@neurocode-ai/app 1.18.8

package/e2e/performance/timeline/navigation-milestones.ts

@@ -0,0 +1,128 @@
+import type { Page } from "@playwright/test"
+
+export type NavigationMilestoneSample = {
+  observedAtMs: number
+  milestones: Record<string, boolean>
+}
+
+export function summarizeNavigationMilestones(samples: NavigationMilestoneSample[]) {
+  const names = Object.keys(samples[0]?.milestones ?? {})
+  const summarize = (matches: (sample: NavigationMilestoneSample) => boolean) => {
+    const first = samples.find(matches)
+    const stable = samples.findIndex(
+      (sample, index) =>
+        index + 2 < samples.length && matches(sample) && matches(samples[index + 1]!) && matches(samples[index + 2]!),
+    )
+    return {
+      firstObservedMs: first?.observedAtMs ?? null,
+      stableObservedMs: stable === -1 ? null : samples[stable + 2]!.observedAtMs,
+    }
+  }
+  return {
+    samples: samples.length,
+    milestones: Object.fromEntries(
+      names.map((name) => [name, summarize((sample) => sample.milestones[name] === true)]),
+    ),
+    all: summarize((sample) => names.every((name) => sample.milestones[name] === true)),
+  }
+}
+
+type NavigationMilestoneProbe = {
+  samples: NavigationMilestoneSample[]
+  stop: () => void
+}
+
+export async function measureNavigationMilestones(
+  page: Page,
+  input: {
+    triggerSelector: string
+    milestones: Record<string, { selector: string; visible?: boolean }>
+    navigate: () => Promise<void>
+  },
+) {
+  await page.evaluate(
+    ({ triggerSelector, milestones }) => {
+      const samples: NavigationMilestoneSample[] = []
+      const streaks = new Map<string, number>()
+      const marked = new Set<string>()
+      let started: number | undefined
+      let running = true
+      const visible = (selector: string) =>
+        [...document.querySelectorAll<HTMLElement>(selector)].some((element) => {
+          const rect = element.getBoundingClientRect()
+          const style = getComputedStyle(element)
+          return rect.width > 0 && rect.height > 0 && style.visibility !== "hidden" && style.display !== "none"
+        })
+      const sample = () => {
+        if (!running || started === undefined) return
+        requestAnimationFrame(() => {
+          setTimeout(() => {
+            if (!running || started === undefined) return
+            const current = Object.fromEntries(
+              Object.entries(milestones).map(([name, milestone]) => [
+                name,
+                milestone.visible === false ? !document.querySelector(milestone.selector) : visible(milestone.selector),
+              ]),
+            )
+            samples.push({
+              observedAtMs: performance.now() - started,
+              milestones: current,
+            })
+            Object.entries(current).forEach(([name, value]) => {
+              if (!value) {
+                streaks.set(name, 0)
+                return
+              }
+              if (!marked.has(`${name}.first`)) {
+                performance.mark(`neurocode.navigation.${name}.first`)
+                marked.add(`${name}.first`)
+              }
+              const streak = (streaks.get(name) ?? 0) + 1
+              streaks.set(name, streak)
+              if (streak === 3) performance.mark(`neurocode.navigation.${name}.stable`)
+            })
+            const all = Object.values(current).every(Boolean)
+            const allStreak = all ? (streaks.get("all") ?? 0) + 1 : 0
+            streaks.set("all", allStreak)
+            if (all && !marked.has("all.first")) {
+              performance.mark("neurocode.navigation.all.first")
+              marked.add("all.first")
+            }
+            if (allStreak === 3) performance.mark("neurocode.navigation.all.stable")
+            sample()
+          }, 0)
+        })
+      }
+      document.addEventListener(
+        "click",
+        (event) => {
+          if (!(event.target instanceof Element) || !event.target.closest(triggerSelector)) return
+          started = performance.now()
+          performance.mark("neurocode.navigation.click")
+          sample()
+        },
+        { capture: true, once: true },
+      )
+      ;(window as Window & { __navigationMilestones?: NavigationMilestoneProbe }).__navigationMilestones = {
+        samples,
+        stop: () => {
+          running = false
+        },
+      }
+    },
+    { triggerSelector: input.triggerSelector, milestones: input.milestones },
+  )
+  await input.navigate()
+  await page.waitForFunction(() => {
+    const samples = (window as Window & { __navigationMilestones?: NavigationMilestoneProbe }).__navigationMilestones
+      ?.samples
+    if (!samples || samples.length < 3) return false
+    return samples.slice(-3).every((sample) => Object.values(sample.milestones).every(Boolean))
+  })
+  const samples = await page.evaluate(() => {
+    const probe = (window as Window & { __navigationMilestones?: NavigationMilestoneProbe }).__navigationMilestones!
+    probe.stop()
+    return probe.samples
+  })
+  return { summary: summarizeNavigationMilestones(samples), samples }
+}

package/e2e/performance/timeline/review-pane-scaling-benchmark.spec.ts

@@ -0,0 +1,312 @@
+import type { Page } from "@playwright/test"
+import { benchmark, expect } from "../benchmark"
+import { setupTimelineBenchmark } from "./session-timeline-benchmark.fixture"
+
+const changedLinesPerFile = 100
+const linesPerSide = changedLinesPerFile / 2
+const fileCounts = [1, 10, 100, 1_000, 10_000]
+const filesPerDirectory = 100
+const readyFrames = 3
+const completionTimeoutMs = Number(process.env.REVIEW_PANE_COMPLETION_TIMEOUT_MS ?? 900_000)
+
+type ReviewPaneScalingSample = {
+  observedAtMs: number
+  logicalRows: number
+  treeRows: number
+  fileRows: number
+  diffLines: number
+  header: string
+  ready: boolean
+}
+
+type ReviewPaneScalingProbe = {
+  startedAt?: number
+  firstTreeRowMs?: number
+  logicalTreeReadyMs?: number
+  firstDiffRenderMs?: number
+  stableReadyMs?: number
+  samples: ReviewPaneScalingSample[]
+  frameTimesMs: number[]
+  longTasks: { startTime: number; duration: number }[]
+  stop: () => void
+}
+
+benchmark.describe("performance: review pane scaling", () => {
+  for (const fileCount of fileCounts) {
+    const changedLines = fileCount * changedLinesPerFile
+
+    benchmark(
+      `${changedLines} changed lines across ${fileCount} ${fileCount === 1 ? "file" : "files"}`,
+      async ({ page, report }) => {
+        benchmark.setTimeout(1_200_000)
+        await page.emulateMedia({ reducedMotion: "reduce" })
+
+        const patchByteLimit = Number(process.env.REVIEW_PANE_PATCH_BYTE_LIMIT ?? Number.POSITIVE_INFINITY)
+        if (Number.isNaN(patchByteLimit) || patchByteLimit < 0)
+          throw new Error(`Invalid REVIEW_PANE_PATCH_BYTE_LIMIT: ${process.env.REVIEW_PANE_PATCH_BYTE_LIMIT}`)
+        const responseBody = JSON.stringify(createScalingDiffs(fileCount, patchByteLimit))
+        await setupTimelineBenchmark(page, {
+          historyTurns: 0,
+          eventBatch: 1,
+          newLayoutDesigns: true,
+        })
+        await page.route("**/vcs/diff**", (route) =>
+          route.fulfill({
+            status: 200,
+            contentType: "application/json",
+            headers: { "access-control-allow-origin": "*" },
+            body: responseBody,
+          }),
+        )
+
+        const expectedRows = fileCount + 2 + Math.ceil(fileCount / filesPerDirectory)
+        const metrics = await measureReviewPaneLoad(page, {
+          expectedFile: reviewFile(0),
+          expectedRows,
+        })
+        const search = await measureBroadReviewSearch(page, fileCount)
+
+        expect(metrics.logicalRows).toBe(expectedRows)
+        expect(metrics.fileRows).toBeGreaterThan(0)
+        expect(metrics.treeRows).toBeGreaterThan(0)
+        expect(metrics.diffLines).toBeGreaterThan(0)
+        expect(search.logicalRows).toBe(fileCount)
+        expect(search.renderedRows).toBeGreaterThan(0)
+        report(
+          { ...metrics, search },
+          {
+            fileCount,
+            changedLinesPerFile,
+            changedLines,
+            additions: changedLines / 2,
+            deletions: changedLines / 2,
+            patchLines: changedLines,
+            patchByteLimit: Number.isFinite(patchByteLimit) ? patchByteLimit : null,
+            payloadBytes: new TextEncoder().encode(responseBody).byteLength,
+            expectedRows,
+          },
+        )
+      },
+    )
+  }
+})
+
+async function measureBroadReviewSearch(page: Page, expectedRows: number) {
+  const filter = page.getByRole("searchbox", { name: "Filter files" })
+  await filter.evaluate((element) => {
+    element.addEventListener(
+      "input",
+      () => {
+        ;(window as Window & { __reviewSearchStartedAt?: number }).__reviewSearchStartedAt = performance.now()
+      },
+      { once: true, capture: true },
+    )
+  })
+  await filter.fill("file-")
+
+  return page.evaluate((expectedRows) => {
+    const startedAt = (window as Window & { __reviewSearchStartedAt?: number }).__reviewSearchStartedAt!
+    return new Promise<{ stableMs: number; logicalRows: number; renderedRows: number }>((resolve) => {
+      let previous = -1
+      let streak = 0
+      const sample = () => {
+        const tree = document.querySelector<HTMLElement>('#review-panel [data-component="file-tree-v2"]')
+        const rows = [...document.querySelectorAll<HTMLElement>('#review-panel [data-slot="file-tree-v2-row"]')]
+        const logicalRows = Number(tree?.dataset.totalRows ?? rows.length)
+        const ready =
+          logicalRows === expectedRows && rows.length > 0 && rows.every((row) => row.textContent?.includes("file-"))
+        streak = ready && rows.length === previous ? streak + 1 : ready ? 1 : 0
+        previous = rows.length
+        if (streak >= 3) {
+          resolve({ stableMs: performance.now() - startedAt, logicalRows, renderedRows: rows.length })
+          return
+        }
+        requestAnimationFrame(sample)
+      }
+      requestAnimationFrame(sample)
+    })
+  }, expectedRows)
+}
+
+function createScalingDiffs(fileCount: number, patchByteLimit: number) {
+  const changes = Array.from({ length: linesPerSide }, (_, index) => {
+    const line = String(index).padStart(3, "0")
+    return `-export const value_${line} = "before"\n+export const value_${line} = "after"`
+  }).join("\n")
+  let patchBytes = 0
+  let capped = false
+
+  return Array.from({ length: fileCount }, (_, index) => {
+    const file = reviewFile(index)
+    const fullPatch = [
+      `diff --git a/${file} b/${file}`,
+      `--- a/${file}`,
+      `+++ b/${file}`,
+      `@@ -1,${linesPerSide} +1,${linesPerSide} @@`,
+      changes,
+    ].join("\n")
+    if (index === 0 && fullPatch.length > patchByteLimit)
+      throw new Error(`REVIEW_PANE_PATCH_BYTE_LIMIT must include the active patch (${fullPatch.length} bytes)`)
+    const patch = !capped && patchBytes + fullPatch.length <= patchByteLimit ? fullPatch : emptyReviewPatch(file)
+    if (patch === fullPatch) patchBytes += fullPatch.length
+    else capped = true
+    return {
+      file,
+      patch,
+      additions: linesPerSide,
+      deletions: linesPerSide,
+      status: "modified" as const,
+    }
+  })
+}
+
+function emptyReviewPatch(file: string) {
+  return [`diff --git a/${file} b/${file}`, `--- a/${file}`, `+++ b/${file}`].join("\n")
+}
+
+function reviewFile(index: number) {
+  return `src/review/d${String(Math.floor(index / filesPerDirectory)).padStart(5, "0")}/file-${String(index).padStart(5, "0")}.ts`
+}
+
+async function measureReviewPaneLoad(page: Page, input: { expectedFile: string; expectedRows: number }) {
+  const toggle = page.getByRole("button", { name: "Toggle review" })
+  await expect(toggle).toBeVisible()
+  await toggle.evaluate((element) => element.setAttribute("data-review-pane-scaling-toggle", ""))
+  await installReviewPaneScalingProbe(page, input)
+  await toggle.click()
+  await page.waitForFunction(
+    () =>
+      (window as Window & { __reviewPaneScalingProbe?: ReviewPaneScalingProbe }).__reviewPaneScalingProbe
+        ?.stableReadyMs !== undefined,
+    undefined,
+    { timeout: completionTimeoutMs },
+  )
+
+  return page.evaluate(() => {
+    const probe = (window as Window & { __reviewPaneScalingProbe?: ReviewPaneScalingProbe }).__reviewPaneScalingProbe!
+    probe.stop()
+    const startedAt = probe.startedAt!
+    const final = probe.samples.at(-1)!
+    const resources = performance
+      .getEntriesByType("resource")
+      .filter((entry) => entry.name.includes("/vcs/diff")) as PerformanceResourceTiming[]
+    const resource = resources.at(-1)
+    const longTasks = probe.longTasks.filter(
+      (entry) => entry.startTime >= startedAt && entry.startTime <= startedAt + probe.stableReadyMs!,
+    )
+    const frameGaps = probe.frameTimesMs.map((time, index) => time - (probe.frameTimesMs[index - 1] ?? 0))
+
+    return {
+      firstTreeRowMs: probe.firstTreeRowMs ?? null,
+      logicalTreeReadyMs: probe.logicalTreeReadyMs ?? null,
+      firstDiffRenderMs: probe.firstDiffRenderMs ?? null,
+      stableReadyMs: probe.stableReadyMs ?? null,
+      responseStartMs: resource ? resource.responseStart - startedAt : null,
+      responseEndMs: resource ? resource.responseEnd - startedAt : null,
+      responseToStableMs: resource ? probe.stableReadyMs! - (resource.responseEnd - startedAt) : null,
+      treeRows: final.treeRows,
+      logicalRows: final.logicalRows,
+      fileRows: final.fileRows,
+      diffLines: final.diffLines,
+      samples: probe.samples.length,
+      maxFrameGapMs: Math.max(0, ...frameGaps),
+      longTaskCount: longTasks.length,
+      longTaskTotalMs: longTasks.reduce((sum, entry) => sum + entry.duration, 0),
+      maxLongTaskMs: Math.max(0, ...longTasks.map((entry) => entry.duration)),
+    }
+  })
+}
+
+async function installReviewPaneScalingProbe(page: Page, input: { expectedFile: string; expectedRows: number }) {
+  await page.evaluate(
+    ({ expectedFile, expectedRows, stableFrames }) => {
+      let running = true
+      let readyStreak = 0
+      const basename = expectedFile.split("/").at(-1)!
+      const longTaskObserver = PerformanceObserver.supportedEntryTypes.includes("longtask")
+        ? new PerformanceObserver((list) => {
+            probe.longTasks.push(
+              ...list.getEntries().map((entry) => ({ startTime: entry.startTime, duration: entry.duration })),
+            )
+          })
+        : undefined
+      const probe: ReviewPaneScalingProbe = {
+        samples: [],
+        frameTimesMs: [],
+        longTasks: [],
+        stop: () => {
+          running = false
+          longTaskObserver?.disconnect()
+        },
+      }
+
+      const sample = (time: number) => {
+        if (!running || probe.startedAt === undefined) return
+        const panel = document.querySelector<HTMLElement>("#review-panel")
+        const tree = panel?.querySelector<HTMLElement>('[data-component="file-tree-v2"]')
+        const rows = panel?.querySelectorAll('[data-slot="file-tree-v2-row"]') ?? []
+        const fileRows = panel?.querySelectorAll('button[data-slot="file-tree-v2-row"]') ?? []
+        const header =
+          panel?.querySelector<HTMLElement>('[data-slot="session-review-v2-file-header"]')?.textContent?.trim() ?? ""
+        const viewers = panel
+          ? [...panel.querySelectorAll<HTMLElement>('[data-component="file"][data-mode="diff"]')]
+          : []
+        const diffLines = viewers.reduce(
+          (sum, viewer) =>
+            sum + (viewer.querySelector("diffs-container")?.shadowRoot?.querySelectorAll("[data-line]").length ?? 0),
+          0,
+        )
+        const observedAtMs = time - probe.startedAt
+        const logicalRows = Number(tree?.dataset.totalRows ?? rows.length)
+        const ready =
+          logicalRows === expectedRows &&
+          fileRows.length > 0 &&
+          header.includes(basename) &&
+          viewers.length === 1 &&
+          diffLines > 0
+        const previous = probe.samples.at(-1)
+        const stable =
+          ready &&
+          previous?.ready === true &&
+          previous.logicalRows === logicalRows &&
+          previous.treeRows === rows.length &&
+          previous.fileRows === fileRows.length &&
+          previous.diffLines === diffLines &&
+          previous.header === header
+
+        probe.frameTimesMs.push(observedAtMs)
+        probe.samples.push({
+          observedAtMs,
+          logicalRows,
+          treeRows: rows.length,
+          fileRows: fileRows.length,
+          diffLines,
+          header,
+          ready,
+        })
+        if (probe.firstTreeRowMs === undefined && rows.length > 0) probe.firstTreeRowMs = observedAtMs
+        if (probe.logicalTreeReadyMs === undefined && logicalRows === expectedRows)
+          probe.logicalTreeReadyMs = observedAtMs
+        if (probe.firstDiffRenderMs === undefined && diffLines > 0) probe.firstDiffRenderMs = observedAtMs
+        readyStreak = !ready ? 0 : stable ? readyStreak + 1 : 1
+        if (readyStreak === stableFrames) probe.stableReadyMs = observedAtMs
+        if (probe.stableReadyMs === undefined) requestAnimationFrame(sample)
+      }
+
+      longTaskObserver?.observe({ type: "longtask", buffered: true })
+      document.addEventListener(
+        "click",
+        (event) => {
+          const toggle = event.target instanceof Element ? event.target.closest("button") : undefined
+          if (!toggle?.hasAttribute("data-review-pane-scaling-toggle")) return
+          probe.startedAt = performance.now()
+          performance.mark("neurocode.review-pane-scaling.click")
+          requestAnimationFrame(sample)
+        },
+        { capture: true, once: true },
+      )
+      ;(window as Window & { __reviewPaneScalingProbe?: ReviewPaneScalingProbe }).__reviewPaneScalingProbe = probe
+    },
+    { ...input, stableFrames: readyFrames },
+  )
+}

package/e2e/performance/timeline/session-parent-hydration-benchmark.spec.ts

@@ -0,0 +1,172 @@
+import type { Page } from "@playwright/test"
+import { expectSessionTitle } from "../../utils/waits"
+import { mockOpenCodeServer } from "../../utils/mock-server"
+import { benchmark, expect, withBenchmarkPage } from "../benchmark"
+import { fixture } from "./session-timeline-stress.fixture"
+import { installStressSessionTabs, stressSessionHref } from "./timeline-test-helpers"
+import { measureSessionSwitch, waitForStableTimeline } from "./session-tab-switch-probe"
+
+type ParentHydrationBenchmarkMode = "natural" | "candidate"
+
+const mode = process.env.SESSION_PARENT_HYDRATION_BENCHMARK_MODE ?? "natural"
+if (mode !== "natural" && mode !== "candidate") throw new Error(`Unknown parent hydration benchmark mode: ${mode}`)
+const userID = "msg_parent_hydration_user"
+const user = {
+  ...fixture.messages[fixture.targetID][0]!,
+  info: { ...fixture.messages[fixture.targetID][0]!.info, id: userID, time: { created: 1700001000000 } },
+  parts: fixture.messages[fixture.targetID][0]!.parts.map((part, index) => ({
+    ...part,
+    id: `prt_parent_hydration_user_${index}`,
+    messageID: userID,
+  })),
+}
+const assistantSeed = fixture.messages[fixture.targetID][3]!
+const assistants = Array.from({ length: 14 }, (_, index) => {
+  const messageID = `msg_parent_hydration_${String(index).padStart(2, "0")}`
+  return {
+    ...assistantSeed,
+    info: {
+      ...assistantSeed.info,
+      id: messageID,
+      parentID: userID,
+      time: { created: 1700001001000 + index * 1_000, completed: 1700001001500 + index * 1_000 },
+    },
+    parts: assistantSeed.parts.map((part, partIndex) => ({
+      ...part,
+      id: `prt_parent_hydration_${String(index).padStart(2, "0")}_${partIndex}`,
+      messageID,
+    })),
+  }
+})
+const messages = [user, ...assistants]
+const target = fixture.sessions.find((session) => session.id === fixture.targetID)!
+const lastID = userID
+const lastAssistant = assistants.at(-1)!
+const lastPart = lastAssistant.parts.at(-1)!
+const lastPartID =
+  lastPart.type === "tool"
+    ? lastPart.id
+    : `${lastAssistant.info.id}:${lastPart.type}:${lastAssistant.parts.filter((part) => part.type === lastPart.type).length - 1}`
+
+benchmark("hydrates an orphaned latest turn after a cold session click", async ({ browser, report }, testInfo) => {
+  benchmark.setTimeout(180_000)
+  const results = [] as Awaited<ReturnType<typeof trial>>[]
+  for (let run = 0; run < 5; run++) {
+    results.push(
+      await withBenchmarkPage(
+        browser,
+        `session-parent-hydration-${mode}-${run}`,
+        (page) => trial(page, mode),
+        testInfo,
+      ),
+    )
+  }
+  const timing = results.map((result) => result.metrics.firstCorrectObservedMs!).sort((a, b) => a - b)
+  report(
+    {
+      results: results.map((result) => ({ ...result.metrics, historyGateCount: result.historyGateCount })),
+      summary: {
+        firstCorrectObservedMs: { min: timing[0], median: timing[2], max: timing.at(-1) },
+        blankSamples: results.map((result) => result.metrics.blankSamples),
+        requestCounts: {
+          list: results.map((result) => result.requestCounts.list),
+          parent: results.map((result) => result.requestCounts.parent),
+        },
+        historyGateCount: results.map((result) => result.historyGateCount),
+      },
+    },
+    { mode },
+  )
+})
+
+async function trial(page: Page, mode: ParentHydrationBenchmarkMode) {
+  const requests: { type: "list" | "parent"; before?: string }[] = []
+  const history = mode === "candidate" ? Promise.withResolvers<void>() : undefined
+  let historyGates = 0
+  await mockOpenCodeServer(page, {
+    sessions: fixture.sessions.filter((session) => session.id === fixture.sourceID),
+    provider: fixture.provider,
+    directory: fixture.directory,
+    project: fixture.project,
+    messageDelay: 50,
+    onMessages: (request) => {
+      if (request.sessionID === fixture.targetID && request.phase === "start")
+        requests.push({ type: "list", before: request.before })
+    },
+    beforeMessagesResponse: (request) => {
+      if (mode !== "candidate" || request.sessionID !== fixture.targetID || !request.before) return Promise.resolve()
+      historyGates++
+      return history!.promise
+    },
+    onMessage: (request) => {
+      if (request.sessionID === fixture.targetID && request.messageID === userID) requests.push({ type: "parent" })
+    },
+    message: (sessionID, messageID) => {
+      if (sessionID !== fixture.targetID || messageID !== userID) return
+      return user
+    },
+    pageMessages: (sessionID, limit, before) => {
+      const items = sessionID === fixture.targetID ? messages : fixture.messages[fixture.sourceID]
+      const end = before ? items.findIndex((message) => message.info.id === before) : items.length
+      const start = Math.max(0, end - limit)
+      return { items: items.slice(start, end), cursor: start > 0 ? items[start]!.info.id : undefined }
+    },
+  })
+  await page.route(`**/session/${fixture.targetID}`, (route) => {
+    const current = new URL(route.request().url()).pathname.startsWith("/api/")
+    return route.fulfill({
+      status: 200,
+      contentType: "application/json",
+      body: JSON.stringify(
+        current
+          ? {
+              data: {
+                ...target,
+                cost: 0,
+                tokens: { input: 0, output: 0, reasoning: 0, cache: { read: 0, write: 0 } },
+                location: { directory: target.directory },
+              },
+            }
+          : target,
+      ),
+    })
+  })
+  await installStressSessionTabs(page, { sessionIDs: [fixture.sourceID] })
+  await page.goto(stressSessionHref(fixture.sourceID))
+  await expectSessionTitle(page, fixture.expected.sourceTitle)
+  await waitForStableTimeline(page, fixture.expected.sourceMessageIDs.at(-1)!)
+
+  const href = stressSessionHref(fixture.targetID)
+  await page.evaluate(
+    ({ href, title }) => {
+      const link = document.createElement("a")
+      link.id = "parent-hydration-target"
+      link.href = href
+      link.textContent = title
+      document.body.append(link)
+    },
+    { href, title: target.title },
+  )
+  const metrics = await measureSessionSwitch(page, {
+    destinationIDs: messages.map((message) => message.info.id),
+    sourceIDs: fixture.messages[fixture.sourceID].map((message) => message.info.id),
+    lastID,
+    requiredPartID: lastPartID,
+    requireBottomAnchor: false,
+    href,
+    switch: async () => {
+      await page.locator("#parent-hydration-target").click()
+      await expectSessionTitle(page, target.title)
+    },
+  }).finally(() => history?.resolve())
+  expect(metrics.firstCorrectObservedMs).not.toBeNull()
+  const requestCounts = {
+    list: requests.filter((request) => request.type === "list").length,
+    parent: requests.filter((request) => request.type === "parent").length,
+  }
+  if (mode === "candidate") {
+    expect(requestCounts.parent).toBe(0)
+    expect(historyGates).toBe(0)
+  }
+  return { metrics, requestCounts, historyGateCount: historyGates }
+}

package/e2e/performance/timeline/session-tab-flash.spec.ts

@@ -0,0 +1,67 @@
+import { benchmark, expect } from "../benchmark"
+import { expectSessionTitle } from "../../utils/waits"
+import { fixture } from "./session-timeline-stress.fixture"
+import {
+  collectCachedRepaintTrace,
+  compressCachedRepaintTrace,
+  installCachedRepaintProbe,
+  waitForCachedRepaintWindow,
+} from "./session-tab-repaint-probe"
+import { waitForStableTimeline } from "./session-tab-switch-probe"
+import {
+  installStressSessionTabs,
+  installTimelineSettings,
+  mockStressTimeline,
+  stressSessionHref,
+} from "./timeline-test-helpers"
+
+benchmark("samples cached session repaint after the click", async ({ page, report }) => {
+  benchmark.setTimeout(120_000)
+  await mockStressTimeline(page)
+  await installStressSessionTabs(page)
+  await installTimelineSettings(page)
+  await page.goto(stressSessionHref(fixture.targetID))
+  await expectSessionTitle(page, fixture.expected.targetTitle)
+  await waitForStableTimeline(page, fixture.expected.targetMessageIDs.at(-1)!)
+  await page
+    .locator(`[data-slot="titlebar-tabs"] a[href="${stressSessionHref(fixture.sourceID)}"]`)
+    .first()
+    .click()
+  await expectSessionTitle(page, fixture.expected.sourceTitle)
+  await waitForStableTimeline(page, fixture.expected.sourceMessageIDs.at(-1)!)
+
+  await installCachedRepaintProbe(page, {
+    targetHref: stressSessionHref(fixture.targetID),
+    destination: fixture.messages[fixture.targetID].map((message) => message.info.id),
+    source: fixture.messages[fixture.sourceID].map((message) => message.info.id),
+    last: fixture.expected.targetMessageIDs.at(-1)!,
+    windowMs: 1_000,
+  })
+
+  await page
+    .locator(`[data-slot="titlebar-tabs"] a[href="${stressSessionHref(fixture.targetID)}"]`)
+    .first()
+    .click()
+  await Promise.all([expectSessionTitle(page, fixture.expected.targetTitle), waitForCachedRepaintWindow(page, 1_000)])
+  const result = await collectCachedRepaintTrace(page)
+  report(compressCachedRepaintTrace(result))
+  expect(result.samples.length).toBeGreaterThan(0)
+})
+
+benchmark("prefetches every open session tab", async ({ page, report }) => {
+  const prefetched = new Set<string>()
+  await mockStressTimeline(page, {
+    onMessages: (input) => {
+      if (!input.before && input.phase === "start") prefetched.add(input.sessionID)
+    },
+  })
+  await installStressSessionTabs(page, {
+    sessionIDs: [fixture.sourceID, fixture.targetID, fixture.childID],
+  })
+  await installTimelineSettings(page)
+  await page.goto(stressSessionHref(fixture.sourceID))
+  await expectSessionTitle(page, fixture.expected.sourceTitle)
+
+  await expect.poll(() => prefetched.has(fixture.childID)).toBe(true)
+  report({ prefetched: [...prefetched] })
+})