Re: [relax-ng-comment] XML 1.1 and a possible RELAX NG technicalcorrigendum

[James Clark <jjc@jclark.com>]

--On 19 December 2001 16:59 -0500 John Cowan <jcowan@reutershealth.com> 
wrote:

> James has in the past suggested that XML name rules should be checked at
> the validation level rather than at the well-formedness level.

I think I suggested merely that they should not be part of some layer other 
than well-formedness, rather than that they should be part of validity.

> In
> pursuit of this, it would be fitting for RELAX NG to support such
> validation.

This sort of checking seems a different kind of thing to what RELAX NG is 
doing.  I think it should be a separate module that can be plugged into the 
processing pipeline independently of validation.

> I propose the following approach, to be instantiated in a RELAX NG
> TC when and if XML 1.1 becomes a W3C Recommendation:
>
> Allow a name to be matched against a datatype, with syntax
> <name><data type="goodname">...</data></name>.  This syntax represents
> the class of names that are lexically equivalent to strings of type
> goodname in some datatype library or other.  In particular, one could
> make use of the XSD2 pattern datatype to specify useful patterns.

The problem with this is that RELAX NG validators have to be able to 
determine whether the intersection of two name classes is empty (e.g. for 
duplicate attribute detection).  This is hard/impossible if you allow 
arbitrary datatypes in name classes.

James