RE: RFC2518 (WebDAV) / RFC2396 (URI) inconsistency

> From: w3c-dist-auth-request@w3.org
> [mailto:w3c-dist-auth-request@w3.org]On Behalf Of Jim Whitehead
> Sent: Wednesday, November 21, 2001 7:07 PM
> To: w3c-dist-auth@w3.org; uri@w3.org
> Subject: RE: RFC2518 (WebDAV) / RFC2396 (URI) inconsistency
>
>
> > > * WebDAV marshals "dav:" URIs that are the name of XML elements as a
> > > {namespace} + {opaque_part} pair.  So, for example,
> > "dav:creationdate" is <D:creationdate xmlns:D="dav:">.
> >
> > RFC2518 says *nothing* about URIs in the DAV: URI scheme. RFC2518 itself
> > never says that an element name or a property "has" a URI.
>
> Section 18 clearly states that "The property names and XML
> elements in this
> specification are all derived from the base URI DAV: by adding a suffix to
> this URI, for example, DAV:creationdate for the 'creationdate' property."

I admit I missed that because it's under "IANA" considerations :-)

> Now, we've had discussion on the WebDAV list to move away from this
> position, and more towards the {namespace identifier} + {name} position
> embedded within the XML namespace draft. This strikes me as a good thing.

The XML namespace recommendation doesn't specify a specific representation.

> > <D:creationdate xmlns:D="dav:"> must be read according to the specs that
> > exist, and this means: an XML element with name "D:creationdate",
> > local name "creationdate" and namespace name "dav:". BTW: this
> should have
> > been "DAV:", right?
>
> URI scheme names are case insensitive (see Section 3.1 of RFC 2396). So,
> "DAV:" and "dav:" are equivalent.

Yes, but XML namespace names are case-sensitive. So <multistatus
xmlns="DAV:"/> and <multistatus xmlns="dav:" /> are different things, and
the latter isn't in the "DAV:" XML namespace (assuming for now that the
notion of "DAV:" XML namespaces exists at all).

> > If you claim that any element or property in WebDAV has a URI,
> > you'd have to answer:
> >
> > - do WebDAV element names and properties share the same namespace?
>
> Yes.
>
> > - what are the URIs (identifiers!!!) for: <cd xmlns="http://a/b/"
> > /> and <d xmlns="http://a/b/c" />?
>
> Following WebDAV concatenation rules:
>
> <cd xmlns="http://a/b/" />  => http://a/b/cd
>
> <d xmlns="http://a/b/c" />  => http://a/b/cd
>
> Again, let me stress that this is what RFC 2518 says, and we have
> discussed
> on the WebDAV list that this is probably not the best path to take going
> into the future.

Ok.

> > It's shorter, but it's invalid (according to the XML NS rec), while the
> > other one is perfectly valid.
>
> OTOH, I haven't heard a compelling reason why the XML NS rec couldn't be
> changed. Certainly this change would have far fewer interoperability
> implications, especially since most namespace implementations
> already allow
> just a URI scheme.

I understand why one would prefer to have a "small" change on somebody
else's spec, but from the feedback I've seen so far I don't think it's going
to happen.

> ...
>
> > *Lack* of interoperability with James Clark's code in JING
> exactly is the
> > reason why we have this discussion. I think we should thank him
> > for actually *using* the grammer in RFC2396 for validation, so this
> > was finally uncovered.
>
> Frankly, I have found this discussion to be incredibly moot, and of little
> value.

I don't think it's moot.

Whether you like it or not, RFC2518 as it's published is broken because it
doesn't conform to what to other (earlier and more basic specifications)
say. It's a pity that this wasn't discovered earlier, but ignoring the issue
won't make it go away.

Received on Wednesday, 21 November 2001 13:19:03 UTC