W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > October to December 2003

RE: DAV Schema Assessment (was Re: rfc2518bis DAV DTD ...)

From: Julian Reschke <julian.reschke@gmx.de>
Date: Fri, 17 Oct 2003 19:19:05 +0200
To: "Eric Sedlar" <eric.sedlar@oracle.com>, <w3c-dist-auth@w3.org>
Message-ID: <JIEGINCHMLABHJBIGKBCAEGPINAA.julian.reschke@gmx.de>

> From: w3c-dist-auth-request@w3.org
> [mailto:w3c-dist-auth-request@w3.org]On Behalf Of Eric Sedlar
> Sent: Friday, October 17, 2003 6:53 PM
> To: 'Julian Reschke'; w3c-dist-auth@w3.org
> Subject: RE: DAV Schema Assessment (was Re: rfc2518bis DAV DTD ...)
>
>
>
> There are two possible use cases for a schema language to describe DAV XML
> syntax.  Case 1 is as a tool like BNF to formally describe the
> protocol.  A
> second is in the case where people would actually do validation
> at runtime.
> I don't think that most WebDAV implementations today actually run DTD
> validation.  For use case 1, XML Schema is fine since there is no
> backwards

I agree that use case 1 is the one we talk about (at least it's the one that
triggered this discussion). I happen to disagree that XML Schema is a good
choice, because *for this particular use case*, it adds only little to what
we already can do with DTDs, however it has the disadvantage of not having a
compact syntax I'm prepared to put into a spec.

(Side note: unless we use Trang to convert the XML Schema to RelaxNG Compact
Notation)

The DTD fragments we have right now at least offer good readability. I'm not
prepared to throw them away just because DTDs aren't en vogue anymore.

> compatibility issue.  If you want to require use case 2, then people would
> have to include a validation engine in their WebDAV implementations.  If
> WebDAV WG chose RELAX NG for use case 2, then all WebDAV vendors
> would have
> to acquire or develop RELAX implementations, which to me, is pushing RELAX

Use case 2 is of no interest to me.

> on people.  If all we care about is case 1, we could always write our
> standards using XML Schema, but just note that the schemas are
> for a formal
> description rather than for use in implementations.

We could, but what would be the advantage compared to DTDs? We would still
have to enumerate all the exceptions caused by WebDAV's extensibility rules.
As far as I understand, we'd only get rid of one single exception (being
namespace-awareness).

> As far as your name-calling, goes, yes, Relax is almost a real
> ISO standard
> like X.400 and other such successes from the ISO.  They're much
> better than
> those poorly specified ones coming from IETF and W3C like SMTP
> and HTTP.  I
> don't know why we have to descend into bashing of particular vendors or
> standards organizations to have this discussion.  W3C is the organization
> that currently owns most of the related standards we deal with in WebDAV,
> such as XML and HTTP, so yes, I believe its standards are more
> relevant than
> ISO ones when a conflict occurs (also note the mail address of this list).

In fact, WebDAV and HTTP are IETF specs. It's just this mailing list that
happens to be hosted by the W3C.

As for which organization is relevant -- this was not supposed to be an
argument *against* W3C or IETF. In fact, in that case I'd have to ask myself
why I'm putting all this work into IETF specification development.

I just wanted to clarify that RelaxNG isn't just an unimportant niche
product: one of it's creators is Jim Clark (editor of XSLT 1.0 rec) and it's
published by Oasis (members including Oracle, SAP and Microsoft, for that
matter). So as far as I'm concerned it has the same credibility as any
competing W3C spec.

Now I do understand that people who have heavily invested into XML Schema
want to see it used. However, that's really not a convincing element if
there are alternatives which seem to be better suited from a technical point
of view.

BTW here are some more alternatives:

- specify an XSLT transformation that will transform any legal WebDAV
message to a format that *can* be DTD validated, or

- specify constraints by a serious of XPath assertions (a la Schematron),
such as

	count(d:allprop|d:prop|d:propname)=1

...and so on.

> Schema was designed by a large committee to meet the needs of a
> wide variety
> of organizations and applications (e.g. structure definition, not just
> validation), as opposed to RELAX, which was designed by one guy.

Two, as a matter of fact.

So does this make XML Schema better per se?

> That's why
> Schema has more market acceptance and RELAX is more elegant.  In general,

So if all we need is a notation for RFCs, what's more relevant? Elegance or
market acceptance?

> the IETF and the WebDAV WG has tried to keep discussions rooted in
> practicality--what's actually out there in the bulk of the users.
>  As far as
> what major vendors are doing, since you mentioned my employer, I
> would point
> out that your employer is also a proponent of XML Schema (quoted from
> http://www.w3.org/2001/05/xml-schema-testimonial.html):
>
> SAP is pleased to see that XML Schema has become a W3C Recommendation. XML
> Schema is a key integration technology for supporting tightly coupled
> business processes through loosely coupled components within and
> outside of
> the company boundary and an essential standard for building and leveraging
> shared knowledge about collaborative services and processes. SAP is
> committed to embracing XML Schema throughout the mySAP.com e-business
> platform by providing XML-based services and leveraging XML Schema to
> support business integration within mySAP Technology.
> -- Dr. Peter Barth, Director Corporate Marketing mySAP Technology
> and mySAP
> Workplace, SAP AG

Actually, greenbytes is my employer.


Julian

--
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
Received on Friday, 17 October 2003 13:19:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:44:05 GMT