[
Lists Home |
Date Index |
Thread Index
]
On Sat, 2002-09-28 at 00:11, John Cowan wrote:
> Simon St.Laurent scripsit:
>
> > See, for instance, the "data versus text" section at the end of:
> > http://books.xmlschemata.org/relaxng/RngBookTextPatterns.html
> >
> > That feels to me like an overstep.
>
> Ah, I see. But the restrictions on the use of datatypes in mixed content
> are primarily for the benefit of validator implementers. It's much easier
> if you can define datatypes with a separate module that gets all the
> available content and then either matches or doesn't, deterministically.
Sorry, I cant't buy this argument :-) !
The datatypes libraries are implemented as independent modules, and they
are so independent that they act on text nodes independently of the
content model (mixed, simple or even -in the case of Relax NG which kind
of extends the notion of text nodes- attributes or lists of tokens).
Having implemented most of the Relax NG specification in my xvif[1]
implementation, I can tell you that "data" patterns in mixed content are
"natively" supported by the derivative method described by James Clark
and that ruling them out would add a significant amount of non trivial
code (which I haven't implemented yet since this is not my priority of
the moment)...
[1] http://downloads.xmlschemata.org/python/xvif/
Eric
--
Rendez-vous à Paris.
http://www.technoforum.fr/integ2002/index.html
------------------------------------------------------------------------
Eric van der Vlist http://xmlfr.org http://dyomedea.com
(W3C) XML Schema ISBN:0-596-00252-1 http://oreilly.com/catalog/xmlschema
------------------------------------------------------------------------
|
