feat(firefox): roll to r1541

[microsoft-playwright-automation]

Browser build: https://github.com/microsoft/playwright-browsers/commit/f81945fd6dce01624d8286736d47c4d014ab3a50. Build has full platform coverage - check that all tests pass there.

[github-actions]

Test results for "tests 1"

5 failed
❌ [firefox-library] › library/har.spec.ts:629 › should have connection details @firefox-ubuntu-22.04-node20
❌ [firefox-library] › library/har.spec.ts:639 › should have security details @firefox-ubuntu-22.04-node20
❌ [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects @firefox-ubuntu-22.04-node20
❌ [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests @firefox-ubuntu-22.04-node20
❌ [firefox-library] › library/har.spec.ts:699 › should return server address directly from response @firefox-ubuntu-22.04-node20

7 flaky

⚠️ [chromium-library] › library/chromium/chromium.spec.ts:301 › should report intercepted service worker requests in HAR `@frozen-time-library-chromium-linux`
⚠️ [chromium-library] › library/video.spec.ts:495 › screencast › should capture static page in persistent context Radoslav Kirilov (@smoke) `@realtime-time-library-chromium-linux`
⚠️ [chromium-library] › library/video.spec.ts:736 › screencast › should work with video+trace `@chromium-ubuntu-22.04-node20`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-library] › library/heap.spec.ts:223 › should not leak workers `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-page] › page/page-goto.spec.ts:90 › should work with Cross-Origin-Opener-Policy `@firefox-ubuntu-22.04-node20`

51241 passed, 1240 skipped

---

Merge workflow run.

[github-actions]

🔴 The Firefox roll breaks 5 HAR serverIPAddress tests

Hi, I'm the Playwright bot — I took a first look at the failing CI.

This is a real regression from the roll, not flake. Firefox r1541 now reports the HAR serverIPAddress as the bare ::1 instead of the bracketed [::1], so library/har.spec.ts's toMatch(/^127\.0\.0\.1|\[::1\]/) no longer matches. All 5 failures share that exact signature, and they reproduce on every retry — 10/10 attempts red on this PR, 0 green.

Details

Caused by this PR

• [firefox-library] › library/har.spec.ts:629 › should have connection details

• [firefox-library] › library/har.spec.ts:639 › should have security details

• [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects

• [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests

• [firefox-library] › library/har.spec.ts:699 › should return server address directly from response

  All five fail identically:

  Expected pattern: /^127\.0\.0\.1|\[::1\]/
  Received string:  "::1"
  

  This is a Firefox-only browser roll (roll-into-pw-firefox/1541), and serverIPAddress is populated by the rolled juggler's networking layer — squarely in the diff's blast radius. The new build emits the loopback IPv6 address unbracketed. The fix is either to restore the bracketed [::1] form in the juggler's HAR reporting, or to relax the test to accept a bare ::1. Chromium and WebKit are unaffected, which is exactly why only the firefox-library project went red.

  Test-results DB confirms it's the roll, not a flake: across the recent window these tests failed on one branch only — this PR (10 failed / 0 passed). On every other PR and branch that ran them they're green (0 failures over hundreds of runs), so the flake hard-rule is not met — there's no run outside this PR where they misbehave.

Pre-existing flake / infra

• The time test runner - frozen job also shows red, but its blob report lists no failing tests — only the 7 flaky tests below, all rescued on retry. That's the frozen-time job's usual flake noise, unrelated to the roll.

• Reported flaky (passed on retry), not caused by this PR: chromium.spec.ts:301, video.spec.ts:495, video.spec.ts:736, browsercontext-cookies-third-party.spec.ts:257 & :470, heap.spec.ts:223, page-goto.spec.ts:90.

_{Triaged by the Playwright bot - agent run}

[github-actions]

Test results for "MCP"

4 failed
❌ [chromium] › mcp/clipboard.spec.ts:19 › clipboard write without permission dialog @mcp-macos-latest-chromium
❌ [chromium] › mcp/clipboard.spec.ts:19 › clipboard write without permission dialog @mcp-windows-latest-chromium
❌ [chromium] › mcp/clipboard.spec.ts:19 › clipboard write without permission dialog @mcp-ubuntu-latest-chromium
❌ [firefox] › mcp/cli-core.spec.ts:394 › --raw on command without output @mcp-windows-latest-firefox

8255 passed, 1361 skipped

---

Merge workflow run.

[github-actions]

🔴 The Firefox roll breaks 5 HAR serverIPAddress tests

Hi, I'm the Playwright bot — I took a first look at the failing CI.

One real regression: Firefox r1541 reports the HAR serverIPAddress for the loopback as the bare ::1 instead of the bracketed [::1], so library/har.spec.ts's /^127\.0\.0\.1|\[::1\]/ no longer matches. All 5 firefox-library failures share that signature and fail on every attempt (2/2 red on this PR, 0 red across ~hundreds of other runs). The MCP clipboard and cli-core reds are pre-existing flakes a firefox roll can't be responsible for.

Details

The diff is firefox-only — browser_patches/firefox/juggler/* plus browsers.json. serverIPAddress is populated by juggler's networking layer, squarely in the blast radius, which is why only firefox-library went red while Chromium and WebKit stayed green.

Caused by this PR

• [firefox-library] › library/har.spec.ts:629 › should have connection details

• [firefox-library] › library/har.spec.ts:639 › should have security details

• [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects

• [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests

• [firefox-library] › library/har.spec.ts:699 › should return server address directly from response

  All five fail identically:

  Expected pattern: /^127\.0\.0\.1|\[::1\]/
  Received string:  "::1"
  

  Test-results DB confirms the roll, not a flake: each failed 2/2 on this PR and 0 times across 356–983 other runs on unrelated SHAs. Fix by restoring the bracketed [::1] form in juggler's HAR reporting, or relaxing the test to accept a bare ::1.

Pre-existing flake / infra

• [chromium] › mcp/clipboard.spec.ts:19 › clipboard write without permission dialog (macOS, Windows, Ubuntu) — the clipboard read returns "" instead of "Hello from Playwright!". This is a Chromium-only test (Firefox is skipped), so a Firefox-only roll can't cause it. It flakes across the DB independent of this PR — 24 failures over 5824 other runs, including on main. The three OSes failing together points at a clipboard-access hiccup in that one MCP run.

• [firefox] › mcp/cli-core.spec.ts:394 › --raw on command without output (Windows only) — TimeoutError: Timeout 5000ms exceeded waiting for aria-ref=e2. Firefox on macOS and Ubuntu passed in the same run; only mcp-windows-latest-firefox failed. That exact bot has a standing flake history — 4 failures over 262 runs on main, plus one-off reds on PRs test: roll stable-test-runner to 1.63.0-alpha-2026-08-10 #42186, chore(tracing): reference trace and har blobs via relative file paths #42192, chore: remove experimental component testing #42168 and [DO NOT LAND] roll ffmpeg to 1012 #42067. Meets the flake bar; this is one more instance, not the roll.

• The 7 tests reported flaky (rescued on retry) — chromium.spec.ts:301, video.spec.ts:495/:736, browsercontext-cookies-third-party.spec.ts:257/:470, heap.spec.ts:223, page-goto.spec.ts:90 — are the usual noise, not caused by the roll.

_{Triaged by the Playwright bot - agent run}

[github-actions]

Test results for "tests 2"

25 failed
❌ [playwright-test] › playwright.artifacts.spec.ts:127 › should work with screenshot: on @frozen-time-runner-chromium-linux
❌ [playwright-test] › reporter-html.spec.ts:307 › created › should not include image diff with non-images @frozen-time-runner-chromium-linux
❌ [firefox-library] › library/har.spec.ts:629 › should have connection details @tracing-firefox
❌ [firefox-library] › library/har.spec.ts:639 › should have security details @tracing-firefox
❌ [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects @tracing-firefox
❌ [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests @tracing-firefox
❌ [firefox-library] › library/har.spec.ts:699 › should return server address directly from response @tracing-firefox
❌ [firefox-library] › library/har.spec.ts:629 › should have connection details @firefox-macos-15-large
❌ [firefox-library] › library/har.spec.ts:639 › should have security details @firefox-macos-15-large
❌ [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects @firefox-macos-15-large
❌ [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests @firefox-macos-15-large
❌ [firefox-library] › library/har.spec.ts:699 › should return server address directly from response @firefox-macos-15-large
❌ [firefox-library] › library/har.spec.ts:629 › should have connection details @firefox-windows-latest
❌ [firefox-library] › library/har.spec.ts:639 › should have security details @firefox-windows-latest
❌ [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects @firefox-windows-latest
❌ [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests @firefox-windows-latest
❌ [firefox-library] › library/har.spec.ts:699 › should return server address directly from response @firefox-windows-latest
❌ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies @firefox-macos-15-xlarge
❌ [firefox-library] › library/har.spec.ts:629 › should have connection details @firefox-macos-15-xlarge
❌ [firefox-library] › library/har.spec.ts:639 › should have security details @firefox-macos-15-xlarge
❌ [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects @firefox-macos-15-xlarge
❌ [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests @firefox-macos-15-xlarge
❌ [firefox-library] › library/har.spec.ts:699 › should return server address directly from response @firefox-macos-15-xlarge
❌ [android-native] › android/device.spec.ts:56 › androidDevice.fill
❌ [android-native] › android/webview.spec.ts:44 › should navigate page externally

23 flaky

⚠️ [chromium-library] › library/chromium/chromium.spec.ts:436 › should produce network events, routing, and annotations for Service Worker (advanced) `@tracing-chromium`
⚠️ [chromium-library] › library/chromium/launcher.spec.ts:37 › should not create pages automatically `@msedge-dev-windows-latest`
⚠️ [chromium-library] › library/inspector/cli-codegen-2.spec.ts:202 › cli codegen › should upload a file via hidden input triggered by button `@chrome-ubuntu-22.04`
⚠️ [chromium-library] › library/chromium/chromium.spec.ts:213 › should intercept service worker requests (main and within) `@chromium-macos-15-large`
⚠️ [chromium-library] › library/chromium/chromium.spec.ts:373 › should produce network events, routing, and annotations for Service Worker `@chromium-macos-15-large`
⚠️ [chromium-library] › library/chromium/chromium.spec.ts:436 › should produce network events, routing, and annotations for Service Worker (advanced) `@chromium-macos-15-large`
⚠️ [chromium-library] › library/video.spec.ts:390 › screencast › should scale frames down to the requested size `@chromium-macos-15-large`
⚠️ [chromium-library] › library/chromium/oopif.spec.ts:179 › should take screenshot `@chromium-macos-14-xlarge`
⚠️ [chromium-library] › library/video.spec.ts:736 › screencast › should work with video+trace `@chromium-macos-15-xlarge`
⚠️ [chromium-library] › library/beforeunload.spec.ts:130 › should support dismissing the dialog multiple times `@chrome-macos-latest`
⚠️ [chromium-library] › library/selector-generator.spec.ts:343 › selector generator › should trim long text `@chrome-macos-latest`
⚠️ [chromium-library] › library/selector-generator.spec.ts:673 › selector generator › should generate noText: no text engine `@chrome-macos-latest`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@tracing-firefox`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@tracing-firefox`
⚠️ [firefox-page] › page/page-goto.spec.ts:90 › should work with Cross-Origin-Opener-Policy `@tracing-firefox`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@firefox-macos-15-large`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-macos-15-large`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@firefox-windows-latest`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-windows-latest`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-macos-15-xlarge`
⚠️ [webkit-library] › library/playwright-client.spec.ts:26 › should connect from a page and drive the same browser `@tracing-webkit`
⚠️ [webkit-library] › library/har.spec.ts:472 › should return receive time `@webkit-windows-latest`
⚠️ [webkit-library] › library/tracing.spec.ts:494 › should produce screencast frames fit `@webkit-macos-15-xlarge`

110414 passed, 4714 skipped

---

Merge workflow run.

[github-actions]

🔴 The Firefox roll breaks the HAR serverIPAddress tests

Hi, I'm the Playwright bot — I took a look at the failing CI.

One real regression from the roll: Firefox r1541 now reports the loopback HAR serverIPAddress as the bare ::1 instead of the bracketed [::1], so library/har.spec.ts's /^127\.0\.0\.1|\[::1\]/ no longer matches. The five affected tests fail on every Firefox platform in this run (ubuntu, macOS, windows, tracing) and on every retry. The test-results DB is unambiguous: 10 failures, all on this PR, 0 across 3034 runs on other SHAs. Everything else that's red is a pre-existing flake a Firefox-only browser roll can't be responsible for — with one Firefox cookie test I'd give a second glance.

Details

The diff is Firefox-only — browser_patches/firefox/juggler/*, preferences/playwright.cfg, and browsers.json. serverIPAddress is populated by juggler's networking layer, squarely in the blast radius, which is why only the firefox-library project went red while Chromium and WebKit stayed green.

Caused by this PR

• [firefox-library] › library/har.spec.ts:629 › should have connection details

• [firefox-library] › library/har.spec.ts:639 › should have security details

• [firefox-library] › library/har.spec.ts:664 › should have connection details for redirects

• [firefox-library] › library/har.spec.ts:686 › should have connection details for failed requests

• [firefox-library] › library/har.spec.ts:699 › should return server address directly from response

  All five fail identically:

  Expected pattern: /^127\.0\.0\.1|\[::1\]/
  Received string:  "::1"
  

  Across the recent window these five failed 10/10 on this PR and 0 times over 3034 runs on unrelated SHAs — the flake hard-rule isn't met anywhere outside this PR. Fix by restoring the bracketed [::1] form in juggler's HAR reporting, or by relaxing the test to accept a bare ::1.

Uncertain — worth a second look

• [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies — hard-failed on @firefox-macos-15-xlarge and was rescued-on-retry on several other Firefox bots in the same run. It has 0 failures over 985 runs outside this PR, so I can't call it a flake, and partitioned-cookie behavior is the kind of thing a Firefox version bump shifts — so it's within the roll's blast radius. It mostly self-rescued and didn't block, but I can't prove it's unrelated. The COOP page-goto.spec.ts:90 flake is a milder instance of the same "only-on-this-PR Firefox flakiness" pattern. Neither is confirmed, but both are worth a glance alongside the HAR fix.

Pre-existing flake / infra

• [chromium] › mcp/clipboard.spec.ts:19 › clipboard write without permission dialog (macOS, Windows, Ubuntu) — a Chromium-only test (Firefox is skipped), so a Firefox roll can't touch it. Flakes independently of this PR: 8 failures across 5 other PRs over 731 runs.
• [firefox] › mcp/cli-core.spec.ts:394 › --raw on command without output (Windows only) — TimeoutError waiting for aria-ref=e2; Firefox on macOS and Ubuntu passed in the same run. Standing flake: 5 failures across 4 other PRs over 722 runs.
• [android-native] › android/device.spec.ts:56 › androidDevice.fill and [android-native] › android/webview.spec.ts:44 › should navigate page externally — Android runs don't exercise juggler's Firefox networking. Both flake across the DB: 12 failures on 2 other PRs over 348 runs.
• [playwright-test] › playwright.artifacts.spec.ts:127 and [playwright-test] › reporter-html.spec.ts:307 (both @frozen-time-runner-chromium-linux) — Chromium test-runner tests; a Firefox browser roll can't change Chromium screenshot or HTML-reporter output. These have 0 failures anywhere in the DB and are the frozen-time job's usual harness noise.
• The remaining tests reported flaky (rescued on retry) — chromium.spec.ts:301/:213/:373/:436, video.spec.ts:390/:495/:736, launcher.spec.ts:37, cli-codegen-2.spec.ts:202, oopif.spec.ts:179, beforeunload.spec.ts:130, selector-generator.spec.ts:343/:673, heap.spec.ts:223, playwright-client.spec.ts:26, har.spec.ts:472, tracing.spec.ts:494 — are the usual cross-engine noise, not caused by the roll.

_{Triaged by the Playwright bot - agent run}

[Yury Semikhatsky (yury-s)]

Updating ::1 formatting in https://github.com/microsoft/playwright-browsers/pull/2501

[Devin Rousso (dcrousso)]

should we close this in favor of #42445?

[Yury Semikhatsky (yury-s)]

should we close this in favor of #42445?

Yes, let's close. As for the fix, I'd wait for https://github.com/microsoft/playwright-browsers/pull/2501 build and roll straight to that as it shouldn't require test updates.