- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Mon, 6 Apr 2009 20:31:33 +0100
- To: Michael Grove <mike@clarkparsia.com>, W3C OWL Working Group <public-owl-wg@w3.org>
Hi Mike! Thanks for the review. I've quite radically updated the document: http://www.w3.org/2007/OWL/wiki/XML_Serialization It's still in a bit of flux as I chase down references and find typos. But I think this is close, in spirit, to the final form. On 6 Apr 2009, at 18:15, Ian Horrocks wrote: > Begin forwarded message: > >> >> ---------- Forwarded message ---------- >> From: Michael Grove <mike@clarkparsia.com> >> Date: Mon, Apr 6, 2009 at 12:58 PM >> Subject: OWL XML Serialization Review >> To: Kendall Clark <kendall@clarkparsia.com> >> >> >> I have just a few comments on the OWL2 XML serialization document >> that >> I wanted to pass along. I think as a vehicle for conveying the XML >> Schema to users, the document is perfectly reasonable. That's one intent. >> The schema >> seems good and everything lines up with the structural specification. >> >> However, I think if some of the intent of the document is as a >> reference for implementors, it needs some more content. It is intended for implementors. >> There's only >> one simple example, and the rest of the document is dominated by the >> schema itself. I think if there were a few more examples, especially >> some non-trivial ones, that would be very helpful. Generally, I don't think we do or should put a lot of examples fin this document. This example is merely intende to convey the flavor when reading causally. There will be examples in the Test Suite and there will be close feature exhaustive examples in the Primer. (And there is a pointer from this example to the primer.) Does that meet your need? >> Furthermore, >> there's a reference in the document that some things in the OWL2 spec >> cannot be captured in the XML schema, but there is not an example, or >> any clear explanation of why that is the case. Does that really matter? Even for an implementor? I mean, what difference does it make *why* such and such a constraint isn't in the schema as long as the *fact* that it is not in schema is document (and, of course, what the constraint *is*). I've tried to make the constraints clear: http://www.w3.org/2007/OWL/wiki/ XML_Serialization#The_Serialization_Syntax There is some advice: http://www.w3.org/2007/OWL/wiki/ XML_Serialization#OWL_2_XML_serialization_ontology_document >> That is useful >> information, Why? I mean, from an implementation perspective, what do you care *why* its not in the schema? How would you use it? (See below for examples.) >> and would have made the document more valuable, >> especially to someone trying to implement the spec. Again, I don't see the *rationale* as being helpful, per se, for *implementation*. >> Of course, if it's not intended as a guide or reference for >> implementors, I think its fine. But if people are supposed to refer >> to this while working on a parser, I think it falls short. Could you say what extra information you'd want? It'd be especially helpful if you'd explain why it would affect your implementation. For example, some groups are not turned into types since that would require multiple inheritance and XML Schema doesn't support MI. How does this help you, qua parser writer? Similarly: ""'On any element, one, and exactly one, of an IRI attribute or an abbreviatedIRI attribute must appear. This constraint is not expressed in the Schema for technical reasons.""" The technical reason is that XML Schema 1.0 does not support this sort of constraint on attributes (though XML Schema 1.1 will). (Such a constraint can easily be checked by program code, of course, or schematron.) I hope it's clear now what the extra constraints *are* (which seems to be to be the key point). Please let me know if any point there is unclear. Cheers, Bijan.
Received on Monday, 6 April 2009 19:27:47 UTC