Philippe Poulard wrote:
>
> I wonder why not use a more abstract level, say simply :
> http://www.rddl.org/natures#Schema
> which could be used for DTD, W3C XML Schema, Relax NG or other schema
>
> What is found within will tell which one is used.

The RDDL natures listed in http://www.rddl.org/natures/ are intended  
to be a useful set of examples. When authoring a RDDL namespace  
document, you can use whichever RDDL natures are considered most  
appropriate.

>
> Moreover, what would happened if I put some Schematron assertions  
> inside a W3C XML Schema ?

I am not sure. Presumably one could have two descriptions of the same  
document, one labelled with a nature "XML Schema" and the other with  
a nature "Schematron" but I am not sure what the implications of that  
might be.

>
> This is the same with catalogs :
> http://www.rddl.org/natures#Catalog
> would suit to SGML catalogs, OASIS catalogs, and others
>
> IMHO, it is more close to the idea that a resource belongs to a  
> class or group.
>
> I noticed in RDDL some assertions that are very discutable :
> http://www.rddl.org/purposes/
> Section "Schema related purposes / Validation"
> -I could validate a document with a DTD after parsing it, even if  
> the document doesn't refer to it (example : because I need to  
> perform XInclusion before the validation)
> -Schema validation can be also combined with the parse phase.

Again, these are intended to be a useful set of properties, but you  
may wish to use your own if these don't fit your needs. The purposes  
documented in http://www.rddl.org/purposes/ are not intended to be an  
exhaustive list.

I am not sure what the objection is, generally, to labelling a schema  
document with the purpose "validation". It seems useful to me, but  
certainly "validation" isn't the *only* thing you might want to do  
with a schema which is the exact reason for having both a RDDL  
purpose and nature.

Jonathan