RE: [xsl] XSLT 2.0: Schema-aware processor: What are the compelling advantages o

[Justin Johansson <procode@xxxxxxxxxx>]

>> Would it be any easier to implement the XSLT beast using RELAX-NG as the
schema language?

>There's an important difference here. One of the reasons that W3C XML Schema
>is such a beast is that it was designed to assign types to elements in your
>instance document as well as giving a yes/no answer on validity. Relax NG
>does validation only, not typing. This feature is heavily exploited in the
>way that schema-aware XSLT processing works.
>
>Michael Kay

>It would be nice (and one reason why I've not used a schema aware
>processor is I have no particular other use or desire to have XSD for my
>documents having already got dtd and relax) but while there are some
>type library extensions for relax, it's basically architechurally
>opposed to the entire concept of an annotated infoset output, on which
>the schema aware XPath2 depends.
>
>David Carlisle

Ouch.  I wonder how that makes James Clark feel?   It does, sadly, limit
RELAX-NG's potential.

Thanks for the explanations though; it's easy to overlook these things.

Personally I adore RELAX-NG due its sheer elegance but this does put a
spanner in the works.

Like many others I find writing XSDs tedious .. particularly when it comes
to schema evolution
and having to upgrade both XML data and XSLT stylesheets.  Of course
without schema a
project with ad hoc XML files is doomed to failure or undesirable longevity.

Does the whole XSD schema evolution thing beg for a product which writes
stylesheets which
can be applied to migrate and simultaneously evolve  XML and XSLT sources?

Microsoft seem to use some XSLT technique like that as part of InfoPath
(when you upgrade the
schema of forms main data source) though even in InfoPath schema evolution
is tiresome.

Justin Johansson
Freelance XML / XSLT / XQuery Developer
Australia

procode(at)tpg(dot)com(dot)au