banana-rdf emphasizes type-safety and immutability, so it can come
with some cost when the underlying implementation is very mutable (I'm
looking at you, Jena and Sesame). We try to keep a clear distinction
between the core concepts and the enhanced syntax that Scala can give
us.
RDF
itself is defined as a record of types. Implementations just have to
plug their own types. And because types alone are not enough, we
introduce the
RDFOps
typeclass, which defines the mandatory operations that an RDF
implementation must
implement. SparqlOps
does the same for SPARQL.
With banana-rdf, you get Diesel, a nice DSL to build and navigate
within pointed graphs (graphs with a pointer to an inner
node). You also get an abstraction for graph stores
(GraphStore), which do not have to be SPARQL engines
(SparqlEngine). Of course, you can serialize and deserialize
most of the RDF syntaxes as well as JSON-LD (RDFa will come soon).
banana-rdf introduces the concept of binders, which let you
bridge the Scala and RDF worlds. Most of the common datastructures are
already available, and you can even map your own classes. Unlike usual
ORM techniques, this does not rely on annotation or reflection.