humanspeak / svelte-motion / 28749693314

60%

main: 57%

push

github

fix(interaction): honor resolved transition for whileTap, delegate to motion springs (#433)

whileTap ignored the resolved transition entirely: a component `transition`
prop / `<MotionConfig>`, an inline `whileTap.transition`, and the base
variant's transition all had no effect on the tap. Both press and release ran
a hardcoded cubic-bezier (`[0.22, 1.1, 0.36, 1]`) — a tween pretending to be a
spring, matching neither framer's spring feel nor a clean settle.

Resolve the gesture transition with framer-motion's precedence instead:
- press:   inline `whileTap.transition` -> component/MotionConfig -> defaults
- release: base variant `animate` transition -> component/MotionConfig -> defaults
- when none is set, pass no transition so motion-dom applies its own per-value
  defaults (scale/other-transform springs, ease for the rest) — exact parity,
  inheriting upstream tuning for free.

The inline `whileTap.transition` is now split out (mirrors splitHoverDefinition)
so it can drive the press and never leaks into the animated keyframes. The
hover-reapply-after-tap path now honors the hover definition's own transition.

The hardcoded tweens originally existed to dodge a spring "velocity
accumulation" runaway during rapid press/release. That was tied to now-removed
manual transform rebasing; motion values are reused across taps so velocity is
continuous. A new rapid-tap e2e hammers 25 press/release cycles at the spring's
resonant cadence and asserts scale stays bounded and settles to rest.

Verified against upstream (~/Github/motion):
- motion-dom/.../utils/default-transitions.ts (getDefaultTransition)
- motion-dom/.../interfaces/motion-value.ts (per-value application)
- motion-dom/.../interfaces/visual-element-target.ts (component precedence)
- motion-dom/.../generators/spring.ts:260 (scale 550/30 is underdamped, ~7%
  overshoot by design — not critically damped)

- Add tap-only rapid-tap demo page + e2e; link from tests index
- Document tap-transition tuning... (continued)

852 of 1647 branches covered (51.73%)

Branch coverage included in aggregate %.

3292 of 5286 relevant lines covered (62.28%)

23.28 hits per line