This release delivers several important fixes and enhancements to RDFLib’s Dataset implementation, resolving long-standing issues and improving consistency across serialization and SPARQL operations. It also introduces new deprecation notices for certain Dataset methods and attributes, which will be removed in the next major release. In addition, this version includes a range of improvements to SPARQL result parsing, typing, and literal handling.
Features
- Added
Dataset.__iadd__support - Dataset's default serialize format is now
trig - Datasets can now add graphs backed by different stores
Fixes and Improvements
- Fixed an issue where the
RecursiveSerializerwould output undeclared prefixes for predicates that contained the base as a substring - Prevented prefix generation for predicates corresponding to the base namespace
- SPARQL Update now correctly inserts into the default graph
- Dataset.parse now returns Self
- N-Quads serialization no longer includes the RDFLib internal default graph identifier
- Static type checkers can now infer the type of
Term.__new__ - Removed automatic date conversion for gYear and gYearMonth literals
- Optional clauses in SPARQL queries can now bind variables
- Fixed reevaluation logic in SPARQL Update between update loops
Maintenance
- Added deprecation notices to certain Dataset methods and attributes
- Use Dataset.graphs instead of Dataset.contexts method
- Use Dataset.default_graph instead of Dataset.default_context
- Deprecate Dataset.identifier entirely.
- Updated type hints for Graph.open() with SPARQLUpdateStore configuration
- SPARQL Result Parsing Improvements
- Simplified and modernized the SPARQL result parsing system:
- These changes maintain backward compatibility while making the SPARQL API more flexible and extensible.
See the CHANGELOG.md for more details.
Full Changelog: 7.2.1...7.3.0