[tmcl-wg] Comments on 2003-02-23 draft

[Robert Barta]

On Tue, Apr 08, 2003 at 09:32:10PM +0200, Lars Marius Garshol wrote:
> | I see this completely outside the scope of TMCL in the sense that
> | when you have a TMCL instance then it should be possible to generate
> | this storage representations (nice idea actually) from that
> | document.
> 
> Up to a point I agree. The point is that TMCL must be designed in such
> a way that schema introspection is *possible*. XML Schema, DTDs, and
> RELAX-NG all allow introspection, but Schematron does *not*, because
> it is opaque. That's the sort of point I am trying to make.

I am now confused by the word 'opaque'.

Sorry to be so insisting, but is your point that Schematron is
specifying allowable patterns 'declaratively' whereas XML schema and
friends do this in a 'constructive' way?

If so, then we have a problem.

> | Just to be sure whether I understand it correctly:
> | 
> | The task is to have, say, a BOOK class
> | 
> |    class BOOK:
> |       title  : string
> |       authors: SET of AUTHOR
> | 
> |    class AUTHOR
> |       name : string
> | 
> | and in a map
> | 
> |    python-for-fun (book)
> |    bn: Python ..
> | 
> |    (is-author)
> |    author: xxxx
> |    opus : python-for-fun
> | 
> | and mediate between the two?
> 
> Yep. I think with a little bit of glue you could do that, but you'd
> need a schema to know whether to make authors a set or a single value,
> and so on.

Well, yes, this is input coming from the ontology.

\rho