- From: Mark Baker <distobj@acm.org>
- Date: Mon, 21 Jan 2002 09:21:37 -0500 (EST)
- To: davidc@nag.co.uk (David Carlisle)
- Cc: xsl-editors@w3.org
David, > > Using the rules defined for the XHTML media type[1], this document is > > an XHTML document because the root namespace is the XHTML namespace. > > If delivered as either application/xml or text/xml, it might > > reasonably > > be expected (see RFC 3023, Section 3, last paragraph) to process it as > > XHTML, not as XSLT. > > I think that should include, in this case, processing <xsl:value-of . That's a possibility, as this use of XSLT predated the XHTML media type draft. But do you think it's a good general rule? There'd be all kinds of exceptions to this rule over time. We'd have to go back, as new exceptions are added, and revise the media type draft. > The modularisation of XHTML can only succeed if an agent which is > processing the XHTML is prepared to hand over control for elements from > another namespace. Exactly right. BTW, I note that the XSLT simplified form isn't limited to just XHTML. Any document type could be used, so this isn't an XHTML specific issue. But the issue you describe holds for all XML languages. FWIW, I'm currently working on an I-D for standardizing this behaviour with respect to namespace dispatch and containment rules. > If a document with top level element being xhtml's > html contans svg elements in the svg namespace and mathml elements in > the mathml namepace then these will (probably) have to be processed (via > some mechanism) by some other process and the resulting rendering passed > back to the html agent for incorporation in the complete document. Yes, absolutely. > I > don't see that handling xsl instructions is radically different from > this. Yes, it isn't *radically* different in that example I provided, but there are two important differences; xsl:value-of can be embedded seemingly anywhere, and XHTML processors don't know to dispatch to alternate processors. Even if the latter were not the case, alternate namespaces should only be used within elements that can support it, otherwise a specific compound media type should be used, in this case maybe application/xhtml+smil. This is all "IMHO" of course, though I've learned a lot from compound document systems in the past (Opendoc and Java Beans, primarily), and I believe the lessons learned about containment rules there should be used here. > Of course if the agent receiving the document can't handle xsl then > tough, but the same is true of sending an xhtml+mathml document to > a non mathml-enabled browser. Right. XHTML doesn't support (except to ignore) non-XHTML namespaces within its documents, so a compound media type is required in all cases. MB -- Mark Baker, Chief Science Officer, Planetfred, Inc. Ottawa, Ontario, CANADA. mbaker@planetfred.com http://www.markbaker.ca http://www.planetfred.com
Received on Monday, 21 January 2002 14:42:38 UTC