chore: 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

@murky-web/oxlint-plugin-solid@1.2.0

Minor Changes

• 8fdeefb: Add two rules for Solid 2.0's split effects and unowned ref callbacks, and
  repair two existing ones.

  solid/no-untracked-effect-read reports a reactive read in the apply phase of a
  two-argument createEffect. That phase runs untracked, so
  createEffect(() => roomId(), (id) => connect(id, theme())) never re-runs when
  theme changes. Nothing reported this: reactivity marks the apply phase as a
  tracked scope, and its model has no state for "matches, but does not subscribe".

  solid/no-owned-primitives-in-ref reports createEffect, onCleanup and
  related primitives inside a ref callback. Solid 2.0 ref callbacks are unowned —
  getOwner() returns null — so the effect is never disposed and the cleanup
  never runs. Directive factories, which create these primitives while they still
  have an owner, are unaffected.

  solid/no-setter-in-effect now also reports an async apply phase that awaits and
  writes the result back into a signal. Its existing check requires every
  statement to be a setter call, which an await breaks.

  solid/prefer-class-object matched only the lowercase classlist. JSX prop
  names are case-sensitive and the Solid 1 prop is classList, so the migration
  rule missed every real call site.

  solid/imports now knows createLoadingBoundary, createErrorBoundary and
  createRevealOrder, so importing one from the wrong module is reported like
  every neighbouring primitive.

  solid/jsx-no-duplicate-props recommended classList, which Solid 2.0 removed
  and the neighbouring rule reports. It now points at the object and array forms
  of class.

@murky-web/typebuddy@1.4.0

Minor Changes

• 6c30131: Repair the oxlint autofixes and close the rules' blind spots.

  Two fixes were producing broken code. prefer-maybe-promise inserted
  return err() directly before a catch block's closing brace, so
  catch { log("boom") } became catch { log("boom") return err(); }, which does
  not parse. require-try-catch rebuilt the function body line by line and
  stripped leading whitespace from every line — including lines inside template
  literals, silently changing string values. It now wraps the body with two
  zero-width insertions and leaves it byte for byte intact.

  Six blind spots closed:

  • The catch branch settles on err(). It was fed to the same code path as the
    try block, so a catch return got two reports with opposite fixes (ok(value)
    and err()), and --fix appended an unreachable return err() after an
    existing return. Returns that hand back a value are now reported without a
    fix — rewriting them to err() would drop the author's fallback, and where
    the default belongs is the author's decision.
  • Returns nested inside an if, loop or switch in a try block are wrapped.
    Only the try block's direct children were, so try { if (flag) return "early"; return "late"; } wrapped just "late".
  • Promise<T> in a type-level signature is reported. TSDeclareFunction,
    TSFunctionType and TSMethodSignature were routed through a check for
    async, which a type signature can never carry, so all three visitors were
    unreachable. Ambient declare context stays exempt: rewriting a third-party
    callback shape would misdescribe that API.
  • MaybePromise and AsyncResult count as awaited return types. Matching only
    Promise meant a bare return; in a MaybePromise<void> function was left
    alone while its Promise<void> twin was fixed — the rule went blind on
    already-migrated code, and on the spelling the package recommends.
  • Expression-bodied async arrows are reported and fixed. async () => fetch(url)
    has no block for the try/catch to live in and escaped entirely.
  • null | undefined is left alone instead of being rewritten to the
    equivalent-but-longer Optional<null>.

  Also fixed: an inserted helper import landed glued to the following statement in
  any file that starts with a comment. Whether to append after an existing import
  was inferred from the anchor's byte offset, which only answers that question in
  a file whose first statement begins at byte zero.