[
Lists Home |
Date Index |
Thread Index
]
James,
I agree with everything you wrote. In particular, I see the issue of the
perceived W3C ownership of the space. At the same time, there is also
widespread perception that XSD has serious flaws, especially the lack of any
attempt to find "80/20" compromises (interesting in light of the recent
discussion on this topic on XML-Dev). I believe that an alternative schema
language could gain widespread adoption, but the perceived authority of the
W3C admittedly makes this an uphill battle.
Two questions:
1) What is the status of type derivation (i.e. inheritance) in RELAX NG?
This is a restatement of my previous post. I seem to recall something about
inheritance being left out of the core spec in the interest of simplicity. I
also recall there being some informal proposal to layer type derivation on
top of RELAX NG, which it is apparently designed to accomodate.
2) What is the status of Xerces support for RELAX NG? My friends in the
trenches tell me that Xerces is widely seen as the de facto standard and
that without said support, no schema language has much of a chance at
widespread adoption.
And a proposal:
As I implied in my previous posts, I am heavily involved in UBL, and this
effort has a great shot at becoming very influential. I am specifically
chairing the Context Methodology Subcommittee, whose task it is, among other
things, to determine whether schemas that have been adapted for business
reasons from a core schema (adding fields, extending or restricting code
lists, etc., etc.) can be strictly derived from the core schema. Without
getting into all the mucky details, it seems clear that there are quite a
few common business requirements that are not covered by XSD. This should
not be interpreted as a criticism of the XSD team; these requirements were
simply not known at the time.
That said, I believe that it would be easier to design an appropriate
inheritance mechanism for RELAX NG that integrates these requirements than
to try to adapt XSD, which is tightly bound up with its current approach to
type derivation. If someone like you or John Cowan were willing to accept
the informal (or formal) status of an advisor to UBL and tailor an
inheritance mechanism to the requirements that we are discovering, this
would give us a strong impetus for putting RELAX NG on equal footing with
XSD in communicating our schema and context methodology work.
I say this only because I personally feel strongly that if the notion of
inheritance is to have any meaning in the XML world, it has to work for
cases like UBL. If we end up making context-modified schemas into
transformations of core schemas instead of derivations, then we have no
polymorphism, and I struggle to see what the value of object-oriented
features is in this case.
Now that I think about it, I also seem to remember reading something that
you wrote about inheritance being out of scope for XML since a) no XML
inheritance schemes support multiple inheritance and b) XML is not a
modeling language, so true modeling languages like UML should be used for
modeling, and XML schemas generated on the basis of UML models. I strongly
dispute this since I feel that polymorphic behavior of XML documents would
be an extremely valuable thing. I'd happily justify this statement if anyone
is interested, but not in this post since it is already way too long.
Cheers,
Matt
> -----Original Message-----
> From: James Clark [mailto:jjc@jclark.com]
> Sent: Tuesday, March 26, 2002 1:10 PM
> To: Matthew Gertner
> Cc: xml-dev@lists.xml.org
> Subject: Re: [xml-dev] RELAX NG Marketing (was RE: [xml-dev]
> Do Names Matter?)
>
>
> My two cents is:
>
> - RELAX NG has achieved depressingly little success so far in
> terms of
> market adoption.
>
> - However, it would be premature to throw in the towel. The
> RELAX NG spec
> was finalized less than 4 months ago. XML is going to be
> around for a long
> time to come and there will be a continuing need for an XML schema
> language. XSD is not suddenly going to become radically
> simpler. Even if
> RELAX NG is only adopted by a small percentage of the XML
> market, the XML
> market is big enough that this is still worthwhile.
>
> - RELAX NG needs more evangelizing. I think for both
> Murata-san and myself
> it would be fair to say that our strengths are in the
> technical department
> more than the marketing department.
>
> - The key difficulty facing RELAX NG is that the W3C is
> perceived as owning
> the XML core standards space; it is very hard for any
> specification to
> compete with a W3C Recommendation in this space, especially when that
> Recommendation has been adopted by the big players including
> Microsoft. It
> will be hard for RELAX NG to compete no matter how brilliantly it is
> evangelized.
>
> - I don't think the name's a big deal; I agree calling it
> pronouncing it
> "Relaxing" sounds a bit funny; I typically pronounce it
> "Relax En Gee". I
> don't think the pronunciation is enshrined in any spec, so this is
> something that is easy to change. But I don't think changing
> the name
> would help: at this point it would just dilute what little
> name recognition
> we have.
>
> - There needs to be a tool to translate RELAX NG into DTDs
> and XSD. This
> would allow people to have the advantage of authoring in an
> easy to learn
> and easy to use schema language but still be able to
> interoperate with the
> rest of the XSD-using world. At the moment, choosing to use
> RELAX NG means
> choosing either (a) not to interoperate with DTD/XSD tools
> and users or (b)
> to maintain two schema versions by hand; it's not surprising in these
> circumstances that people don't choose RELAX NG. Maybe this
> is wishful
> thinking, but I think if people could automatically generate
> decent DTDs
> and XSD schemas from RELAX NG schemas, then they would be
> much more likely
> to consider authoring in RELAX NG.
>
> James
>
>
>
>
>
|
