humanspeak / svelte-motion / 26317570612

72%

main: 3%

push

github

feat(hooks): add useFollowValue (#376)

* feat(hooks): add useFollowValue

Add `useFollowValue` — the generalised follow hook accepting any
ValueAnimationTransition (spring / tween / inertia / keyframes) and
returning an augmented motion-dom MotionValue. `useSpring` becomes a
thin wrapper that delegates to `useFollowValue` with `type: 'spring'`
baked in, so the spring API is unchanged.

Ships with a 13-test spec, a /docs/use-follow-value reference page,
and a /examples/use-follow-value demo (six pointer-driven followers
across four spring variants and two tweens — the canonical
"one source, many personalities" pattern from the framer-motion docs).

Adopt docs-kit 2026.5.38, which introduces EmbeddedExampleV2 — the
brutalist-mono panel cousin of ExampleV2 designed for inlining a live
demo inside a doc page. The local Example.svelte wrapper is now a
thin re-skin that delegates to it, so all 17 existing doc-page embeds
inherit the new bar styling.

Misc:
- Carve overload-syntax exception into the .coderabbit.yaml
  utils/*.ts path rule (TypeScript forbids arrow syntax on overloads;
  useFollowValue / useSpring / useCycle follow function-decl).
- Ignore root-level .png artifacts (Playwright sign-off screenshots).

* fix(test): make useFollowValue .destroy() test deterministic

The previous assertion read `v.get() === 0` synchronously after
`src.set(100)` on a default-spring follower. Springs are async, so
the assertion could pass as a false positive even if the follow had
never been torn down — the test wasn't actually proving anything.

Switch to a `type: 'tween', duration: 0` follower with a `waitFor`
poll helper that proves the source→follower path is wired BEFORE
calling `.destroy()`, then pumps several frames post-destroy to
confirm subsequent source changes don't propagate.

Addresses CR feedback on PR #376.

497 of 640 branches covered (77.66%)

Branch coverage included in aggregate %.

25 of 26 new or added lines in 2 files covered. (96.15%)

1965 of 2795 relevant lines covered (70.3%)

12.53 hits per line