On Fri, 2005-12-02 at 09:20 +0000, Michael Kay wrote:
> > 
> > All arguments for XHTML everywhere eventually boil down to arguments
> > that rather than 
> > 
> > <monty>
> >   <python/>
> > </monty>
> > 
> > I should write:
> > 
> > <div class="monty">
> >   <span class="python"/>
> > </div class"monty">
> > 
> > No bloody thank you.
> 
> I've recently seen people trying to use the latter approach for adding
> extensibility to financial messages, in the apparent belief that writing
> xs:any in a schema was somehow against the spirit of things.
> 
> But doesn't it come back to architectural forms? (Perhaps I'll understand
> them one day...)

Yes!  Yes!  Yes!  We need architectural forms painfully.  SGMLers sold
us out by making them very unnecessarily gnomic.  OK just kidding,
SGMLers, but when even Mike Kay admits he doesn't grok them, someone
feel asleep at the tutorial wheel.

Anyway, I went from not understanding the concept at all to presently
thinking they're the solution for a lot of problems in XML today.  I
still don't understand them in the SGML original, but Schematron's
Abstract Patterns (SAPs), and Ricko's patient explanations of how these
are just AFs naturally cast into the Schematron framework have made me
realize the power of the idea (John Cowan's and Jeni Tennison's
explorations here on XML-DEV a few years ago also helped me).

I've distilled the technical nuts and bolts of SAPs into the second half
of this article:

Discover the flexibility of Schematronabstract patterns
http://www.ibm.com/developerworks/xml/library/x-stron.html


-- 
Uche Ogbuji                               Fourthought, Inc.
http://uche.ogbuji.net                    http://fourthought.com
http://copia.ogbuji.net                   http://4Suite.org
Articles: http://uche.ogbuji.net/tech/publications/