Skip to content

v0.29.4

Choose a tag to compare

@Goldziher Goldziher released this 27 Jun 10:04
v0.29.4
c89fc17

Changed

  • tooling: extend the no-project-special-casing pre-commit hook to reject the xberg and
    crawlberg downstream product names (case-insensitive, including camelCase and separator
    variants), and consolidate the brand allowlist so the xberg-io org namespace and the xberg.io
    domain stay permitted while xberg-io/xberg and bare xberg mentions are still caught. Neutralize
    the xberg-named Java/enum test fixtures to generic sample names.

Fixed

  • e2e: keep public Ruby and Elixir test calls on configured method names and
    resolve $mock_url placeholders inside typed JSON-array arguments across
    generated language e2e suites.

  • e2e: resolve $mock_url placeholders for Ruby object arrays, Elixir typed
    object arguments, and Kotlin/PHP typed object setup while allowing Elixir e2e
    calls to target keyword-opts public facades.

  • e2e: avoid Elixir typed-object variable collisions and align Kotlin typed
    object mock URL fallbacks with the generated mock-server harness.

  • node: remove downstream internal DTO names from generated trait-bridge
    return-value comments.

  • ffi: honor [crates.ffi].exclude_types when generating cbindgen.toml.
    Excluded Rust-only helper DTOs are now omitted from the header prelude forward
    declarations and emitted in [export].exclude, keeping C and cgo headers from
    leaking types that the FFI layer does not expose.

  • java/kotlin-android: route configured trait-bridge lifecycle functions through the generated
    bridge APIs instead of also emitting ordinary FFI wrappers. This keeps raw Rust functions such as
    register_document_extractor from shadowing typed host interfaces (IDocumentExtractor,
    IRenderer) with dangling DocumentExtractor/Renderer parameter types or JSON-string JNI
    declarations.