[
Lists Home |
Date Index |
Thread Index
]
To come back to the RELAX NG marketing issue, I really don't see any domain
where XML Schema is preferable to RELAX NG, apart from tool support.
And why has XML Schema a better tool support ? Because it is so. Because
everybody let the market decide instead of making an decision in good faith.
Because XML Schema was proposed by the W3C, which is sponsored by vendors.
So vendors influenced vendors, and what we get is a vendor-friendly
technology (expensive to implement (even in open-source), expensive to use,
with nice formal specs to please nerds and make sure that common people
can't get it).
Just have a look at the xml-dev archive and count the number of threads with
this synopsis :
Q] How do I do xxxx in a DTD / XML Schema ?
A1] You can, by doing (very complicated explanation) / You cannot, sorry
A2] However, you can do it with RELAX NG by doing (way more simple
explanation).
To me, it's pretty obvious. What is the point in using XML Schema ?
So what does RELAX NG really needs ? Tool support.
Why is tool support so problematic ? Because each tool vendor has to
implement the support for a given schema language himself, even if it is not
its core business.
Why can't he reuse already existing code ? Partly because until now,
parsing, schema parsing and validation are mangled in the same code base, so
that you can't take a part without taking all, which is not acceptable for
many reasons, both technical and economical. Schema parsing and validation
is not considered as an independant module (apart in the JARV, Jing and Sun
MSV projects), so this slows the progress of schema languages towards
technical quality (simplicity, flexibility, completeness), and bind its
progress to political and short-term economical decisions.
That is the reason why I really believe in the necessity of a careful
definition of independent concerns in XML (parsing, abstract grammars,
validating, transport, etc.), and the definition and progressive refining of
APIs for those different concerns. Each time different concerns are not
properly separated, each time they are not properly orthogonalized, it is a
slowing factor, and the opportunity for politics to jump in the bandwagon.
And now, for the mother of all politic battlefields : the definition of
APIs... :)
Regards,
Nicolas
>-----Message d'origine-----
>De : Stefano Debenedetti [mailto:sdebenedetti@e-tree.com]
>Envoy? : mercredi 27 mars 2002 17:10
>? : Nicolas LEHUEN
>Cc : 'Matthew Gertner'; 'Bullard Claude L (Len)';
>'xml-dev@lists.xml.org'
>Objet : Re: [xml-dev] Capitalism and XML (was RELAX NG Marketing)
>
>
>I don't think this is off-topic, it all started up as a long thread on
>schema interoperability, XML itself is about interoperability.
>
>I like very much the terminology proposed by Manos Batsis but
>I agree it
>is unfortunately not suitable for discussing interoperability
>on xml-dev
>so I'll just propose another synthesis:
>
>While dicussing interoperability one must ask who's to interoperate.
>
>That's why after having computers of any kind interoperate
>with respect
>to pure data (ASCII, Unicode, XML, ...), after making software using
>this data interoperable by means of schemas (XML again,
>XML-Schema, RNG,
>...), it is now a matter of discussion how interoperability between
>those schemas could be acheived by human beings and companies (RDF &
>Semantic web, Web services, hell: even "generalized
>integration between
>clients and arbitrary objects and services written for popular
>application servers" [1] would do).
>
>So this is not OT IMHO, and my personal thought about this is:
>
>If a spec is clear enough to grant interoperability for the masses,
>who's going to need to pay a software only for its interoperability?
>
>Everybody would be able to implement the spec or find a suitable open
>source implementation to do whatever with.
>
>If this doesn't happen now, as the first century of computers runs
>towards its end, it is because there is a precise will for this not to
>happen. Interoperability is not wanted for "anybody", despite
>marketing.
>
>So the question is: who wants _people_ to seamlessy integrate the
>vocabularies they need, having interoperable pieces of software do the
>right thing about each of them instead of messing it all up?
>
>Only the open source community in its wildest dream.
>
>The others are people believing they need to sell/buy a product that
>allows that, so interoperability in their vision is meant just for
>people willing to buy such software. Whether such products usually end
>up being interoperable between them or not is a subject it
>would be too
>sad for me to detail any further here.
>
>Think about email threads. A single spec would maybe not be enough to
>acheive that, but if we still don't have some sort of open
>standard for
>threading on our desktop it is because everybody thinks it is
>better for
>her or him (as a company) to sell/buy this functionality as a piece of
>some highly pricy collaborative and/or learning software product.
>
>(For open-standard-on-a-desktop here I mean some configuration options
>all threading-capable-black-box-software we have installed on our
>desktop should have in common. Just the same way anybody (..ehm..) is
>able nowadays to find the POP or SMTP configuration options on any
>(..uhm..) email client.)
>
>I think lack of interoperability among email clients is between
>responsibles for slowing the decrease (if there is any) of
>overall world
>ignorance, thus augmenting overall world poverty. By cohincidence it
>happened also to be between the causes of one of the most impressive
>enrichments of world history: Mr. Gertner, do you think the gain the
>world had from having Microsoft is worth more than human knowledge and
>capability to communicate? Do you think
>Microsoft+Sun+Ibm+Oracle+Cisco+Adobe+Macromedia are? Not in
>any way and
>not even if the whole process of human knowledge sharing was
>relying on
>them only.
>
>So why are we waiting for these companies to give us the standard for
>interoperability of knowledge (when they even failed on CSS)? They
>won't, not until they are sure enough the profit for them
>would be more
>than the loss. That's how SGML came to light, for example, but,
>according to xml-dev, it seems highly unlikely that anything
>of the kind
>is going to happen now and that's why people are starting once
>again to
>react: "hell, we're going to do that (XML-Schema-like-facility)
>ourselves", "any marketing expert is volunteering?" ..come on,
>I really
>appreciate the spirit behind the effort but that can't always
>be left to
>heroism, unless you think that what we need to make these
>companies give
>us we want is really a hero, I don't think so.
>
>Let's rethink the framework, keep it simple, make it function.
>
>Let the W3C be the companies' public commitment to public
>interest, let
>it be their commitment to clarity and interoperability, let the
>community be their judge and requirements-setter. Anything more than
>that, like policies towards open source software for example, should
>stay in their creepy top-secret business policies.
>
>:-) Thank you for reading so far, kind regards
>
>ciao
>ste, human. Lefty only by history: my grandfather had the same job
>qualification I have ("mechanographical operator")
>
>[1] this quote is an extension of what was stated as "a major design
>goal" for some of their future products by the CTO of one of the web
>heavyweights we all know about and use the software of so often. My
>answer to him on that NDA-protected-thread was that XForms was the way
>to go to acheive that, not anything proprietary. That's why he's a CEO
>and I'm a "ceo" which in Venetian dialect means "little kid".
>
>
|
