[dsdl-discuss] Re: Three debatable features for Part 7

Martin,

> Amajor problem with the Duerst proposal is it lack of identifiable objects.
> It does not name collections, and only allows named collections to be
> identified by URL. It would appear that you can next references
> indefinitely, but there are no rules re conflict resolution. This is OK with
> a simple add-only model, but not if you allow exceptions to rules (e.g.
> <range>A-Z</range><except>AEIOU</except>

I do not understand your point. Are you happy if we use URNs rather than URLs?
What do you mean by "no rules re conflict resolution"?

> > Even if we do not borrow this idea, Schematron (as a host language) can
> > mimic it by using two Schematron rules (one for the kernel
> > and another for the hull). However, RELAX NG or DTD do not have this
> > option.
>
> Why can't RELAX handle it, providing we tell it that the language it is
> defined in is DSDL7? I thought RELAX allowed any set of datatypes to be
> applied, and connot see why you can't apply named character sets as a
> datatype of a specified kind? (Is there any restriction that prohibits
> multiple datatypes being defined for an element in different environments?)

I am afraid that I was not clear enough. I was wondering what happens
if Part 7 DOES NOT introduce kernels and hulls. In Schematron, we can
still mimick kernels and hulls. One rule can reference to a collection
of definitely-OK characters and another rule can reference to another
collection of probably-OK characters. These rules can specify
different error message. However, in RELAX NG, we cannot mimick kernels
and hulls when Part 7 DOES NOT have them.

Cheers,

-- 
MURATA Makoto <murata@hokkaido.email.ne.jp>
--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)