Skip to content

Sync with upstream patchright (through v1.61.1) - #7

Open
StudentWan wants to merge 23 commits into
mainfrom
users/danielwan/sync-upstream-v1.61.1
Open

StudentWan wants to merge 23 commits into
mainfrom
users/danielwan/sync-upstream-v1.61.1

Conversation

@StudentWan

Copy link
Copy Markdown
Owner

Merges 22 commits from Kaliiiiiiiiii-Vinyzu/patchright into our fork.

What's coming in

Our fork's delta

Against the merge base, our only remaining divergence was the Edge NTP crash fix (9 lines). It merged cleanly and is intact at driver_patches/crNetworkManagerPatch.ts:575-583. The earlier js-handle and expose_binding fixes are no longer part of our delta — absorbed upstream or reverted by 699c377.

Conflict resolution

One conflict, in patchright.patch, resolved by taking upstream's side.

That file is a generated documentation artifact — its own first line says it "is not used, it is only for documentation", and .github/workflows/patch_file_updater.yml regenerates it wholesale via diff -ruN playwright patchright. The conflict came from upstream's Kaliiiiiiiiii-Vinyzu#222 reformat (41,063 insertions / 6,575 deletions to that file alone) colliding with our 9-line insert. Hand-resolving it would be overwritten by the next CI patch run.

Not yet verified

npm run typecheck and npm run patch have not been run on this merge. npm run patch is the meaningful gate — our patches use ts-morph getMethodOrThrow / getClassOrThrow, which throw at patch time if upstream's refactor moved anything they reach for.

Worth a look during review

Our NTP early-return sits before the if (event.resourceType !== 'Document') return; guard, whereas upstream's reformatted version puts that guard first. The merge preserved our ordering, so behavior is unchanged — but it means every ntp.msn request gets an explicit Fetch.continueRequest, including subresources, not just Documents.

🤖 Generated with Claude Code

Vinyzu and others added 23 commits May 10, 2026 13:31
…nyzu#199)

* chore: scaffold fix-v1.60.0-25698100548

* Port to v1.60.0

Note: Might revert some test skips later

* Resolve Review

* Revert Regressions

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Vinyzu <50874994+Vinyzu@users.noreply.github.com>
Co-authored-by: Sulthon Zainul Habib <4921059+sulthonzh@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
…ods in JSHandle class and remove unnecessary context initialization
…nyzu#216)

Co-authored-by: Vinyzu <50874994+Vinyzu@users.noreply.github.com>
Co-authored-by: Federico Pomar <1781508+fmpomar@users.noreply.github.com>
…nyzu#219)

Co-authored-by: Vinyzu <50874994+Vinyzu@users.noreply.github.com>
Sync with Kaliiiiiiiiii-Vinyzu/patchright (22 commits, through v1.61.1 fixes,
JSHandle refactor, visibility checks, and the Kaliiiiiiiiii-Vinyzu#222 driver refactor/formatting).

Conflict in patchright.patch resolved by taking upstream's side. That file is a
generated documentation artifact (regenerated by .github/workflows/patch_file_updater.yml)
and is not used to patch the driver. The Edge NTP crash fix remains intact in
driver_patches/crNetworkManagerPatch.ts, which merged cleanly.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

This branch had an error being deployed

1 failed deployment
pr-artifact-build — 9c36c24e Deployed Jul 29, 2026 by StudentWan via ​ #8
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants