Repository navigation
test(mcp-management): click and tick the fake clock inside act - #27915
Conversation
Make website/src/test/McpManagement.test.tsx stop emitting React act() warnings, and drop its row from website/integration/act-warning-baseline.json so the guard in website/integration/setup.ts checks it.
First Principles Review (Opus 5.5) — ⏭️ skippedRevision |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing to check: the change only touches one test file and the act-warning baseline, so no user-visible control changed and no screenshot was needed. [UX-REVIEWED] e9bd2ce Description read[DESCRIPTION-READ] 6faa497cdedb5f52d99417405d63f73145b2c70914d53fac5dee405d09311df1 That sha256 names everything this verdict read: the PR description, plus the 3 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the change only touches tests, and it removes the warnings at their source by ticking the fake clock inside act instead of hiding them. Notes
[DESIGN-REVIEWED] e9bd2ce Description read[DESCRIPTION-READ] 6ee7fd0a3c6d8f2f906411ce0a9364227a87636eeee3682abcf84a5dafcd1ed1 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
bolichen97
left a comment
There was a problem hiding this comment.
Approving at e9bd2ce.
This is test-only. McpManagement.test.tsx moves its clicks and fake-clock ticks inside act:
btn.click()andsharing.click()becomefireEvent.click.- Each bare
vi.advanceTimersByTimeAsyncis wrapped inact. vi.waitForunder fake timers becomes a newtickUntilhelper. It retries the same check, advances the fake clock by 50 ms insideactbetween tries, and runs the check one last time after its timeout so a failure still reports the real assertion.
Every assertion is unchanged; only how the test waits changes. With the act warnings gone, the file is removed from act-warning-baseline.json, so a new warning there now fails the run.
Gates at this head: 0 failing checks; Design, UX, Opus and GPT pass (First Principles skipped); no review requests changes; mergeable. Only test shards and PR Readiness were still running, and auto-merge waits for them.
Problem / Motivation
Goal:
website/src/test/McpManagement.test.tsxstops emitting React act() warnings, so its row leaveswebsite/integration/act-warning-baseline.jsonand the guard inwebsite/integration/setup.tschecks it like any file outside the list.Before this change, a run of the file emitted 36 act() warnings (late state updates in
McpManagement): a state update landed after the test's last barrier. Because the file is on the baseline, the guard kept those warnings out of the log.Why it matters
website/docs/testing.md(Rules every test keeps) treats an act() warning as a finding: the update lands after the assertions, so whether the test sees it depends on timing. The baseline is shrink-only, and each file that leaves it is held to the rule from then on.Not a goal
What changed (motivation → approach → change)
The tests now wait for the state the late update produces before they finish, so the update lands inside the test's barrier.
website/src/test/McpManagement.test.tsx:import { act, render, screen, cleanup, waitFor, fireEvent, within } from '@testing-library/react'website/src/test/McpManagement.test.tsx:/**vi.waitForticks the fake clock outside act; this polls the same way, each tick inside act. */website/src/test/McpManagement.test.tsx:await act(async () => { await vi.advanceTimersByTimeAsync(50) })website/src/test/McpManagement.test.tsx:await tickUntil(() => screen.getByText('fresh-mcp'), { timeout: 5000 })website/src/test/McpManagement.test.tsx:await act(async () => { await vi.advanceTimersByTimeAsync(2 * 1000) })website/src/test/McpManagement.test.tsx:await tickUntil(() => screen.getByRole('button', { name: M('share_all_confirm', { count: 2 }) }), { timeout: 5000 })website/src/test/McpManagement.test.tsx:await tickUntil(website/src/test/McpManagement.test.tsx:await tickUntil(() => screen.getByText('fresh-mcp'), { timeout: 5000 })website/integration/act-warning-baseline.json: removedsrc/test/McpManagement.test.tsx.Backwards compatibility
Compatible: test-only change.
Removes nothing: a baseline row is a tolerance the guard stops needing, not a capability.
Tests
KIROCREW_ACT_STRICT=1 npx vitest run src/test/McpManagement.test.tsxfromwebsite/: 7 of 51 tests fail on act() warnings.integration/setup.refactor.determinism.test.tspasses (the baseline stays sorted, unique and existing);npx eslintis clean on the file.Manual verification
N/A: test-only change; the strict runs above are the check.
Screenshots / video
Why no screenshot: nothing renders; this change touches one test file and a test baseline only.
No issue exists; burn-down of the act() warning baseline that #17673 introduced.