[dsdl-discuss] Re: Response to rest of Martin's comments

> In Schematron, abstract patterns can references other abstract patterns,
> abstract rules can reference other abstract rules, and in XML entities
> can reference other entities. (I need to add wording about that...good)

Even if it is doable in Schematron, I think that such descriptions should
be schema-language independent rather than Schematron-dependent.

> Consequently, I think there is no additional power in Murata-san's
> proposal. If Murata-san provides an example, I am happy to express
> it in Schematron, to help judgement. It is the convenience that
> is the issue, and the expected/indended use.

Thanks for your offer. The XML doc at the following URI provides
six lists of kanjis. Each list is intended for one grade in elementary
schools. Some local governments use kanjis up to grade 3 only, so that
non-Japanese can understand documents. To validate such documents, I
would like to create the union of three lists easily.

http://www.asahi-net.or.jp/~eb2m-mrt/dsdl/educationKanji.xml

> However, as I have said many times, I only made the Part 7 binding
> because I was asked to show how it was done. Actually, doing it was
> a good exercise for clarifying part 3, and so has been worthwhile
> even if we ditch it. Just as with Part 4, I think it is perfectly
> OK to have features that are primarily intended to augment Part 2:
> that goes with layering and supporting plurality--different
> languages have different needs.

Although I certainly care RELAX NG, I also think that we should allow
(a variation of) the DTD language to take advantage of character
repertorire constraints. I can also imagine that such constraints
are useful for other parts of DSDL.

> I think Murata-san and I agree that there are some constraint
> that would be better expressed as constraints on the semantics
> of all documents of a schema, and some constraints that would
> be better expressed as constraints on certain documents.

Yes, I agree.

>The
> first might indeed be better expressed piggy-backed on the
> RELAX NG schema (if there is one) and the second might be better
> expressed with Schematron schema. But I expect this may be too
> subtle for users: they may prefer a single mechanism.

I am not convinced if Schematron should be the only mother language for
character repertorire constraints and you are not convinced if RELAX NG
should be the only for them. Just like the DB community introduced
SQL/XML and XQuery, I think we should introduce both and let users
choose.

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)