Re: are many URIs ultimately relative? was RE: are 'cid' URLs relative?

On Fri, May 26, 2000 at 11:11:36AM -0400, Jonathan Borden wrote:
> Michael Mealling wrote:
> > On Fri, May 26, 2000 at 09:36:00AM -0500, Al Gilman wrote:
> > > At 10:53 PM 2000-05-25 -0400, Jonathan Borden wrote:
> > > [sorry to have to belabor the correction of a minor red herring...]
> >
> > Not really. This actually applies to several URI schemes....
> >
> 
> Ok then what about the "file" or "news" scheme as pointed out by Michael
> Rys:
> 
> > ... Another example would be two namespace
> > names using textually equal, absolute URIs that are not globally unique,
> > such as "news:ibm.aplsv" or "file:foo.txt". These URIs are absolute but
> not
> > globally unique because they depending on the associated newsserver or
> > fileserver, and may be different when dereferenced by the same process in
> > different contexts.
> 
> The point being that many commonly used URI schemes in fact employ context
> information for resource resolution and are in this sense relative.

Yep. There are several URI schems that are not complete in and of 
themselves. I'm writing one up right now. The Common Name Resolution
Protocol has a URI scheme with a case where the authority[1] part is
taken to be a list of network nodes taken from the users local 
configuration (much like the news scheme which is dependent on what
the user has configured as their nnntp server).

I think what your asking WRT to XML Namespaces is whether or not these
types of URIs are appropriate for identifying a namespace. I think
I would have to say that they aren't. 

Do these URIs  follow the semantics of the URI space in general? 
Yes. 

Does that mean that there may be some subset of the URI space that is 
innappropriate for identifying an XML namespace? 
I think so. 

How do you do define that subset?
By specifying that they have to be globally unique. 

Is that acceptable to everyone here? 
I doubt it....

In other words: I think the correct solution here is that the XML
Namespace document must say that Namespaces must be identified by
what the ABNF calls "absoluteURI" and that the URIs used by XML
Namespaces must be globally unique. But then again, I'm of the camp
that thinks you eventually want to resolve these things into some
sort of document that describes the namespace.

-MM


[1] RFC 2396 ABNF http://www.ietf.org/rfc/rfc2396.txt

-- 
--------------------------------------------------------------------------------
Michael Mealling	|      Vote Libertarian!       | www.rwhois.net/michael
Sr. Research Engineer   |   www.ga.lp.org/gwinnett     | ICQ#:         14198821
Network Solutions	|          www.lp.org          |  michaelm@netsol.com

Received on Friday, 26 May 2000 11:46:37 UTC