- From: Bijan Parsia <bparsia@cs.manchester.ac.uk>
- Date: Mon, 13 Apr 2009 21:54:36 +0100
- To: RDF Data Access Working Group <public-rdf-dawg@w3.org>
On 13 Apr 2009, at 21:35, Bijan Parsia wrote: > On 13 Apr 2009, at 21:03, Seaborne, Andy wrote: > [snip] >> I am expecting it to capture the AST for a query and it would be >> useful to more formally define the abstract syntax. > > There has a been a lot of positive feedback from OWL implementors on > the use of UML to define the abstract syntax...I'm not a huge fan, > myself. Sorry. I'm a huge fan of the precise definition of the language in an abstract form. *UML*, itself, gives me some heebie jeebies, but the subset we're using seems absolutely fine. I prefer XML for this sort of conceptual modeling (or something close to XML) because it gets you closer to multilanguage APIs. There can be problems due to limitations and infelicities in the XML abstract model (e.g., XML Schema and WSDL define a distinct "component model" to define their languages because the XML infoset doesn't deal with things like inclusion...HTML5 style definition seem to handle this with api extensions; so, for example; one could define an OWL ontology with imports by adding a method to an Ontology DOM element that gave access to the directly imported Ontology DOM elements). It's often hard to tell what's "mere" fashion and what's sense. These things seem useful: 1) the description of the language suggests an API for manipulating that language; 2) the description of the language abstracts away from many lexical details; 3) the description is concrete enough to be "directly" implemented 1 gives you 3 in a way that respects 2. Anyway, I don't think there's an intention to rewrite the main SPARQL spec (is there?) so this is more on the supplementary side of things. Cheers, Bijan.
Received on Monday, 13 April 2009 20:55:35 UTC