RE: XSLT/XPath 2.0 (was "Identifying two tags...")

I agree with a lot of what Elliotte Rusty Harold 
said [snipped below]. To me, this points to a 'processing
mode' that other people have mentioned. I also 
agree with Bryan on the issue of 'schema neutrality'. 
If you need schema validation, it should work whether 
that is XML Schema, RELAX NG, or Schematron. 
 
> I don't want any of the core functionality to depend on schema types. 
> For instance, I do not want a function like starts-with(), round() or 
> getYear() to fail simply because there is no schema available. I do 
> not want equality comparisons between 2 and 2 to return false because 
> the first 2 is an integer and the second 2 is a string.

Or at least, if you are in 'non-schema' mode, then this should 
not fail but be coerced to a specific type just as XPath 1.0 
does. 
> 
> I do not want XSLT processing to fail because the instance document 
> has an xsi:schemaLocation attribute that points to a nonexistent 
> schema.

Isn't this related to the parser though? Lots of the XSLT engines 
fail a document that references a DTD that can't be located -- 
but only if the parser is invoked in validating mode. This seems
like something that should be a parser switch just as it is with DTDs.

[snipped]
> Any feature which can be implemented without PSVI information (e.g. 
> getting the year from 1999-12-21) should be implemented without PSVI 
> information.
> 
> I do not want any features to work differently because of schemas. 

This one seems almost impossible although I agree with the idea. Part 
of my concern is that so much of what I think I currently understand
is going to change and I'm going to have to relearn it with a 
different and more complex set of rules. 

Sara

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