RE: DTD for supported-method-set (3.1.3)

If I might suggest that this can be solved by using XML Schemas, which are
now a full W3C recommendation.  This will also solve the problem of the DTDs
having so many "any" elements, which is a looser spec that may lead to
interoperability problems.

> -----Original Message-----
> From: ietf-dav-versioning-request@w3.org
> [mailto:ietf-dav-versioning-request@w3.org]On Behalf Of Julian Reschke
> Sent: Wednesday, October 24, 2001 7:29 AM
> To: Clemm, Geoff; ietf-dav-versioning@w3.org
> Subject: RE: DTD for supported-method-set (3.1.3)
>
>
> > From: ietf-dav-versioning-request@w3.org
> > [mailto:ietf-dav-versioning-request@w3.org]On Behalf Of Clemm, Geoff
> > Sent: Wednesday, October 24, 2001 4:22 PM
> > To: ietf-dav-versioning@w3.org
> > Subject: RE: DTD for supported-method-set (3.1.3)
> >
> > ...
> >
> > Actually, in looking at the DAV:expand-property example, we clearly did
> > not intend for the attributes to be in the DAV: namespace (since they
> > are not qualified with "D:" to indicate they are in the DAV: namespace).
> > So I'll change that to "the intention was for every element to
> be declared
> > in the DAV:namespace and every attribute to be in the default
> namespace".
>
> Correction: not in the default namespace, in *no* namespace. Attributes do
> not inherit a default namespace.
>
> >    Well.
> >
> >    First of all, this is a very uncommon way to use attributes. In
> >    vocabularies like XSLT, XSL-FO or XHTML, attributes are declared
> >    with no namespace and rely on their context. Namespaced prefixes
> >    usually only appear in vocabularies where the attributes can appear
> >    on elements in other namespaces, such as xml:lang (XML itself),
> >    xlink:role (XLink) or xsi:type (XML Schema).
> >
> > OK, I'm sold (:-).
> >
> >    So besides that this would be very strange, the DTD fragment
> >    doesn't specify it (well, it can't really, because DTDs do not know
> >    about namespaces).
> >
> > Yeah, that doesn't really argue either way, since as you say,
> > DTD fragments cannot declare anything about namespaces.
> >
> >    So my suggestion would be
> >
> >    a) to define that the name attribute is in no namespace,
> >
> > I'll go with that.
> >
> >    b) fix the DTD fragment,
> >
> > I wouldn't do that (since it would be a non-standard usage of DTDs).
>
> Correct. I should have said: either a) or b). a) was preferred anyway.
>
> >    c) add examples to this (and maybe other) definitions. (Sorry).
> >
> > I'll try to squeeze in an example of the DAV:supported-method-set
> > property in the final editing pass.
> >
> >    >    In retrospective, just defining properties in terms of DTDs --
> >    > while DTDs
> >    >    can by definition not be normative for WebDAV (*) --
> isn't a good
> > idea.
> >    >    (*) DTDs can not properly capture namespace information.
> >    > Neither can they
> >    >    adequately model the extensible model defined in section C.2.2
> >    > of RFC2518
> >    >
> >    >
> > (<http://www.greenbytes.de/tech/webdav/rfc2518.html#rfc.section.C.2.2>).
> >    >
> >    > I'll let the folks that like DTD's respond to that (:-).
> >
> >    Did I open a can of worms? DTDs can not be used for normative
> > descriptions
> >    of XML vocabularies like WebDAV's. Even the W3C doesn't attempt to do
> > that.
> >
> > Well, if it's a can of worms, it's one we inherited from 2518 (:-).
> > I did at one time suggest we purge DTD's from the spec, but this was
> > not something the working group supported.
>
> Interesting.
>
>
>

Received on Wednesday, 24 October 2001 13:36:43 UTC