test(512): flush LibrariesScreen scan-poll chain deterministically instead of waitFor timeouts #558
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: timothy/ersatztv#558
Reference in New Issue
Block a user
Fixes #512.
What
web/src/screens/LibrariesScreen.test.tsx→ "stops scan polling and refreshes sources once when scans complete" was CI-load-flaky. It fired the poll tick viarunPollTickbut discarded the promiseloadScanStatusesreturns, so the progress-clear state update and the fire-and-forgetloadSources()refetch it triggers landed on real microtasks afteract()had already resolved. The test out-waited that race with twowaitFor(…, { timeout: 5000 })calls, which were marginal on a starved CI VM (timed out on PR #509 / run 918); prior timeout bumps (#447) were diminishing whack-a-mole.Fix
Drive the chain to completion deterministically instead of out-waiting it.
runPollTicknow, insideact(async () => …):await Promise.all(intervalHandlers.map(h => h()))— settles the scan-status poll promise the handler returns (and its progress-clearsetState);await flushMicrotasks()(a singlesetTimeout(0)macrotask) — drains the entire microtask queue, including the fire-and-forgetloadSources()refetch, which by spec has fully settled before any macrotask runs.The mock fetches resolve synchronously on microtasks and
setIntervalis the only mocked timer (setTimeoutstays real), so one macrotask turn deterministically completes both chains with no wall-clock delay. Both original assertions are preserved verbatim — progress label75%gone, and exactly one extra/api/v1/media-sourcesfetch — and the twowaitForcalls + the 20s test-level budget are dropped.Verification
LibrariesScreensuite green 15/15 consecutive runs (npm test -- LibrariesScreen --runin a loop).npm run lintclean;npm run buildsucceeds (only the pre-existing unrelated chunk-size advisory).LibrariesScreen.tsx/libraries.tsuntouched; the component was observable deterministically from the harness alone.Done-when
waitForwall-clock timeouts to observe the poll chainPure web-test change — no server-state effect (independent cross-model review skipped per
process.independent-review-rubric; covered by orchestrator cold-read of the diff + the 15/15 determinism loop).🤖 Generated with Claude Code
The "stops scan polling and refreshes sources once when scans complete" test fired the poll tick via runPollTick but discarded the promise loadScanStatuses returns, so the progress-clear state update and the fire-and-forget loadSources() refetch it triggers landed on real microtasks AFTER act() had resolved. The test out-waited that race with two waitFor({ timeout: 5000 }) calls, which were marginal on a starved CI VM (timed out on PR #509/run 918); prior timeout bumps (#447) were diminishing whack-a-mole. Drive the chain to completion deterministically instead of out-waiting it: runPollTick now awaits the promise the handler returns (settling the progress-clear) and then yields to a single macrotask (setTimeout 0) to drain the microtask queue — including the fire-and-forget loadSources() refetch — all INSIDE act(async () => ...). The mock fetches resolve synchronously on microtasks and setInterval is the only mocked timer, so one macrotask turn completes both chains with no wall-clock delay. The two waitFor calls are replaced by synchronous assertions (progress label gone; exactly one extra /api/v1/media-sources fetch), and the 20s test-level budget is dropped. Verified: LibrariesScreen suite green 15/15 consecutive runs; npm run lint and npm run build clean. No production code changed. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Fixes #512Adversarial review (orchestrator cold-read of the diff + determinism loop). Pure web-test leaf, no server-state effect → cross-model independent review skipped per
process.independent-review-rubric, stated here. Checks: both original assertions preserved verbatim (progress75%gone; exactlybefore + 1/media-sourcesfetch) — not weakened;runPollTickawaits the handler promises then drains onesetTimeout(0)macrotask insideact(), which by spec settles the whole microtask queue (incl. the fire-and-forgetloadSources()refetch) before the macrotask fires — race-free with no wall-clock wait;setIntervalis the only mocked timer sosetTimeout(0)stays real; no production code touched. 15/15 consecutive loop runs green; lint + build clean.Review-verdict: MERGEABLE @
b5596f55⚠️ CI reds on this PR are infra, not code — the runner host hit
No space left on deviceduring the build step. Tracked in server-management#654. On the last run, Functional E2E + all PR-Gates went green here onb5596f5; only the disk-starved EF-migration/Build jobs fast-failed. Merge is pending a clean rerun once the runner-host capacity issue is resolved — no code change needed.