Re: comments and questions

[James Clark <jjc@jclark.com>]

From: "Franck Delahaye" <franck.delahaye@akazi.com>
To: <relax-ng-comment@lists.oasis-open.org>
Sent: Monday, June 04, 2001 7:10 PM
Subject: comments and questions

>  what is left for 100% of the syntax validation is:
> define and validate FM-UML relations in FM-XML:
> I think that the cross-reference pattern of "relax ng" seems ok for that.

If you find it to not to be enough, please let us know.

> validate XPath typed attributes:
> several attributes of FM-XML should contain an XPath expression (just like
> the "select" attribute of XSLT templates)
> validate Java typed attributes:
> several attributes of FM-XML should contain a subset of Java expressions.
> I'd like to know:
> if my problem is in the scope of RELAX NG, or if I took a wrong way ?

RELAX NG is datatype-neutral: it allows you to plug in any library of
datatypes that you want.  RELAX NG doesn't build in an XPath datatype or a
Java expression datatype, any more than it builds in a double datatype or a
int datatype.

> if there will be a public alpha version of a RELAX NG validator,

In due course, I expect there will be.

> and if
> James Clark's TREX validator will become a RELAX NG validator ?

I plan to adapt my TREX validator for RELAX NG.

> What  would be a great thing is to provide some kind of accompagnying
> "validation component interfacing standard" to be abble to build easely a
> "multi language" validator starting from TREX, Java, VB.. validation
> component. But this may be not realistic...

We have had discussion about a standard interface for datatyping libraries
which would allow you to plug in components in Java for validating element
content.

James