[elezar]

Summary

Program rows relied on the program's 900 s per-test default, because the runner passed neither --timeout nor --grace. One hung test could use much of the 45-minute run timeout. The nextest path stopped a test after 120 s.

• Derive the per-test timeout (30 s x 4 = 120 s) from shared constants, and generate the nextest profile from them so the two paths cannot drift apart.
• Pass --timeout and --grace (10 s, nextest's default) to the program.
• Add a test for the generated nextest config.

Related Issue

Part of #5416 (first item only; the issue stays open). Groundwork for #5333, which propagates this budget into testkit's inner calls.

Testing

• cargo fmt --check, cargo clippy --all-targets -- -D warnings and cargo test for tests/runner pass.
• The full e2e lane was not run locally.

[github-actions]

Compatibility review (advisory)

New or updated breaking changes; see the full report for all findings.

1. cli — No explicit risk acceptance found

openshell-test-runner run --program changes its child invocation from test-program run --junit PATH (900-second default) to test-program run --timeout 120 --grace 10 --junit PATH.

• Impact: Existing prebuilt testkit suites with tests requiring 120–900 seconds now receive SIGTERM and report timeout failures with exit code 1, without a runner flag to restore their previous budget.
• Action: Preserve the 900-second --program default in v0.1.8 and make the 120-second limit opt-in through a documented runner flag.
• PR: test(runner): pass an explicit per-test timeout and grace to program rows #5450

Review details

• Execution: complete
• Candidate: PR #5450 (29a08e491d961743824200fac035e389f911d875)
• Baseline: 7873aaa3b8a3529459034cfce3079077a1a1e67b (7873aaa3b8a3529459034cfce3079077a1a1e67b)
• Reviewed: rust-sdk, python-sdk, go-sdk, typescript-sdk, api-behavior, cli, configuration, policy, helm, state

Advisory review; publication gates are unchanged.

Branch check run