Skip to content

fix(v7): Only add main process context lines to renderer frames in the app directory - #1475

Open
timfish wants to merge 2 commits into
v7from
ccr-6ee1cfbd-3n8uz3-v7
Open

timfish wants to merge 2 commits into
v7from
ccr-6ee1cfbd-3n8uz3-v7

Conversation

@timfish

@timfish timfish commented Oct 15, 2026 •

Copy link
Copy Markdown
Collaborator

Backport of #1469 to v7.

Renderer stack frames mostly hold renderer URLs, which the main process ContextLines integration cannot resolve. Preload script frames are the exception: they are files in the app directory, and the main process adds context lines to them.

This marks events from renderers in sdkProcessingMetadata and wraps the ContextLines integration. For renderer events, it only adds context lines to frames whose file is inside app.getAppPath(). Main process and utility process events are not changed. The wrapper also applies when users add contextLinesIntegration themselves.

The marker uses sdkProcessingMetadata, so it is not sent to Sentry.

…p directory

Renderer stack frames mostly hold renderer URLs. Preload script frames
are files in the app directory and keep their main process context
lines. Renderer frames outside the app directory get none.

timfish commented Oct 15, 2026

Copy link
Copy Markdown
Collaborator Author

Two E2E jobs failed, both on windows-2022 with Electron 38.8.6. The failures do not look related to this change, so I re-ran them once.

  • Shard 2: all tests that ran passed, including the preload error test. Then the Vitest worker exited unexpectedly.
  • Shard 4: all 22 tests passed. Then an unhandled envelope assertion from the Electron net Tracing test failed the run. That test sends no renderer errors, so it does not reach the changed code.

The other 99 jobs passed, and #1451 passed these same jobs on v7.


Generated by Claude Code

Also check that the ContextLines integration gets wrapped.

timfish commented Oct 15, 2026

Copy link
Copy Markdown
Collaborator Author

E2E Tests (windows-2022, 42.11.6, 4) failed on b66fc10. All tests that ran passed. Then the Vitest process ran out of JavaScript heap after the native crash tests. The only change in b66fc10 is a Symbol in place of a string key, and this job passed on the previous commit, so I will re-run it once when the run finishes.


Generated by Claude Code

@timfish
timfish requested a review from mydea October 15, 2026 14:19
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.

1 participant