[richiemcilroy]

Follow-up to #2666, from capy's review: the soak cases' scripted holds had fixed lengths, so a short CAP_VERIFY_SOAK_SECS put a hold over the end of the stimulus. With 30 s, the Studio soak's 10 s hold started at 20 s and ran to the end. The stimulus never showed its last picture, and the case hung until its timeout.

Holds now scale with the length and end by five sixths of it: Studio 3 s at 1/3 and 10 s at 2/3, Instant 5 s at 1/2, each capped at a tenth or a sixth of the length. At 1800 s the holds are unchanged.

CAP_VERIFY_SOAK_SECS=30 on a MacBook Pro: both soak cases pass (Studio 35.9 s, Instant 35.0 s).

Confidence Score: 5/5

The PR appears safe to merge.

Summary

The PR scales scripted holds for shorter soak runs while keeping the original hold lengths at 1800 seconds.

• Soak recordings scale their holds to finish before the stimulus ends.

_{Reviews (1) · Last reviewed commit: "test(verify): soak holds end before the ..." · Reviewed by Greptile}

[capy-ai]

Capy found no issues.