It looks like this is mainly a description of how to implement a service that does what I think could be very important things: • Providing canonical stable IDs for bnodes in the context of a specific dataset. Personally, bnodes that can't be round-tripped irritate me no end. • Setting the stage for treating datasets the way git treats code, with SHA hashes acting as fundamental identifiers, and providing a basis for an algebra that can translate any dataset into any other dataset through a series of diff-patches. But the intended audience seems to be implementers. Looks like there's an implementation in js: https://github.com/digitalbazaar/rdf-canonize Python: https://github.com/amagioss/py_rdfc10 Rust: https://github.com/zkp-ld/rdf-canon I'm not seeing anything on the JVM.
The test suite seems to mention two java implementations: Corese and RDF-URDNA. https://w3c.github.io/rdf-canon/reports/#test-subjects It also looks like JENA https://github.com/apache/jena/issues/2105.