Skip to content

fix(tests): follow-ups from the timing-flakes review (#2150) #2155

Closed as not planned

Description

@devswha

Follow-up to #2150 (notes from the merge pass's review; none blocked CI).

  • Direct runs can hit Bun's 5 s default timeout on a busy host (omo MEDIUM, server/codex.test.ts ~902-926): the ratio method makes the two bounded-time cases take ~1.3 s and ~0.9 s idle (vs 0.1/0.3 s before). bun test server/codex.test.ts without --timeout, as docs/development.md describes, failed 12/12 under a 12-copy load; CI and check fast pass --timeout 60000 and are unaffected. Give these cases an explicit timeout (as ci-browser.test.ts does) or shrink the inputs.
  • The thread CPU clock helper in server/codex.test.ts ~21-33 is a copy of src/lib/markdown.test.ts ~248-259; share it.
  • Without a thread clock (Windows) the fallback process.cpuUsage() counts all threads in ~15.6 ms steps, which could push the large-answer ratio near its bound.
  • The ci-browser hang guard's error message drops stderr (scripts/ci-browser.test.ts ~90-91).

Activity

  1. devswha commented on Oct 28, 2026

    @devswha
    OwnerAuthor

    Closing under the new follow-up bar: a follow-up issue is kept only for a reproduced, user-visible MEDIUM finding. The reproduced MEDIUM timeout was fixed by 391b906 (#2169); the remaining items are below the bar. Below the follow-up bar: LOW/unreproduced; recorded on PR #2150 and in this issue for reference.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions