Merge pull request #5815 from deepseek-harness/test/web-cot-toggle-flake-20261007

test(web): bound plugin switch waits by the profile write chain

apps/web/tests/README.i18n.yaml

@@ -6,8 +6,8 @@
   en: 3e8ecd3509fa944f
   zh: 3fd3ff9fcd1e1e90
 /apps-web-browser-e2e/completion-observations:
-  en: ae500691f64d1e43
-  zh: 2bfc53b52d210b24
+  en: 4a06ba5463b53bb3
+  zh: 70863c70e15c6746
 /apps-web-browser-e2e/these-are-host-face-tests:
   en: 8c406d423c8e2137
   zh: a66a144fd146f56c

apps/web/tests/README.md

@@ -20,6 +20,8 @@ State-sensitive cases use Workspace, admission, attachment, and model-stream bar
 
 [Idle-submission handoff](idle-submission-handoff.e2e.ts) waits up to the scaffold's 30-second turn-completion budget for Host frames and browser placement. Vitest owns scaffold teardown so a replay-consumption error does not replace the failed scenario assertion.
 
+Plugin-manager switches wait up to `PLUGIN_TOGGLE_SETTLE_MS` (10 seconds) for the Host's profile write and the store's follow-up inventory read, the two steps that republish `aria-checked`; that constant's JSDoc records the measured chain.
+
 Explicit scrolling uses `scrollIntoView` from `support.ts`: it resolves the locator again when its old element detaches and checks connection in the same browser task as native scrolling. Scenarios retain their visibility and geometry assertions after scrolling.
 
 ## These are Host-face tests

apps/web/tests/README.zh.md

@@ -20,6 +20,8 @@ pnpm --filter @deepseek-ai/dsh-web-frontend exec playwright install --with-deps
 
 [空闲提交交接](idle-submission-handoff.e2e.ts)对 Host 帧和浏览器中的消息归位的等待使用 scaffold 的 30 秒 Turn 完成预算。Vitest 负责 scaffold 清理，使 replay 消耗检查错误不会覆盖场景中失败的断言。
 
+插件管理器的开关为 Host 的 profile 写入与 store 随后的 inventory 读取预留 `PLUGIN_TOGGLE_SETTLE_MS`（10 秒），这两步完成后才会重新发布 `aria-checked`；链路的实测值记录在该常量的 JSDoc 中。
+
 显式滚动使用 `support.ts` 的 `scrollIntoView`：旧元素脱离 DOM 时重新解析 locator，并在同一个浏览器任务中检查连接状态、执行原生滚动。各场景保留滚动后的可见性与几何断言。
 
 ## 这些是 Host 面的测试

apps/web/tests/cot-translation-paid.e2e.ts

@@ -15,7 +15,9 @@ import {
   acknowledgeReloadConnectionLoss, assertFixtureInventory, captureStableAria, compareOrRefreshGolden, fixtureUserPrompts,
   launchWebScaffold, readPersistedEvents, watchConsole, webSnapshotMode, type WebScaffold,
 } from './scaffold.ts'
-import { connectFreshWorkspaceZh, expandOwningTurnProcess, saveFailureShot, ZH_BROWSER_LOCALE } from './support.ts'
+import {
+  connectFreshWorkspaceZh, expandOwningTurnProcess, PLUGIN_TOGGLE_SETTLE_MS, saveFailureShot, ZH_BROWSER_LOCALE,
+} from './support.ts'
 
 const BUNDLE_NAME = '@deepseek-ai/dsh-experimental-cot-translation-bundle'
 const BUNDLE = fileURLToPath(new URL('../../../packages/experimental/cot-translation-bundle', import.meta.url))
@@ -168,7 +170,8 @@ it.skipIf(MODE === 'record')('reuses saved native Account Flash translations acr
   const panel = settings.locator('[data-plugin-panel]'), card = panel.locator(`[data-plugin-package="${BUNDLE_NAME}"]`)
   const toggle = card.getByRole('switch'); await toggle.waitFor()
   expect(await toggle.getAttribute('aria-checked')).toBe('false')
-  await toggle.click(); await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('true')
+  await toggle.click()
+  await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
   await card.getByRole('button').first().click()
   const provider = panel.getByLabel('翻译服务', { exact: true })
   await provider.waitFor(); expect(await provider.inputValue()).toBe('bing')

apps/web/tests/cot-translation.e2e.ts

@@ -12,7 +12,9 @@ import {
   acknowledgeReloadConnectionLoss, assertFixtureInventory, captureStableAria, compareOrRefreshGolden, fixtureUserPrompts,
   launchWebScaffold, readPersistedEvents, watchConsole, webSnapshotMode, type WebScaffold,
 } from './scaffold.ts'
-import { connectFreshWorkspaceZh, expandOwningTurnProcess, saveFailureShot, ZH_BROWSER_LOCALE } from './support.ts'
+import {
+  connectFreshWorkspaceZh, expandOwningTurnProcess, PLUGIN_TOGGLE_SETTLE_MS, saveFailureShot, ZH_BROWSER_LOCALE,
+} from './support.ts'
 
 const BUNDLE_NAME = '@deepseek-ai/dsh-experimental-cot-translation-bundle'
 const BUNDLE = fileURLToPath(new URL('../../../packages/experimental/cot-translation-bundle', import.meta.url))
@@ -198,7 +200,7 @@ it.skipIf(MODE === 'record')('persists Bing reasoning translation across reopen
   await toggle.waitFor()
   expect(await toggle.getAttribute('aria-checked')).toBe('false')
   await toggle.click()
-  await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('true')
+  await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
   await card.getByRole('button').first().click()
   const provider = panel.getByLabel('翻译服务', { exact: true })
   const target = panel.getByLabel('目标语言', { exact: true })
@@ -315,7 +317,7 @@ it.skipIf(MODE === 'record')('persists Bing reasoning translation across reopen
 
   await panel.getByRole('button', { name: '返回插件列表', exact: true }).click()
   await toggle.click()
-  await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('false')
+  await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
   await body.waitFor({ state: 'detached' })
   await reasoning.getByText(ORIGINAL, { exact: true }).waitFor()
   await reasoning.locator('[data-disclosure-row]').first().click()

apps/web/tests/plugin-manager.e2e.ts

@@ -16,7 +16,9 @@ import {
   SCAFFOLD_DEFAULTS_BUNDLE, assertFixtureInventory, captureStableAria, compareOrRefreshGolden,
   launchWebScaffold, watchConsole, webSnapshotMode, type WebScaffold,
 } from './scaffold.ts'
-import { ZH_BROWSER_LOCALE, connectFreshWorkspaceZh, openSettings, saveFailureShot } from './support.ts'
+import {
+  ZH_BROWSER_LOCALE, connectFreshWorkspaceZh, openSettings, PLUGIN_TOGGLE_SETTLE_MS, saveFailureShot,
+} from './support.ts'
 
 const SNAPSHOT_DIR = fileURLToPath(new URL('./expected/plugin-manager', import.meta.url))
 const MANAGER_EXPECTED = join(SNAPSHOT_DIR, 'manager.expected.md')
@@ -643,7 +645,7 @@ describe('web e2e: plugin manager', () => {
       await toggle.click()
       try {
         await expect.poll(() => teamRows().filter(entry => entry.fiber?.state === FiberState.ACTIVE).length, { timeout: 20_000 }).toBe(3)
-        await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('true')
+        await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
         await action.waitFor({ timeout: 20_000 })
         await action.getByRole('button', { name: '智能体团队', exact: true }).click()
         const teamPanel = teamPage.getByRole('dialog', { name: '智能体团队', exact: true })
@@ -897,10 +899,10 @@ describe('web e2e: plugin manager', () => {
     const bundles = async () => (JSON.parse(await homeFile('profiles', 'scaffold', 'package.json')) as {
       dsh: { profile: { bundles: string[] } }
     }).dsh.profile.bundles
-    await expect.poll(bundles, { timeout: 10_000 }).toEqual([...SCAFFOLD_BUNDLES, '@fixture/bundle'])
+    await expect.poll(bundles, { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toEqual([...SCAFFOLD_BUNDLES, '@fixture/bundle'])
     // A live profile: the row mounts once the whole tree recomposed, the switch is on, and nothing waits for a restart.
     await expect.poll(() => mounted()?.fiber?.state, { timeout: 20_000 }).toBe(2)
-    await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: 10_000 }).toBe('true')
+    await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
     expect(await panel.getByText(/下次启动生效/).count()).toBe(0)
     const snapshot = await captureStableAria(page, '[data-plugin-panel]', scaffold.workspaceCwd)
     await compareOrRefreshGolden(LIVE_EXPECTED, snapshot, MODE)
@@ -910,8 +912,8 @@ describe('web e2e: plugin manager', () => {
     await rowSwitch.waitFor({ timeout: 10_000 })
     expect(await rowSwitch.getAttribute('aria-checked')).toBe('true')
     await rowSwitch.click()
-    await expect.poll(async () => (await homeFile('profiles', 'scaffold', 'cordis.patch.yml')).includes('fixture-row'), { timeout: 10_000 }).toBe(true)
-    await expect.poll(() => rowSwitch.getAttribute('aria-checked'), { timeout: 10_000 }).toBe('false')
+    await expect.poll(async () => (await homeFile('profiles', 'scaffold', 'cordis.patch.yml')).includes('fixture-row'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe(true)
+    await expect.poll(() => rowSwitch.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
     // A disabled entry keeps its disposed fiber; only an active one counts as mounted.
     await expect.poll(() => mounted()?.fiber?.state, { timeout: 20_000 }).not.toBe(2)
     await rowSwitch.click()
@@ -920,7 +922,7 @@ describe('web e2e: plugin manager', () => {
 
     await toggle.click()
     await expect.poll(() => mounted()?.fiber?.state, { timeout: 20_000 }).not.toBe(2)
-    await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('false')
+    await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
     expect(tripwire.pageErrors).toEqual([])
   }, 90_000)
 
@@ -959,8 +961,8 @@ describe('web e2e: startup-applied plugin management', () => {
       expect(mounted()?.fiber?.state).toBeUndefined()
       await toggle.click()
       // The selection is saved and the switch turns on, but nothing mounts before the next start; a toast says so.
-      await expect.poll(bundles).toEqual([...SCAFFOLD_BUNDLES, '@fixture/bundle'])
-      await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('true')
+      await expect.poll(bundles, { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toEqual([...SCAFFOLD_BUNDLES, '@fixture/bundle'])
+      await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
       await page.getByText('更改将在下次启动生效', { exact: true }).waitFor({ timeout: 10_000 })
       expect(mounted()?.fiber?.state).toBeUndefined()
       // The pack's page lists its rows from their declarations, with no live entry to switch.
@@ -970,8 +972,8 @@ describe('web e2e: startup-applied plugin management', () => {
       await panel.getByRole('button', { name: '返回插件列表' }).click()
 
       await toggle.click()
-      await expect.poll(bundles).toEqual(SCAFFOLD_BUNDLES)
-      await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('false')
+      await expect.poll(bundles, { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toEqual(SCAFFOLD_BUNDLES)
+      await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
       expect(tripwire.pageErrors).toEqual([])
     } finally {
       await browser?.close()

apps/web/tests/support.ts

@@ -41,6 +41,15 @@ export const ZH_BROWSER_LOCALE = 'zh-CN'
 /** Same-day anchor for seeded event times and the Asia/Shanghai browser clock. */
 export const WEB_FIXTURE_TIME = Date.parse('2026-01-15T12:00:00+08:00')
 
+/**
+ * Budget for a plugin-manager switch to republish its state after a click: the
+ * click starts a Host profile write, and the store's follow-up `load()` re-reads
+ * its inventory before the switch shows the new value. That chain measured
+ * 0.5-0.8 s on an idle host and 2.0-4.8 s across 32 samples in a 16-worker lane,
+ * so Vitest's 1-second `expect.poll` default expires first.
+ */
+export const PLUGIN_TOGGLE_SETTLE_MS = 10_000
+
 /** How often a pinned browser clock advances, in milliseconds. */
 const PINNED_CLOCK_STEP_MS = 250
 

apps/web/tests/worktree-opt-in.e2e.ts

@@ -10,7 +10,7 @@ import type {} from '@deepseek-ai/dsh-experimental-worktree'
 import { chromium } from 'playwright'
 import { expect, it, onTestFailed, onTestFinished } from 'vitest'
 import { launchWebScaffold, watchConsole } from './scaffold.ts'
-import { connectFreshWorkspace, newEnglishPage, saveFailureShot } from './support.ts'
+import { connectFreshWorkspace, newEnglishPage, PLUGIN_TOGGLE_SETTLE_MS, saveFailureShot } from './support.ts'
 
 const BUNDLE = '@deepseek-ai/dsh-experimental-tool-worktree'
 const exec = promisify(execFile)
@@ -80,8 +80,8 @@ it('persists the Git Worktrees switch and removes its capability while retaining
   expect(tools()).not.toContain('create_worktree')
 
   await toggle.click()
-  await expect.poll(() => rows().length).toBe(2)
-  await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('true')
+  await expect.poll(() => rows().length, { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe(2)
+  await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
   expect(await selectedBundles()).toContain(BUNDLE)
   expect(tools()).toContain('create_worktree')
   const created = await scaffold.ctx.worktrees.create(agent, { name: 'gui-opt-in' })
@@ -95,8 +95,8 @@ it('persists the Git Worktrees switch and removes its capability while retaining
   await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('true')
   expect(tools()).toContain('create_worktree')
   await toggle.click()
-  await expect.poll(() => rows().length).toBe(0)
-  await expect.poll(() => toggle.getAttribute('aria-checked')).toBe('false')
+  await expect.poll(() => rows().length, { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe(0)
+  await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
   expect(await selectedBundles()).not.toContain(BUNDLE)
   expect(scaffold.ctx.get('worktrees')).toBeUndefined()
   expect(tools()).not.toContain('create_worktree')