Changelog for Spira https://github.com/rdf-ruby/spira
- Spira#using_repository now returns the given repository
- General updates to bring up to date.
- Implemented #validate, #validate! (refactored from #save!)
- Force to_a on query results when constructing to force the promise-like semantics of SPARQL::Client
- Bumped the version dependency on rdf-isomorphic to 0.3.0
- Added support for before_create, after_create, before_save, after_save, after_update, before_destroy and after_destroy hooks.
- Switch RDF.rb dependency to >= instead of ~>
- Use RDF::URI.intern on URIs generated via base URIs
- Added a Spira::Types::Native, which will return the RDF::Value for a given predicate directly without any serialization or dserialization.
- Fix a bug with Spira::Types::Any that prevented blank node objects
- Added Spira::Resource#copy, #copy!, #copy_resource!, and #rename!
- Fix a bug with fragment RDF::URI arguments that prevented correct URI construction
- Added Spira::Resource.project(subject, attributes, &block), which creates a new instance without attempting to perform any base_uri logic on the given subject. This provides a supported API entry point for implementors to create their own domain-specific URI construction.
- Updating a value to nil will now remove it from the repository on #save!
- Tweaks to dirty tracking to correctly catch both changed and updated values. All tests pass for the first time with this change.
- Change gemspec name to work with bundler
- Remove type checking for repository addition. More power in return for slightly more difficult error messages.
- Repositories added via klass, *arg are now only instantiated on first use instead of immediately.
- RDF::URI#as, RDF::Node#as, Resource.for, and Resource#new can now all accept a block, which yields the new instance and saves it after the block.
- Clarify error message when default repository is not setup
- Added a weak-reference identity map for each instance. Any circular references in relations will now return the original object instead of querying for a new one.
- Use a weak-reference identity map when iterating by class.
- When serializing/unserializing, duck typing (:serialize, :unserialize) is now permitted.
- Added Resource.[], an alias for Resource.for
- Resource.each now correctly raises an exception when a repository isn't found
- Added #exists?, which returns a boolean if an instance exists in the backing store.
- Added #data, which returns an enumerator of all RDF data for a subject, not just model data.
- #save! and #update! now return self for chaining
- Implemented #update and #update!, which allow setting multiple properties at once
- Existing values not matching a model's defined type will now be deleted on #save!
- Saving resources will now only update dirty fields
- Saving resources now removes all existing triples for a given predicate if the field was updated instead of only removing one.
- Implemented and documented #destroy!, #destroy!(:subject), #destroy!(:object), and #destroy!(:completely). Removed #destroy_resource!
- has_many collections are now Sets and not Arrays, more accurately reflecting RDF semantics.
- The Any (default) property type will now work fine with URIs
- Added ResourceDeclarationError to replace various errors that occur during invalid class declarations via the DSL.
- Raise an error if a non-URI predicate is given in the DSL
- Small updates for RDF.rb 0.2.0
- Implemented dirty field tracking. Resource#dirty?(:name) will now report if a field has not been saved.
- Relations can now find related classes in modules, either by absolute reference, or by class name if they are in the same namespace.
- Fix a bug with default_vocabulary in which a '/' was appended to vocabularies ending in '#'
- Fix a bug with the Decimal type where round-tripping was incorrect
- Fix some error messages that were missing closing parentheses
- Added a Decimal type
- Small updates for RDF.rb 0.2.0 compatibility
- Add a Spira::Base class that can be inherited from for users who prefer to inherit rather than include.
- Resource#new returns to the public API as a way to create a resource with a new blank node subject.
- Bumped promise dependency to 0.1.1 to fix a Ruby 1.9 warning
- Rework error handling when a repository is not configured; this should always now raise a Spira::NoRepositoryError regardless of what operation was attempted, and the error message was improved as well.
- A '/' is no longer appended to base URIs ending with a '#'
- Resources can now take a BNode as a subject. Implemented #node?, #uri, #to_uri, #to_node, and #to_subject in support of this; see the yardocs for exact semantics. RDF::Node is monkey patched with #as, just like RDF::URI, for instantiation. Old code should not break, but if you want to add BNodes, you may be using #uri where you want to now be using #subject.
- Implemented #each on resource classes, allowing classes with a defined RDF type to be enumerated
- Fragment URIs are now used as strings, allowing i.e. Integers to be used as the final portion of a URI for classes with a base_uri defined.
- Added an RDF::URI property type
- Implemented #to_rdf and #to_uri for increased compatibility with the RDF.rb ecosystem
- Initial release