- From: Steven Pemberton <steven.pemberton@cwi.nl>
- Date: Fri, 25 Jan 2002 14:29:09 +0100
- To: <www-tag@w3.org>
- Cc: "HTML WG" <w3c-html-wg@w3.org>
I think it is a proof of the worth of the TAG that the media types issue is finally being taken seriously in W3C. With XHTML as The Great Integrator, the HTML Working Group has been fighting this one since at least 1999. Our opinion then was that media types were Not Very Useful for heterogeneous documents, and that application/xml would have to do, with namespaces doing the rest. But the message from Norman Walsh has truly warmed my heart:: > From: Norman Walsh <Norman.Walsh@Sun.COM> > In general, is there really any value in declaring specific media > types for XML vocabularies? > > Imagine that I've got text/foo+xml and text/bar+xml. If I send a > document that's just 'foo' or just 'bar', those may have value. But as > soon as I start mixing foo and bar together, I don't see that there's > any right answer as to what media type I should use. Precisely what we were saying three years ago! But why am I so pleased? Because it was exactly Sun's outspoken and impassioned attack on our position (http://lists.w3.org/Archives/Member/w3c-ac-forum/1999JulSep/0020) that forced us to drop all mention of media types from XHTML 1.0, and to ask the XML community to please solve the problem. This involved the setting up of the ietf-xml-mime list to discuss the issues, which ended in the creation of RFC 3023 (XML Media Types). That RFC was not the perfect answer in our eyes, but it at least *allows* you to use application/xml and namespaces, so we felt vindicated a little. If we can now finally agree that XML documents can be classified with application/xml and let the namespaces do the rest, the HTML WG will be more than delighted to go back and sweep up the mess! Steven Pemberton Chair, HTML Working Group PS. If this is the start of a trend, let me predict that some of the other architectural issues that the HTML WG has been fighting for years will re-emerge here, such as identifying profiles with namespaces and opaque linking. I hope so!
Received on Friday, 25 January 2002 08:28:32 UTC