Sorry, don't really understand what you mean by your statement below. The XSL document may not be readable but the XML can be set to be as readable and descriptive as we want it to?
XML has no semantics. Your custom markup won't be compatible with anything. Without XSLT transformation to [X]HTML it is meaningless to any software.
XSLT is not supported by web spiders and by many browsers.
AFAIK using XSLT on client-side limits you to IE6 and Gecko browsers *only*.
Everyone else gets unusable page.
XSLT is a great technology, and there is one simple step to make it absolutelty compatible - transform on the server-side. If you output [X]HTML you're going to be compatible with almost every single web client out there.
I think that XSLT + XML won't save much (if anything) compared to resulting XHTML.
In terms of bandwidth: XML already adds markup to your data. XSLT is very verbose, and stylesheet will probably take as much as extra data it could generate.
Without HTTP pipelining (in IE6) 2 small files usually load slower than one larger.
XSLT gives savings when it generates lots of repetitive data, but then HTTP compression is so much better in eliminating repetition.
I don't think that XSLT saves server cpu either. If you have static data you can transform it once and serve XHTML.
If you have dynamic data you have to build DOM tree anyway. If you absolutely don't want to waste any cpu cycle on transformation, you can generate XML data using XHTML tags.
-- regards, Kornel Lesiński
****************************************************** The discussion list for http://webstandardsgroup.org/
See http://webstandardsgroup.org/mail/guidelines.cfm for some hints on posting to the list & getting help ******************************************************
