humanspeak / svelte-motion / 30035816765

62%

main: 57%

push

github

feat(gesture): upstream parity — velocity continuity, first-start idle, wildcard keyframes (#453)

* docs(plans): add upstream-parity-next3 batch — velocity continuity, first-start wipe, wildcard keyframes

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* docs(plans): dispatch parity 001 + 002 to opus executors on feature-parity

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* test(controls): first-ever start must travel from non-neutral idle

Adds a red e2e anchor: the beam idles at its initial variant (scaleX
0.16), yet the FIRST controls.start('launch') teleports straight to
scaleX 1 (from-side wipes to 1, WAAPI runs 1->1). Samples the beam from
the moment of the click and asserts no early frame reaches ~1 and a
mid-flight sample sits between idle and the endpoint.

Fails on unfixed code: early scaleX jumps to 1 at ~111ms.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(controls): first-ever start animates from the non-neutral idle

The first controls.start() from a non-neutral idle (e.g. beam scaleX
0.16) produced no visible motion: WAAPI reads its from-value off the
live computed style, but the reactive inline transform is rewritten off
the idle keyframes the instant a command is received, so the transform
reads back as the property default (scaleX 1) and the animation runs
1->1.

Two changes in startAnimationControlsDefinition + the inline-style
ternary:
- Hold the idle keyframes as the inline base during a first in-flight
  command (ternary falls back to initialKeyframes while
  lastAnimationControlsTarget is still undefined).
- Race-free: expand each idle-defined channel of the first command into
  an explicit [from, to] keyframe pair, mirroring upstream where the
  motion value holds the idle value as the from. Only the first command
  needs this; later commands animate from the held target.

Full animation-controls e2e suite green (incl. the 005 stop-freeze trio
and 006 detach pair). New first-start regressi... (continued)

985 of 1823 branches covered (54.03%)

Branch coverage included in aggregate %.

134 of 186 new or added lines in 4 files covered. (72.04%)

3745 of 5855 relevant lines covered (63.96%)

23.12 hits per line