- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Fri, 20 Apr 2007 10:36:21 +0100
- To: Shane McCarron <shane@aptest.com>
- CC: Ben Adida <ben@adida.net>, Fabien Gandon <Fabien.Gandon@sophia.inria.fr>, GRDDL Working Group <public-grddl-wg@w3.org>, public-rdf-in-xhtml-tf@w3.org
I can see that leads us essentially to paradox. (of course the set of valid xhtml is also infinite). The spec is written only for valid xhtml, and that seems appropriate, but the expectation is that implementations do the right thing with invalid xhtml. My own implementation even runs a tidy algorithm on stuff that either has an html mimetype or a root element of html Some of the features we wish to test do require grddl attributes on the root element, which leads us to either: a) tests with invalid input or b) some kind of DTD hackery To date the WG has been following (a). We're hoping to move test cases to last call very soon, and if anyone feels strongly enough about this topic, it would be an appropriate area for a last call comment. Jeremy Shane McCarron wrote: > > > > Jeremy Carroll wrote: >> >> >> From a GRDDL perspective, there is an expectation of handling invalid >> XHTML (like other user-agents) > I guess. And from that perspective, you really should be testing both. > Although, by definition, the collection of invalid XHTML is effectively > infinite. > -- Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Friday, 20 April 2007 09:36:52 UTC