chore(release): version packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@dunky.dev/state-machine-dom@0.1.0

Minor Changes

• #32 e6caca8 Thanks @ivanbanov! - Add @dunky.dev/state-machine-dom — the DOM half of the bindings translation,
  shared by every DOM target. The aria-* attribute projection and the payload
  adapters (onValueChange/onWheel/onScroll/onScrollEnd → neutral
  payloads, with preventDefault bound to its event) were byte-identical in the
  React and Solid normalizers; they now live once, in this package, and each
  target keeps only what genuinely differs: its handler prop names
  (onChange/onDoubleClick vs onInput/onDblClick), the focusable →
  tabindex casing (tabIndex vs tabindex), and its value serialization
  (React passes ARIA booleans through; Solid stringifies them).

  No API change for consumers of the React or Solid packages — normalize
  behaves exactly as before; the shared package becomes a dependency of both.
  The motivation is drift-proofing: a payload-adapter fix previously had to be
  applied to each DOM target by hand, and had already diverged once.

Patch Changes

• Updated dependencies [c8e94e4]:
  • @dunky.dev/state-machine-bindings@0.4.1

@dunky.dev/native-state-machine@0.4.0

Minor Changes

• #62 caafda4 Thanks @ivanbanov! - The modal binding now reaches React Native instead of being dropped:
  normalize() maps it to aria-modal, the web-aligned alias RN routes to
  accessibilityViewIsModal.

  normalize({ modal: true }); // { 'aria-modal': true }

  It was dropped on the assumption RN had no element-attr analog. It does —
  aria-modal sits in the same alias block as aria-hidden, which this
  normalizer already targets. The effect is iOS-only (VoiceOver stops reading
  siblings of the modal surface); Android has no sibling-inerting equivalent, so
  it degrades to a no-op there, the same per-platform fan-out aria-hidden
  already relies on.

  Without it, a dialog authored once in core announced as modal on the web and as
  an ordinary view on native — the substrate-agnostic contract leaking a hole
  exactly where a screen reader user would notice it.

Patch Changes

• #60 c8e94e4 Thanks @ivanbanov! - Rolls back 0.4.0's translation-contract surface — it shipped by mistake.
  DROPPED_HANDLERS/DROPPED_ATTRS are removed, and HandlerTargets/
  AttrTargets are now the partial rename maps
  (Partial<Record<HandlerKey, string>>) instead of exhaustive
  null-accounting records.

  What stays is the part that mattered: maps and drop sets are typed by the
  real vocabulary keys (HandlerKey/AttrKey), so a typo'd or unknown entry
  is a compile error instead of a silent leak.

• Updated dependencies [e6caca8, 6deab75, 6deab75, b70bedc, c8e94e4, e6caca8]:

  • @dunky.dev/react-state-machine@0.3.4
  • @dunky.dev/state-machine-utils@0.4.0
  • @dunky.dev/state-machine@0.3.3
  • @dunky.dev/state-machine-bindings@0.4.1

@dunky.dev/state-machine-utils@0.4.0

Minor Changes

• #66 6deab75 Thanks @ivanbanov! - Remove the dead exports: the positioning module (Placement, Side,
  PositioningOptions, placementToSide, pickSide), memo, and
  composeHandlers. Nothing in the repo ever consumed them — mergeProps
  composes handlers through its own private helper, and positioning was
  speculative vocabulary for floating components that don't exist yet.
  mergeProps is now the package's whole surface. Minor (not patch) because
  the symbols were publicly exported: any external import of them breaks.
  Positioning will come back designed against a real floating component when
  one lands.

• #66 6deab75 Thanks @ivanbanov! - Export composeHandlers — the handler-pair composition mergeProps has
  always applied to overlapping on* props, now public: the consumer handler
  runs first, and the library handler is skipped when the consumer prevented
  default (the first argument's defaultPrevented, per Radix/Ark conventions).
  No behavior change anywhere — mergeProps calls the same function; it was
  just private before.

  import { composeHandlers } from "@dunky.dev/state-machine-utils";
  
  const onClick = composeHandlers(consumerOnClick, libraryOnClick);

@dunky.dev/solid-state-machine@0.3.0

Minor Changes

• #32 e6caca8 Thanks @ivanbanov! - Add @dunky.dev/solid-state-machine — the Solid bindings target.

  A first-class Solid bridge (not a React re-export): useMachine mirrors the
  connector's snapshot into a Solid createStore (via reconcile) so reading a
  field in JSX is fine-grained, runs the lifecycle through onSettled/onCleanup,
  keeps props fresh with a tracked setProps effect, and runs each
  ComponentEffect as its own dep-tracked createEffect(compute, apply).
  useSelector returns a Solid accessor. normalize maps the agnostic bindings
  to Solid DOM props (onInput, onDblClick, tabindex) and mergeProps
  applies Solid's class concat + single-object style merge. The same connect
  and machine config run unchanged across React, Solid, React Native, and OpenTUI.

  Targets Solid 2.0 as a first-class citizen: the peer range is solid-js
^2.0.0-rc.1. Solid 1.x is not supported — 2.0 removed the surface a 1.x
  bridge would stand on (solid-js/store, single-argument createEffect,
  onMount) and 1.x lacks the root exports this package imports, so, like the
  rest of the Solid ecosystem (router, TanStack, solid-primitives), the majors
  are version-split. Two consumer-facing 2.0 behaviors: writes commit on the
  microtask queue (call flush() in tests before asserting), and JSX comes from
  the renderer package ("jsxImportSource": "@solidjs/web", render from
  @solidjs/web).

Patch Changes

• #32 e6caca8 Thanks @ivanbanov! - Add @dunky.dev/state-machine-dom — the DOM half of the bindings translation,
  shared by every DOM target. The aria-* attribute projection and the payload
  adapters (onValueChange/onWheel/onScroll/onScrollEnd → neutral
  payloads, with preventDefault bound to its event) were byte-identical in the
  React and Solid normalizers; they now live once, in this package, and each
  target keeps only what genuinely differs: its handler prop names
  (onChange/onDoubleClick vs onInput/onDblClick), the focusable →
  tabindex casing (tabIndex vs tabindex), and its value serialization
  (React passes ARIA booleans through; Solid stringifies them).

  No API change for consumers of the React or Solid packages — normalize
  behaves exactly as before; the shared package becomes a dependency of both.
  The motivation is drift-proofing: a payload-adapter fix previously had to be
  applied to each DOM target by hand, and had already diverged once.

• #32 e6caca8 Thanks @ivanbanov! - preventDefault on the adapted payloads (ChangePayload, WheelPayload) now
  actually works. normalize() used to copy the native event's preventDefault
  onto the payload detached from its event, so the first connect() to call
  payload.preventDefault() would throw TypeError: Illegal invocation — native
  DOM methods require this to be a real Event. The payload now carries a
  closure bound to the originating event:

  // connect() side — this used to throw, now suppresses the default as promised
  onValueChange: (payload) => {
    payload.preventDefault?.();
  };

  Latent until now (no in-repo connect() calls it yet), but it is the behavior
  the bindings contract promises, so it's fixed in both DOM targets before a
  component relies on it.

• #32 e6caca8 Thanks @ivanbanov! - A ComponentEffect body now runs untracked, so its authored deps list is the
  whole re-run contract — identical to the React target's dep array. Previously
  the body executed inside the tracking scope, so any prop the effect merely read
  became a hidden dependency and re-ran it (cleanup + re-subscribe) on changes to
  props it never declared.

  const escape: ComponentEffect<M, Props> = [
    (machine, props) => {
      void props.onEscapeKeyDown; // read, but NOT a dep — no longer re-runs on change
    },
    ["closeOnEscape"], // ONLY this prop re-runs the effect, on every target
  ];

• Updated dependencies [e6caca8, 6deab75, 6deab75, b70bedc]:

  • @dunky.dev/state-machine-dom@0.1.0
  • @dunky.dev/state-machine-utils@0.4.0
  • @dunky.dev/state-machine@0.3.3

@dunky.dev/state-machine@0.3.3

Patch Changes

• #65 b70bedc Thanks @ivanbanov! - Harden the notify and teardown paths across the core, and cut hot-path
  allocations:

  • A listener removed during a notify pass no longer fires again when a nested
    notify (a send from inside a subscriber) rebuilds the iteration snapshot
    mid-pass — unsubscribing is now final even under re-entrancy. The same
    guarantee now holds for connector and store subscribers, and the mechanism
    lives in one shared primitive instead of three near-copies.
  • A state cleanup that throws no longer skips the remaining cleanups or leaves
    the pass populated: every cleanup runs (timers and subscriptions all
    release), the first error is rethrown after the pass, and the next stop
    cannot double-run them.
  • A sync() rule or combine().subscribe() disposed by hand now detaches from
    the composition's registry — long-lived groups with subscribe/unsubscribe
    churn no longer grow it without bound, and stop() no longer re-runs
    hand-run disposers.
  • Computed recompute is allocation-free: dep keys/values live in reused
    buffers and are captured at read time, so the old post-pass that re-read
    every dep is gone. In the benchmark suite this lands recompute ~1.5× and
    4-deep computed chains ~1.6× faster.
  • machine.select is built once and reused instead of allocating a fresh
    facade object on every property access, so its identity is stable (safe for
    dependency arrays).
  • Dropped the internal write-only version counter — bumped on every notify,
    read by nothing.

@dunky.dev/opentui-state-machine@0.3.4

Patch Changes

• #60 c8e94e4 Thanks @ivanbanov! - Rolls back 0.4.0's translation-contract surface — it shipped by mistake.
  DROPPED_HANDLERS/DROPPED_ATTRS are removed, and HandlerTargets/
  AttrTargets are now the partial rename maps
  (Partial<Record<HandlerKey, string>>) instead of exhaustive
  null-accounting records.

  What stays is the part that mattered: maps and drop sets are typed by the
  real vocabulary keys (HandlerKey/AttrKey), so a typo'd or unknown entry
  is a compile error instead of a silent leak.

• Updated dependencies [6deab75, 6deab75, c8e94e4]:

  • @dunky.dev/state-machine-utils@0.4.0
  • @dunky.dev/state-machine-bindings@0.4.1

@dunky.dev/react-state-machine@0.3.4

Patch Changes

• #32 e6caca8 Thanks @ivanbanov! - Add @dunky.dev/state-machine-dom — the DOM half of the bindings translation,
  shared by every DOM target. The aria-* attribute projection and the payload
  adapters (onValueChange/onWheel/onScroll/onScrollEnd → neutral
  payloads, with preventDefault bound to its event) were byte-identical in the
  React and Solid normalizers; they now live once, in this package, and each
  target keeps only what genuinely differs: its handler prop names
  (onChange/onDoubleClick vs onInput/onDblClick), the focusable →
  tabindex casing (tabIndex vs tabindex), and its value serialization
  (React passes ARIA booleans through; Solid stringifies them).

  No API change for consumers of the React or Solid packages — normalize
  behaves exactly as before; the shared package becomes a dependency of both.
  The motivation is drift-proofing: a payload-adapter fix previously had to be
  applied to each DOM target by hand, and had already diverged once.

• #60 c8e94e4 Thanks @ivanbanov! - Rolls back 0.4.0's translation-contract surface — it shipped by mistake.
  DROPPED_HANDLERS/DROPPED_ATTRS are removed, and HandlerTargets/
  AttrTargets are now the partial rename maps
  (Partial<Record<HandlerKey, string>>) instead of exhaustive
  null-accounting records.

  What stays is the part that mattered: maps and drop sets are typed by the
  real vocabulary keys (HandlerKey/AttrKey), so a typo'd or unknown entry
  is a compile error instead of a silent leak.

• #32 e6caca8 Thanks @ivanbanov! - preventDefault on the adapted payloads (ChangePayload, WheelPayload) now
  actually works. normalize() used to copy the native event's preventDefault
  onto the payload detached from its event, so the first connect() to call
  payload.preventDefault() would throw TypeError: Illegal invocation — native
  DOM methods require this to be a real Event. The payload now carries a
  closure bound to the originating event:

  // connect() side — this used to throw, now suppresses the default as promised
  onValueChange: (payload) => {
    payload.preventDefault?.();
  };

  Latent until now (no in-repo connect() calls it yet), but it is the behavior
  the bindings contract promises, so it's fixed in both DOM targets before a
  component relies on it.

• Updated dependencies [e6caca8, 6deab75, 6deab75, b70bedc]:

  • @dunky.dev/state-machine-dom@0.1.0
  • @dunky.dev/state-machine-utils@0.4.0
  • @dunky.dev/state-machine@0.3.3

@dunky.dev/state-machine-bindings@0.4.1

Patch Changes

• #60 c8e94e4 Thanks @ivanbanov! - Rolls back 0.4.0's translation-contract surface — it shipped by mistake.
  DROPPED_HANDLERS/DROPPED_ATTRS are removed, and HandlerTargets/
  AttrTargets are now the partial rename maps
  (Partial<Record<HandlerKey, string>>) instead of exhaustive
  null-accounting records.

  What stays is the part that mattered: maps and drop sets are typed by the
  real vocabulary keys (HandlerKey/AttrKey), so a typo'd or unknown entry
  is a compile error instead of a silent leak.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
dunky-state-machine	Ready	Preview	Aug 22, 2026 1:51pm