- From: Robin Berjon <robin.berjon@expway.fr>
- Date: Wed, 16 Feb 2005 17:42:51 +0100
- To: Dare Obasanjo <dareo@microsoft.com>
- Cc: www-tag@w3.org
Dare Obasanjo wrote: >>> No, they won't. At least not if you are using MSXML or System.Xml in the >>> .NET Framework. The same problem exists with xml:base today. In both >>> libraries, the assumption we made was that the XML namespace would be >>> unchanging. > >> Based on what grounds did you decide to make such a bold assumption? > > [Dare Obasanjo] That decision was made before my time but given the > fact that this thread started because of similar discussions around > other specifications I don't think it is as unreasonable as you claim > to think that the number of names within a namespace will be > unchanging. It's one thing to believe that a rule applies in vocabulary design (which is what sparked this thread) and another to imagine a restriction to be in the specification that's simply not there when implementing an XML framework that is supposedly generic. It's an obvious request for trouble. >>> For this reason, we don't allow users to specify a schema >>> for the XML namespace but instead always use an internal schema with a >>> fixed list of attribute declarations {xml:lang, xml:space}. > > Is there anything in the XML Schema spec that makes this behaviour > conformant? > > [Dare Obasanjo] Yes. Schema locations are hints not directives. A XML > Schema validator can ignore locations and use schemas it already knows > about. Oh right sorry, I'd forgotten that XML Schema was thoroughly broken in its specifying of xs:import. Still, it seems like a strange reason to expect the xml:* namespace to be fixed. The whole point of reserving names in the spec is so that future specs can add new ones! -- Robin Berjon Research Scientist Expway, http://expway.com/
Received on Wednesday, 16 February 2005 16:42:45 UTC