Release 3.1.11

• Favor absolute URIs for term attributes and properties, but allow pnames transitionally.
• Update local vocabularies to use absolute URIs in property values and keys.
• Change Term#properties to use an empty hash with a default_proc which does the magical lookup from attributes.
• Add noDoc option to Vocabulary writer to eliminate the Yard documentation.

Release 3.1.10

• Add Term#terms and Statement#terms which returns an array including itself, and array including the non-nil non-statement terms, respectively.
• Rename #has_foo? methods as simply #foo? to go with Ruby style. Old accessors remain as aliases. In some cases (statement?, term?) these were already in use, so now are overloaded.

Release 3.1.9

• Use an RDF::OrderedRepo in CLI if the ordered option is provided and the gem is loaded; otherwise, use an extended Array.
• Include shacl and shex gems in CLI formats tried.

Release 3.1.8

• Update RDF* mode:
  • Eliminate PG mode, rdfstar is not boolean valued.
  • Change CLI from --rdf-star to --rdfstar.
  • Update RDF* spec reference.
• Allow the CLI repository to be set per command. (Required for N3 reasoning).
• Skip some exponentiation tests in JRuby.
• Enable CI on all ruby versions on GitHub Actions.
• Create RDF.config for configuring cache sizes. Allows a general size, URI and Node specific cache sizes to be configured at startup time. Fixes #256.
• Add wrap_transaction keyword parameter to RDF::List#initialize. For #417.
• Update faraday and faraday_middleware to ~>1.x.

Release 3.1.7

• Fix decimal and double #round method to round towards positive for negative numbers, as called for in XPath function.
• Update the public domain dedication information to indicate that it applies to all repositories in the organization, and requested on a PR.
• Update URI#relativize based on JSON-LD Context#remove_base. This covers more cases where the base URI can be used.
• Fix the graph passed to new terms
  • It always passed the RDF::Vocabulary class instead of the actual vocab

3.1.6

For RDF::List:

• fix graph used to insert values into a list when not provided.
• use self.class, rather than RDF::List to allow it to be more easily subclassed. (for N3)

For RDF::Query:

• In Query#optimize! partition required from optional patterns so that optional patterns always come last.
• Add Variable#to_base as an alias of #to_s.
• Deprecate Query::Pattern#variable_names.
• Add #var_values and Query::Pattern#var_values to return the appropriate value from a statement (including embedding) associated with that value.
  This abstracts pattern matching to allow for embedded statements, and other things that can be used in a pattern (e.g., lists).

For RDF::Literal:

• Update RDF::Literal::Decimal#round, #floor, #ceil to return integers along with RDF::Literal::Double variations. Previously, they returned the same type.
• Add RDF::Literal::Numberic#** and RDF::Literal::Numeric#% based on XPath definitions.
• Add native support for Rational to xsd:double. Update Literal::Numeric#** to use the type of the native result, rather than always double.

Other:

• Fix bug in vocabulary writer with %() strings.

Release 3.1.5

• Without linkeddata loading, attempt to manually load all other serialization gems.
• Require 'rdf/cli' when used, rather than depend on autoload.
• Set the result encoding of CGI.encode to UTF-8.
• Add --ordered option to CGI to use an extended array instead of a repository to preserve order of triples.
• Use RDF::URI.parse instead of ::URI.parse when using NetHttpAdapter.

Release 3.1.4

• Explicitly require 'strscan' in ntriples/reader.
• Set log_statistics after level has been assigned.
• Use CGI.unescape instead of the deprecated URI.decode. Fixes #416.

Release 3.1.3.

add RDF::Vocabulary.register and make appropriate changes.

This is necessary, as simply defining a vocabulary is no longer sufficient to have it register a prefix for the vocabulary.

Release 3.1.2

Important new features include provisional support for RDF* and updates to query optimization.

• Cleanup etc/doap and references for developers.
  • Now uses https://unlicense.org/1.0/ for dap:license.
• RDF::Vocabulary
  • Update Vocabulary writer to output vocabulary classes that better document the vocabulary and reduce Yardoc errors.
  • Don't try to document vocabulary terms which aren't valid as Ruby attribute names.
  • Optimize Vocabulary.expand_pname to use Vocabulary.vocab_map, to look up terms by their standard prefix.
    • Vocabulary.vocap_map uses re-wored RDF::VOCABS, and is overrided by rdf-vocab, if it is included.
    • Also added Vocabulary.from_sym, which will return the vocabulary associated with the class_name symbol (e.g., RDF::Vocabulary.from_sym(:OWL) => RDF::OWL.
    • Adds RDF::Vocabulary.limit_vocabs which is used to limit the vocabularies returned by RDF::Vocabulary.each and can substantially reduce load time when the rdf-vocabs gem is used.
• Core Model
  • Make RDF::Statement an RDF::Resource, which allows a statement to appear as either the subject or object of another statement.
  • Update N-Triples and N-Quads readers and writers to accept and emit such statements; for the reader this requires using the rdfstar option, with value of either :PR or :SA.
  • An RDF::Enumerable may express support for by responding with true to #supports?(:rdfstar). The memory-based RDF::Repository supports RDF*.
  • Add Statement#embedded? to discriminate statements including statements.
  • Check for support for rdfstar before inserting embedded statements.
• RDF::Query
  • Support query patterns with recursive sub-patterns for subject or object.
    • Variables include those from sub-patterns.
    • Added Pattern#eql? which does recursive checks on sub-patterns.
    • Solution binding also binds to sub-patterns.
    • In Repository#query_pattern, consider recursive patterns for subject or object to match against recursive statements for subject or object.
    • Merging solutions also adds solutions found by binding recursive patterns to statements.
      ( This supports the SPARQL BIND(<<?a ?b ?c>> AS ?t) pattern).
  • Order Query::Solutions#variable_names based on Query::Solutions#project.
  • Binding to a variable now requires a solution (term is deprecated). This makes binding consistent with other uses.
  • In Solution#merge! don't merge a pattern over a statement. (BIND embedded triple use case).
  • Add :optimize option to Query.execute.
  • Remember graph_name when dupping a Query.
  • Invert costs in Pattern, as it ends up that favoring objects before subjects is much more expensive.