixed, 1 notes. Code review: 0 blocking, 0 fixed, 0 notes. Claims review: 0 blocking, 0 fixed, 0 notes. Rounds: 1. Tokens: 23.5M. Runs: 7. Cost USD: 1.8082. Agent seconds: 2527. Assessment: complete *Written by an agent.*
Closes #916
Status
Implement: done (3/3 green) Review: round 2 + fix-check passed, 2 fixed Hardware: n/a Gate: waiting for the stack
Done when
pnpm vitest run ui/srcpasses with the new tests. Delivered byui/src/data/api.test.ts(a read with a tag),ui/src/state/queries.test.ts(a route's ETag,a lease renewal, two tests) overui/src/data/api.ts,ui/src/data/live/connection.ts,ui/src/state/queries.ts,ui/src/model/renewal.ts; 49 files, 482 tests pass.If-None-Match. Delivered bye2e/console/full-fetch.ts(fetchFull) used by every stub that calledroute.fetch(); the stubbed specs (update, long-lists, activity, workers-gateway-stops-mid-fetch) pass in Chromium; the whole lane is CI's.Assumptions
origin/task/914andorigin/task/915merged in, because the spec's query layer and inbox live there.@tanstack/react-queryin Fallow'signoreDependencies— it is a bundled console library kept as a dev dependency, like its siblings in that list.Review
Spec review: 0 blocking, 0 fixed, 0 notes. Code review: 2 blocking, 2 fixed, 0 notes. Claims review: 0 blocking, 0 fixed, 1 notes.
Rounds: 2. Tokens: 22.1M.
Runs: 14. Cost USD: 5.4613. Agent seconds: 3723.
Assessment: complete
Written by an agent.