Version Packages

[gtbuchanan]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and publish to npm yourself or setup this action to publish 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

@gtbuchanan/cli@0.7.0

Minor Changes

• 6b1453b: Verify that published packages declare no private workspace dependencies

  gtb verify (scope manifest) now asserts that every install-time
  workspace: dependency of a published package is itself published. pnpm
  rewrites the specifier to the linked package's concrete version at pack
  time without checking that the version ever reached the registry, so a
  private workspace package ships in the tarball as a dependency no
  consumer can resolve.

  Covers dependencies, peerDependencies, and optionalDependencies.
  devDependencies are exempt (publishing strips them) as is anything
  bundleDependencies actually bundles — an explicit name list covers any
  field, while true covers dependencies alone, so a workspace peer or
  optional dependency stays checked.

Patch Changes

• c897a61: Make gtb publish resilient to partial release failures

  A release run no longer aborts on the first problem it meets. Failures are
  collected and re-thrown together, so one package or channel can't strand the
  rest, and the run reports everything that went wrong instead of only the first
  thing.

  The skip-if-exists check now reads a single gh release list rather than a
  gh release view per package, and a failed listing raises instead of being
  read as "nothing is released" — which previously sent every package on to a
  create that could only fail. A create GitHub rejects because the tag is
  already taken now counts as released, covering both a race with the listing
  and a tag reserved by a deleted immutable release.

@gtbuchanan/vitest-config@0.2.0

Minor Changes

• 59b6616: Add a testTimeout option to configure, configurePackage, and
  configureGlobal

  Raise it for a package whose tests do seconds of real work — building an
  ESLint config, starting a TypeScript project service, spawning a child
  process. On a machine running the rest of the build in parallel, that
  work exceeds vitest's 5s default and the suite fails under load while
  passing in isolation.

  The trade is real: this is a duration bound, so a higher one catches a
  performance regression later. It buys reliability, because a wall-clock
  limit can't distinguish a slower test from a busier machine. Size it to
  the worst contention the suite runs under, not to the test's own cost.

  testTimeout was already accepted by the e2e entry points; this makes
  it consistent for source tests.

  Fixes testTimeout: 0 on the e2e entry points, where vitest's
  documented way to disable the limit was dropped by a truthiness check
  and silently fell back to the default. hookTimeout derives from it
  there, so both were affected.