HTTP Protocol for RDF Stores

Last week the W3C SPARQL Working Group has published a number last call working drafts for SPARQL 1.1. Much have been already said on various fora on the new features of SPARQL 1.1, like update, entailment regimes, property paths; I will not repeat here. But I think it is worthwhile calling attention on one of the documents that may not be seen as a “core” SPARQL query language document, namely the Graph Store HTTP Protocol.

Indeed, this document stands a little bit apart. Instead of adding to the query (and now also update) language, it concentrates on how the HTTP protocol should be used in conjunction with graph stores. I.e., what is the meaning of the well known HTTP verbs like PUT, GET, POST, or DELETE for graph stores, what should be the response codes, etc. It is important to emphasize that this HTTP behaviour is not bound to SPARQL endpoints; instead, it is valid for any Web sites that serve as a graph store. This could include, for example, a Web site simply storing a number of RDF graphs with minimal services to get or change the content of those. (In this respect, this document is closer to, e.g., the Atom Publishing Protocol which includes similar features for ATOM data, and which also plays an important role for technologies like, for example, OData.) Because such setups, i.e., “just” stores of RDF graphs without a SPARQL endpoint, are fairly frequent, it is important to have these HTTP details set. So… worth looking at this document and send feedbacks to the Working Group! (Use the public-sparql-dev@w3.org mailing list for comments.)