fix(e2e): lower REFERENCE_CALIBRATION_MS to 18000 for stability margin

REFERENCE=20000 gave multiplier < 1.0 on fast CI (0.97× → 58s timeout)
causing 7 scenario timeouts. REFERENCE=18000 (below min observed 19.4s)
ensures multiplier ≥ 1.0 (1.08× → 65s timeout on typical CI).

Light Playwright timeouts (10s) are working correctly - all failures
were cucumber step timeouts, not element-finding timeouts.
This commit is contained in:
linonetwo 2026-05-04 22:53:24 +08:00
parent cec3834161
commit 40f8b97a59

View file

@ -15,11 +15,11 @@ import path from 'path';
*/
/**
* Baseline calibration time on reference GitHub Actions CI.
* Measured empirically: smoke test (16 steps + app launch + filesystem watch) 20s.
* This is the "1×" reference point. All timeout calculations scale from this.
* Fastest observed calibration time on GitHub Actions CI.
* Setting this below the minimum ensures multiplier stays 1.0
* on typical CI, providing stability margin for heavy operations.
*/
const REFERENCE_CALIBRATION_MS = 20000; // ~20s on baseline CI
const REFERENCE_CALIBRATION_MS = 18000; // fastest observed CI (~19.4s), ensures multiplier ≥ 1.0
/**
* Minimum step timeout budget on the reference (1×) machine.