- From: Dominique Hazaël-Massieux <dom@w3.org>
- Date: Mon, 16 Feb 2004 11:30:56 +0100
- To: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Cc: public-rdf-in-xhtml-tf@w3.org
- Message-Id: <1076927455.30689.26.camel@stratustier>
Le ven 13/02/2004 à 18:01, Jeremy Carroll a écrit : > I am not very up on GRDDL but get the impression that you are not using the > stylesheet PI, despite it having been explicit created to give an XML > mechanism corresponding to the HTML link attribute .... It is in fact a mechanism corresponding to the HTML link element, when used in a context of a style sheet, not the general recommended linking mechanism; in particular, it seems suited for browsers to display (or read vocally, ...) a generic XML document, not to help interpreting its meaning. The generic XML linking mechanism is XLink; it might be a good idea to try to rely on XLink for GRDDL, but the currently poor support of XLink in XML tools and the very basic types of linking needed for GRDDL doesn't add up very much in favor of that. > As is, any XML format wanting to use the grddl attribute has to explicitly > permit it, whereas PIs can be added to any format. IMHO, PI are really a not very good way to store information in the XML infoset, given that they cannot be parsed generally speaking (ie, they don't have a notion of attribute, sub-elements, etc.); as far as I remember, James Clark regrets nowadays to have used processing instructions for the styling mechanism in XML, and many people argue in favor of deprecating them in an upcoming version of XML; in short, I don't think it would be reasonable to build an XML application targeted for wide interoperability based on processing instructions (again, that's only my opinion) I agree that the fact that GRDDL creates a new attribute in a new namespace using a new mechanism is suboptimal; but on the bright side, the namespace-based mechanism [1] is much more powerful and relies on the widely-deployed and understood namespace mechanism. Maybe only the latter will get implemented and deployed; but document-based semantics still seems like an interesting application of GRDDL, I would think. 1. http://www.w3.org/2004/01/rdxh/spec#ns-bind Dom -- Dominique Hazaël-Massieux - http://www.w3.org/People/Dom/ W3C/ERCIM mailto:dom@w3.org
Received on Monday, 16 February 2004 05:30:58 UTC