[JacobStephens2]

Summary

Give the historical Rust pin test enough Trial time for Python process startup on loaded macOS runners.

 trial_installs_its_historical_rust_pin_before_the_session
-    fixture.execution(5)
+    fixture.execution(30)

Both Arms still run concurrently, and other tests' timeouts stay unchanged.

Closes #1422

Evidence

• Before: Historical Rust pin Trial test times out on loaded macOS runners #1422 records both Arms timing out with the 5-second budget on a loaded macOS runner. The original test also passed locally, consistent with the reported flake.
  After: cargo test --bin thirdshift evals::experiment::execution::tests::trial_installs_its_historical_rust_pin_before_the_session -- --exact passes with the 30-second budget.
• The same exact test fails when the install step is temporarily removed (pinned Rust 1.99.0 download refused by Trial proxy) and when it is temporarily called twice ("install\ninstall\n" versus "install\n"). Both mutations were restored before committing.
• cargo test --bin thirdshift evals::experiment::execution::tests: 21 passed.
• cargo fmt --check: passed.
• /usr/bin/time -p scripts/affected-tests main: passed the full Rust suite via the cargo test fallback; 1434.61 seconds wall time, including 64 seconds of compilation. Unit tests: 1529 passed, 3 ignored; all integration suites passed.

Test seam: Trial execution through Execution::prepare and Execution::run, using the existing fake Arms and Rust tool fixtures.

Merge Danger

Door: two-way

Blast Radius: tests

One test can now wait up to 30 seconds when its Trial stalls. The change is easy to revert.

Review coverage

Reviewed against main with thirdshift-code-review: Standards 0 findings; Spec 0 findings. Both reviewers read the only changed file, src/evals/experiment/execution/tests.rs. Unread changed files: none. Generated-output exclusions: none.

Unaddressed findings

• Standards: none.
• Spec: none.
• Security: none.

Built with codex codex-cli 0.164.0 · gpt-6.1-sol · xhigh

Security review outcome

No unaddressed introduced findings.