- From: Sandro Hawke <sandro@w3.org>
- Date: Mon, 21 May 2007 22:48:07 -0400
- To: public-rif-wg@w3.org
There are several questions in-line in http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax that I think are best answered in e-mail. (That page has already gotten to be very hard to read, with comments from four people.) I've also tried to clean up the page a bit. * There was a link to "http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax#preview" which doesn't make sense. I removed it. * Adrian writes: We do need to choose property names that indicate relationships between classes, rather than just repeat the class name. E.g. "Person has a companion that is a Dog" will look better when serialized than "Person has a dog that is a Dog". I agree, but that's a comment on the abstract syntax, not the XML syntax, I think. * I clarified 1.2 and 1.3, and removed comments from Adrian and Gary, because I think my re-writes address them (or at least change what they need to say). * Gary writes: "name v. ref makes it clear whether you are "defining" a new Dog or referencing a Dog" Yeah, RDF and OWL intentionally don't do that -- they see no difference between a "definition" and a "reference" -- in either case you're just providing some more information about something. For RIF, I think it's fine to say the first occurance in a serialization provides all the properties. * in 3.4 -- I'm just talking about re-using the b-node syntax hack, to fit "local names" where the parser is expecting a URI. I should re-write that section to explain that, somehow. (or someone can beat me to it....) * in 3.5 <Dog id="ns:taiko"> is considered by many to be a bad design because it combines the xml namespace prefixes with the URI scheme names, with possibly confusing results. Someday, the IETF might define a URI scheme called "ns", and then the software which used xmlns:ns wouldn't be able to use that scheme. More likely, some vendor might start to deploy "xml" URIs which would cause unexpected conflicts with the reserved xml namespace prefix in some implementations. Other people think this is a paranoid fantasy and that we shouldn't worry about it. (Or were you thinking the id would ONLY be a qname, and never a raw URI? If so, that doesn't work because some URIs (like ones that end in "/") cannot be written as qnames.) * I haven't studied Gary's semi-stripe-skipping design enough to have an opinion or comments on it yet, sorry! -- Sandro
Received on Tuesday, 22 May 2007 02:48:11 UTC