[Pavel Feldman (pavelfeldman)]

Summary

• Revert feat(test-runner): experimental esbuild transpilation #43320: remove the PLAYWRIGHT_EXPERIMENTAL_ESBUILD transform, its tests and the optional esbuild peer dependency.
• The test runner will support bun instead (chore(test runner): run tests under bun, add parse error locations #43402).

[github-actions]

Test results for "tests 1"

1 failed
❌ [chromium-library] › library/video.spec.ts:762 › screencast › should work with video+trace @chromium-playwright-x64-ubuntu24-64-core-node22

4 flaky

⚠️ [playwright-test] › ui-mode-test-watch.spec.ts:185 › should watch new file `@playwright-x64-server2025-64-core-node22`
⚠️ [chromium-library] › library/chromium/chromium.spec.ts:301 › should report intercepted service worker requests in HAR `@chromium-playwright-x64-ubuntu24-64-core-node22`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@firefox-playwright-x64-ubuntu24-64-core-node22`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-playwright-x64-ubuntu24-64-core-node22`

21764 passed, 524 skipped

---

Merge workflow run.

[github-actions]

Test results for "MCP"

5 failed
❌ [chrome] › mcp/request-blocking.spec.ts:79 › blocked takes precedence (hostname) @mcp-macos-latest-chrome
❌ [chrome] › mcp/video.spec.ts:23 › should work with recordVideo (persistent) @mcp-macos-latest-chrome
❌ [chrome] › mcp/files.spec.ts:201 › closing browser during download does not crash the server @mcp-windows-latest-chrome
❌ [msedge] › mcp/http.spec.ts:92 › http transport browser lifecycle (isolated) @mcp-windows-latest-msedge
❌ [msedge] › mcp/http.spec.ts:132 › http transport browser sigint @mcp-windows-latest-msedge

5149 passed, 614 skipped

---

Merge workflow run.

[github-actions]

Hi, I'm the Playwright bot and I took a first look at the CI failures.

🟡 Three failures are confirmed pre-existing flakes; three remain uncertain

This PR removes the opt-in experimental esbuild transform and its related tests and dependency. It does not change the video, download, request-blocking, or MCP HTTP lifecycle paths that failed.

Details

Overall assessment

I found exact unrelated failure history for three tests. The other three failed only in two MCP jobs with browser-connection or lifecycle symptoms, but the test-results database has no unrelated failure for those exact tests. Under the CI triage rules, they remain uncertain until a rerun passes or the same tests fail on an unrelated SHA.

Pre-existing flake / infra

• [chromium-library] › library/video.spec.ts:762 › screencast › should work with video+trace — the same test failed with the same Cannot read properties of undefined (reading 'file') error on unrelated SHAs, including a038fe7 and d96199f. The database has seven unrelated failing runs and 570 passing runs for this test.
• [chrome] › mcp/files.spec.ts:201 › closing browser during download does not crash the server — the same Windows Chrome test and empty download-progress output failed in #43198 and unrelated push runs 37515631161 and 37354633372.
• [msedge] › mcp/http.spec.ts:132 › http transport browser sigint — the same Windows Edge test failed with the same missing close browser event in unrelated push runs 38655853177 and 38466762334.

Uncertain

• [chrome] › mcp/request-blocking.spec.ts:79 › blocked takes precedence (hostname) and [chrome] › mcp/video.spec.ts:23 › should work with recordVideo (persistent) — both timed out with MCP error -32000: Connection closed in the same macOS Chrome job. The PR does not touch these MCP paths, but the database has no unrelated failure for either exact test across 400 recent runs. A successful job rerun would clear them.
• [msedge] › mcp/http.spec.ts:92 › http transport browser lifecycle (isolated) — timed out in the Windows Edge job. The PR does not touch MCP HTTP lifecycle handling, but the database has no unrelated failure for this exact test across 391 recent runs. A successful job rerun would clear it.

_{Triaged by the Playwright bot - agent run}

[github-actions]

Hi, I'm the Playwright bot and I took a first look at the CI failures.

🟢 The failure is a confirmed pre-existing flake

library/video.spec.ts:762 › screencast › should work with video+trace has the same failure on the PR's base SHA and on unrelated runs. This PR is clear.

Details

The PR removes the experimental esbuild transform and its related tests and build configuration. It does not change video capture, tracing, or this test's code path.

Pre-existing flake / infra

• [chromium-library] › library/video.spec.ts:762 › screencast › should work with video+trace — all attempts here found no screencast-frame, then failed with Cannot read properties of undefined (reading 'file'). The exact test, CI project, bot, and error also occurred on base SHA 5c42a39, where it passed after two retries. In the refreshed three-day test-results data for this bot, it passed 15 of 20 runs and failed 5; retries rescued 12 of those passes.

_{Triaged by the Playwright bot - agent run}