Re: [xml-dev] RNG vs. XSD : is the use of abstract types and polymorphis

["Simon St.Laurent" <simonstl@simonstl.com>]

On 3/14/12 9:03 PM, Rick Jelliffe wrote:
> I like what Norm wrote.I'd add a couple of things.

As do I, though my things are a little different.

 From my perspective, W3C XML Schema was a trainwreck because its 
foundation structures came from the languages people wanted to use to 
process XML with, not from much precedent in markup itself.  Classical 
inheritance was the way of Java and C++ and many others, and was 
apparently a siren call few could resist.

(At the time I was doing my XML programming in Java, a choice I've since 
concluded was akin to building a garden with a cement mixer.)

XSLT 1.0 may provide a bit of contrast.  I still think leaping from the 
annotation model of CSS to the transformation model of XSLT was drastic 
overkill for the at least theoretical application of SGML on the Web. 
However, unlike W3C XML Schema, XSLT itself was built on the very 
different, and I now think saner foundations of functional programming 
languages.

Yes, it killed OOP and procedural programmers that variables weren't 
variable.  People who hoped XSLT would convert objects to and from XML 
were disappointed to find that it was really about transformations of 
XML to XML.  Those limits, however, are massive strengths when you need 
XSLT to do the kinds of transformations it was built to do.

It's not really a surprise to me that RELAX NG and XSLT 1.0 have a 
common founder in James Clark, who reliably looked beyond the "why can't 
we make it all look like Java" demands of the day.

I can imagine inheritance working well in a schema language.  It might 
be sane, for instance, to create a data type language using prototypal 
inheritance - what JavaScript is good at - rather than locking itself 
into the hierarchies of classical inheritance.  Alas, I haven't had time 
to play with that.

Maybe that's for the best, though.

Thanks,
-- 
Simon St.Laurent
http://simonstl.com/