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

@galaxy-foundry/foundry@0.1.0

Minor Changes

• #240 87f9a46 Thanks @jmchilton! - Add browser-safe ./meta subpath export (foundryCliMeta) so consumers can render per-subcommand documentation from static program metadata without invoking commander or shelling out to foundry --help. The bin is refactored: buildProgram() is now importable from @galaxy-foundry/foundry's internal program.ts, and the bin entry point is a thin parse-argv shim.

• #239 ceb66c9 Thanks @jmchilton! - Restructure publishable packages: introduce the unified foundry CLI bundling all validate-* subcommands plus a summarize-nextflow wrapper. The summarize-nextflow package now owns its own schema and self-validates without a foundry dependency. The four standalone schema packages (summary-nextflow-schema, summary-cwl-schema, galaxy-tool-discovery-schema, galaxy-tool-summary-schema, tests-format-schema) are folded into either summarize-nextflow (producer-co-located) or foundry (orphans).

Patch Changes

• Updated dependencies [ceb66c9]:
  • @galaxy-foundry/summarize-nextflow@0.1.0

@galaxy-foundry/note-schema@0.1.0

Minor Changes

• #449 1ead2a7 Thanks @jmchilton! - cli-command names the document a page summarizes source_url, constrained to a URL, and owes
  one exactly where the page really is a summary of something else.

  The field is the sibling Foundry's name and constraint for the same idea, so one spelling now
  means one thing across both instances rather than three fields sharing a stem and disagreeing.
  It is distinct from the schema kind's upstream, which stays: there the field records where a
  vendored artifact came from, and a cross-field rule keys off whether it points outside this
  repository. Summarizing an external command is not vendoring it.

  The accompanying validator rule replaces a blanket "must declare upstream" with the condition
  that actually holds. A command this repository implements has no second place for a reader to
  check, and eight pages pointed at one program.ts in this very tree; those now carry no
  source_url and are rejected if they grow one. Every other cli-command page must declare one.
  The split keys off foundryCliMeta — the same program metadata the corpus check already
  imports — rather than a name pattern, so a command added to our own CLI is classified by the
  thing that knows.

• #439 75b5c46 Thanks @jmchilton! - Stop erasing three shapes on the way out of the package: a Mold's typed references, a Mold's
  declared artifacts, and a pipeline's phase grammar. Adds KindReference, OutputArtifact,
  InputArtifact, PipelinePhase and PhaseBranchItem.

  The references and the phase grammar were declared z.ZodType<unknown> while the schemas
  beside them built the exact object; the artifact declarations were simply unexported. A
  consumer that wanted to RENDER any of them had to describe it again from the outside, and
  three did — a nine-field interface copied from the reference schema by reading it, two more
  copied from the mold schema, and a component prop typed any[] for phases.

  Every one of these is z.infer of its schema rather than hand-written, which is the part
  that matters. An interface beside a schema does not fail against it when the fields are
  optional: drop trigger from the reference object, or schema from the output artifact, and
  the output stays assignable to an interface that declares them optional — everything
  compiles, and a row quietly stops rendering. referenceShape is hoisted out of
  buildKindContext so the schema and the type are one description.

  branchItem also loses a z.lazy wrapper that was never doing anything: the union does not
  refer to itself, so there was nothing to defer.

  The reference vocabularies stay string rather than the contract's enums. They come from
  reference_contract.yml at runtime, and a compile-time literal union would be a second,
  staler copy of a registry that is already the authority.

  Not breaking: unknown accepted everything, so every existing annotation still holds. A
  consumer that was asserting any of these shapes can drop the assertion.

• #387 a10e1d6 Thanks @jmchilton! - Consume the shared license table from @galaxy-foundry/license-policy instead of loading a
  repo-local license-policy.yml.

  Breaking for importers. The barrel no longer re-exports loadLicensePolicy,
  findLicensePolicyPath, licenseIds, isValidLicenseId, resolveLicenseRow,
  LICENSE_POLICY_FILE, LICENSE_REF_RE, LicenseRow, CastMode, or
  RedistributionPolicy. Import them from @galaxy-foundry/license-policy directly. Only
  the LicensePolicy type is still re-exported, since callers must name it to build the
  buildNoteSchema options object.

  buildNoteSchema is unchanged — it still takes a licensePolicy. Callers that used to
  pass loadLicensePolicy(repoRoot) now pass bundledPolicy().

• #389 bf39a96 Thanks @jmchilton! - Consume the meta_tags.yml format from @galaxy-foundry/tag-registry instead of a
  repo-local loader. src/tags.ts is deleted; the facet vocabulary stays at our repo root,
  because that package deliberately ships none — facets are the browse axes of one domain.

  Breaking for importers. The barrel no longer re-exports loadTagRegistry,
  tagRegistry, buildTagIndex, TagRegistryFile, Facet, or FacetInfo. Import them
  from @galaxy-foundry/tag-registry directly. Only the TagRegistry type is still
  re-exported, since callers must name it to build the kind-context options — the same
  arrangement as LicensePolicy.

  buildNoteSchema and buildKindContext are unchanged: both still take a tags registry
  of the same shape.

  The package validates what neither loader did. loadTagRegistry now refuses a registry
  with no facets block, a facet missing label/description, a tag with no gloss, or a
  tag two facets both declare — naming the file. Previously these arrived as an undefined
  somewhere downstream.

Patch Changes

• #453 e9bd4b6 Thanks @jmchilton! - Take @galaxy-foundry/license-policy 0.4, which renames licenseIdFromFilePath to
  licenseFileIdFromPath and LicenseFile.licenseId to LicenseFile.id.

  Nothing this package imports was renamed — bundledPolicy, isValidLicenseId and
  resolveLicenseRow are unchanged, and the only type re-exported from the barrel is still
  LicensePolicy. The bump is here because a caret range on a 0.x version pins the minor, so an
  importer resolving both packages would otherwise be held at 0.3.

@galaxy-foundry/summarize-nextflow@0.1.0

Minor Changes

• #239 ceb66c9 Thanks @jmchilton! - Restructure publishable packages: introduce the unified foundry CLI bundling all validate-* subcommands plus a summarize-nextflow wrapper. The summarize-nextflow package now owns its own schema and self-validates without a foundry dependency. The four standalone schema packages (summary-nextflow-schema, summary-cwl-schema, galaxy-tool-discovery-schema, galaxy-tool-summary-schema, tests-format-schema) are folded into either summarize-nextflow (producer-co-located) or foundry (orphans).

foundry-site@0.0.2

Patch Changes

• Updated dependencies [1ead2a7, 87f9a46, e9bd4b6, 75b5c46, a10e1d6, bf39a96, ceb66c9]:
  • @galaxy-foundry/note-schema@0.1.0
  • @galaxy-foundry/foundry@0.1.0