On Thu, 25 Nov 1999, Oren Ben-Kiki wrote:

> DSD has specified as one of its goals:
> 
> "... expressive power of DSDs should be close to that of XSLT ..."
> 
> This raises the question of why not to use XSLT directly, now that we have
> the <xsl:message> construct (a document is valid if and only if applying the
> XSLT "schemasheet" does not invoke any <xsl:message> calls).

See 
	http://www.ascc.net/xml/resource/schematron/schematron.html
for a schema language built using XPaths on top of XSLT. See also
	http://www.xmlhack.com
for some further thoughts along these lines.

I think the troubles with using XSL directly as a schema language are:
	1) it is too complicated to use directly: writing XSL is still 
programming, and I think Schema users may find that the extra power gets 
in the way; 
	2) reasoning about what you are defining becomes difficult: XSL 
makes a fine validation language as it is (i.e., where one is defining 
some ad hoc constraints) but I think a schema language needs to provide 
elements which correspond to the objects being declared: C is not a 
schema language but it can be used as a validation language, along the 
same lines;
	
	
Rick Jelliffe


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list