Skip to content

test(drivers): the codex-computer-use idle test gives two calls in a row room on a busy runner - #700

Merged
earakely-scale merged 1 commit into
mainfrom
edgararakelyan/codex-driver-idle-test-flake
Nov 15, 2026
Merged

earakely-scale merged 1 commit into
mainfrom
edgararakelyan/codex-driver-idle-test-flake

Conversation

@earakely-scale

@earakely-scale earakely-scale commented Nov 15, 2026 •

Copy link
Copy Markdown
Collaborator

What

tst/unit/drivers/codex_computer_use_driver_test.py::test_a_repl_left_idle_is_closed_without_another_call_and_its_session_starts_afresh
failed on a busy CI runner (#699's unit (2/2)):

AssertionError: assert (False)
 +  where False = <built-in method startswith of str object>('1 in ')
 +    where ... = 'None in 6467'.startswith
  • The test sets IDLE_S to 0.2 s. On that runner more than 0.2 s passed between its set 1 and its get, so the
    driver's sweep closed the REPL in between, and get found a fresh one.
  • The fix:
    • IDLE_S is now 1 s, a module constant that says why: well over two calls in a row on a busy runner.
    • _gone waits up to 5 s for the REPL to close, instead of 2 s.
  • The test still checks the same thing: a REPL left idle is closed without another call, and the next call starts
    afresh. It takes about a second longer.

Proof

  • The driver's tests: 16 passed locally; the idle test passed 3 times in a row.
  • main's recent unit runs pass, so it's a timing flake; this widens the window it needs.

🤖 Generated with Claude Code

RetriggerView in GreptileConfidence Score: 5/5

The PR appears safe to merge.

Summary

Gives the idle test more room on busy runners.

  • The idle test gives two calls room before checking REPL closure.

No actionable issues found. Tests were not run during this review.

Reviews (1) · Last reviewed commit: "test(drivers): the codex-computer-use id..." · Reviewed by Greptile

…row room on a busy runner

With IDLE_S at 0.2 s, a busy CI runner let the driver close the REPL between the test's set and get, so the get found a
fresh one ("None in 6467") and the test failed. The idle window is now 1 s, and the wait for the REPL to close is 5 s.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@earakely-scale
earakely-scale requested a review from a team as a code owner November 15, 2026 17:19
@earakely-scale
earakely-scale merged commit 50fda36 into main Nov 15, 2026
35 checks passed
@earakely-scale
earakely-scale deleted the edgararakelyan/codex-driver-idle-test-flake branch November 15, 2026 17:41
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