xml-dev - Re: Must DTDs constrain order?

[Eric van der Vlist <vdv@dyomedea.com>]

Tim Bray wrote:
> 
> At 09:55 AM 28/10/00 +0200, Eric van der Vlist wrote:
> >I begin to think that XML vocabularies should take example on XML itself
> >and that when designing a vocabulary one should start by defining the
> >infoset (the data model), a canonical form (a very strict syntax) before
> >defining one (or more if needed) "user XML syntax(es)" and canonization
> >algorithms with reference implementation(s).
> 
> Er, this might indeed be a good idea, but XML itself does not serve
> as a supporting example.  

Maybe as a supporting counter example then ;=) 

> The XML syntax was designed first.  The
> infoset came later.  

If the XML infoset had come before, we would probably have avoided some
incoherence between the XPath and DOM data models to name few.

(I don't want to sound critical or rewriting the XML history, I am just
trying to learn from this and from my own experience.)

> Some people are horrified at this, feeling that
> the infoset (or grove, depending on your religion) is The Real Thing
> and the syntax merely an ephemeral expression of it.  I disagree, but
> that horse's corpse lies about 100,000 messages back on the xml-dev
> trail, having been beaten to horse tartare. -T

Speaking about XML, it would have been difficult (or impossible) to
specify a data model (or infoset or grove) before the exposure to
applications such as DOM or XPath.

For less generic specifications, it might still be possible...

Eric
-- 
------------------------------------------------------------------------
Eric van der Vlist       Dyomedea                    http://dyomedea.com
http://xmlfr.org         http://4xt.org              http://ducotede.com
------------------------------------------------------------------------