[
Lists Home |
Date Index |
Thread Index
]
>-----Message d'origine-----
>De : Simon St.Laurent [mailto:simonstl@simonstl.com]
>Envoyé : vendredi 18 janvier 2002 18:28
>À : Nicolas LEHUEN
>Objet : RE: [xml-dev] RDDL (was RE: [xml-dev] Negotiate Out The Noise)
>
>
>On Fri, 2002-01-18 at 11:09, Nicolas LEHUEN wrote:
>> A document type cannot be guessed from the list of
>namespaces it uses.
>
>I don't really believe in such things as document types - though I
>suppose you're correct that a mere list of namespaces is inadequate if
>you really want to nail down precisely what type of document it might
>be, since the list doesn't define the possible interactions. Only the
>document itself really does that...
Well, DOCTYPE has always proven useful until now, but it only points to
DTDs. So why drop the idea ? Pointing to a catalog of meta-data files (DTDs,
schemas, human-readable documentation, etc.) would be very useful...
>> There
>> should be a way to bind a document to a series of meta-data
>resources, and
>> that's what I thought RDDL should be.
>
>I gave that a go once with XPDL:
>http://www.simonstl.com/projects/xpdl/
Thanks for the pointer. I'll have a look at it right now.
>There's also the XHTML meta tag and its cousins if you want to go that
>route.
I don't want to go the XHTML route, but the XML route. I'd like an
equivalent of the DOCTYPE, but more opened. I don't want to answer the
riddle 'what is at the end of a namespace URL', I just want a useful tool.
>> But it is not, it is only a way to
>> bind resources to a given namespace. If I followed your
>logic, to validate a
>> RDDL document using RDDL, I would load the RDDL for the
>XHTML, RDDL and
>> XLink namespaces. Now I don't have one, but three resources
>directories in
>> which I'm supposed to find a DTD. Great, I've got three :
>the DTD for XHTML,
>> the DTD for RDDL, and the DTD for XLink. How do I make my
>computer select
>> the good one, i.e. the RDDL DTD, instead of the two other
>(and especially
>> the XHTML one, since the root element html
>
>That's not a problem of RDDL - that's a general validation issue.
>Schema mathematics (except for RELAX NG) are currently wretched. I'd
>like to see RDDL files contain modules which only describe their
>particular namespace, not the entire concoction. Using those is
>currently difficult, however, so we're stuck with what we've got.
>
>Incremental progress, not the perfect solution you appear to
>be seeking.
>
>> xmlns="xmlns="http://www.w3.org/1999/xhtml"; is found in both
>XHTML and RDDL
>> ?).
>
>Yep. That's the current sad state of the schema art.
It's just because a schema is in no way associated to a namespace. Just
think of namespaces as the first half of a QName : a schema is just a set of
constraints on where and when some QName can/have to appear in an XML
document. Namespaces do not imply any particular schema, and vice-versa. If
you see it this way, schemas aren't so wretched after all. We just have to
lower the impression namespace can make on us.
>> You bet it will ! As soon as someone will try to use RDDL for RDDL,
>> XHTML+SVG, XHTML+MathML, people will begin to discover that
>a namespace is
>> not a document type, and that mixing different namespaces in the same
>> document create new document types, different from each
>original 'default
>> document types' possibly associated to the namespace. From
>there, they will
>> throw out RDDL and try to think seriously about the problem.
>
>Nah. Maybe they'll get rid of their expectations about schemas and/or
>the whole fuzzy notion of document types instead.
>
>It sounds like you want a complete solution. I don't think there are
>any complete solutions in XML, just a box of parts. Change your
>expectations, and you might be a lot happier.
I think you're a bit pessimistic here. DOCTYPE does work. Why not keep the
idea and point to a catalog of meta-data resources, each identified by a
different role/purpose/language ? This does not seem unfeasable.
Regards,
Nicolas
>--
>Simon St.Laurent
>Ring around the content, a pocket full of brackets
>Errors, errors, all fall down!
>http://simonstl.com
|
