Skip to content

test: wait for a just-spawned child to appear in the process table before asserting the helper lists it (main Windows CI red after #940) - #943

Merged
debugmcpdev merged 1 commit into
mainfrom
fix/process-table-test-listing-race
Oct 17, 2026
Merged

debugmcpdev merged 1 commit into
mainfrom
fix/process-table-test-listing-race

Conversation

@debugmcpdev

Copy link
Copy Markdown
Collaborator

Main went red on the Windows lane right after #940 (run 39001840388): tests/unit/test-utils/process-table.test.ts > lists a live node child under its image asserted on the first scan right after spawn(), which on windows-latest came back empty while the child was alive — the second case's scan 300 ms later listed the same pid. The test now polls for the listing (up to 15 s). Test-only; #941 and #942 fail the same lane for the same reason and are re-updated once this lands.

🤖 Generated with Claude Code

…fore asserting the helper lists it (main Windows CI red after #940)

On windows-latest the first scan right after spawn() came back empty while the child was alive; a scan 300 ms later listed it. The test now polls for the listing (up to 15 s) instead of asserting the first answer.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@codecov

codecov Bot commented Oct 17, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

@debugmcpdev
debugmcpdev merged commit 3f53e4f into main Oct 17, 2026
14 checks passed
@debugmcpdev
debugmcpdev deleted the fix/process-table-test-listing-race branch October 17, 2026 04:50
debugmcpdev added a commit that referenced this pull request Oct 17, 2026
…iltered assertion (Flake Hunt red on main after #943) (#944)

The second case of the process-table unit test queried the table right
after spawn(), the same race #943 fixed for the first case: the pid is
alive, so it survives the 300 ms poll, and the image filter then drops it
because Win32_Process has not listed it yet (Flake Hunt 39027849623,
windows-latest, seed 516586464: expected [] to equal [5740]). Wait for the
listing first, with the sibling case's timeout, and say on the helper that
the image check presumes a process old enough to be listed.

Co-authored-by: JF <john.franklin@gmail.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
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