And yes, the SPIN spec is of course platform neutral and implementations for Sesame (and other engines) would be easy. In fact we are talking to a large database vendor to add native SPIN support on the server side.

Thanks for the feedback.

I hadn’t come across that feature of SPIN. Its an interesting extension, but I wonder whether it would be classified as an extension of the query language or of the SPARQL processor? Its a fine line as you could probably argue that Property Functions are processor features rather than language extensions per se.

btw, as I understand it both ARQ and Sesame allow run-time extension of the SPARQL engine, e.g. by registering new functions and/or dynamic class loading. This is not quite the same as SPIN but it would be possible to implement the same kind of behaviour based on that underlying capability I think.

I also think that functions and magic properties are a good place to start. However, there is another category of extensions missing, which is “meta-extensions”. In particular I’d like to make you aware of the SPIN framework’s features to define new functions and magic properties so that they can be dynamically discovered on the Semantic Web. Each function (or (as of the upcoming SPIN 1.1) magic property) has a URI that can be dereferenced to get a complete executable description of the function’s arguments as well as a nested body query (typically another SELECT query) or embedded JavaScript code. With such a meta-extension, users are able to add any number of new extensions, and the language designers don’t even need to do anything else. See http://spinrdf.org