CRAN release: 2020-09-27
- don’t build vignette on machines that don’t have packages listed in Suggests (these packages must only be used conditionally)
CRAN release: 2020-07-08
Rewrote logic around how
as_xmlused to fill in a default value that pointed to a local copy of
eml.xsd. Now, it automatically fills in a web-resolvable location to make validating documents easier out of the box.
Re-worked the logic for handling
schemaLocationwhen serializing to XML:
schemaLocationis present on the
emldobject, it is used verbatim (no change in API here) regardless of the
schemaLocationis absent on the
as_xml(..., schemaLocation = TRUE)causes a value to be guessed.
as_xml(..., schemaLocation = FALSE)explicitly prevents a value from being filled in when serialized.
as_xml(..., schemaLocation = "Some value)explicitly sets the provided value.
emld::eml_validateno longer depends on
schemaLocationto determine the correct XSD to use during schema validation and now uses two helpers (See below) to find the correct schema file. See #52 & #45.
emld::eml_versionnow allows specifying the version without the
eml-prefix, like `eml_version(“2.1.1”), and will throw a warning when it gets output that doesn’t ’look right rather than silently failing.
Fixed a bug where the EML 2.1.1 units dictionary was being used for EML 2.2.0 docs which would cause spurious validation errors. See #56.
Developer-facing (non-exported) changes:
- Added two new helper methods:
find_real_root_name(doc : xml_document) : list(prefix : character, name: character)which returns the namespace prefix and the local name of the root element on an
guess_root_schema(doc : xml_document) : list(module : character, version : character, namespace : character)which returns the module, schema version, and namespace URI of the root element on an
schemaLocationis now ignored during roundtrip testing because of the new (above) behavior of
emldwith respect to
- Roundtrip testing can now handle documents that are supposed to be invalid but still roundtripped. Specify intentionally invalid files by adding “invalid” (case insensitive) to the filename in
- Minor tweaks to the README. Thanks @jeanetteclark
CRAN release: 2020-02-05
- Fixed serialization bug for
- Fixed validation bug: packageId is now used as an identifier for checking uniqueness, and no more errors for annotation elements in additionalMetadata. #49
- Fixed validation bug: XPath was referencing the element rather than the attribute
CRAN release: 2019-10-11
CRAN release: 2019-03-06
- Implemented changes requested by rOpenSci review, as detailed in #30