RE: Namespace name: better to use URN or URL?

[Nicolas LEHUEN <nicolas.lehuen@ubicco.com>]

Title: RE: Namespace name: better to use URN or URL?

Well in fact you're right, building a unique URN is as "easy" as building a unique URL, since you can use a naming convention inspired from Java package names (reversed domain name).

 

Regards,

Nicolas

-----Message d'origine-----
De : David Hunter [mailto:david.hunter@mobileq.com]
Envoyé : jeudi 30 août 2001 18:03
À : xml-dev@lists.xml.org
Objet : RE: Namespace name: better to use URN or URL?

This is how I feel too, however, I think we may be in the minority.  Maybe the difference is the word "philosophically" - in my philosophy, a URN provides a better fit, but I think many (most?) people prefer a URL, since they can then put something there, which will provide a benefit in their situation.  Philosophically, they look at the whole situation, and specifically XML Namespaces, differently than I do.

(The point has also been made that it's easier to create a unique URL than a unique URN, but I'm not convinced that that's true.)

Side note:  RDDL changes things a little bit, for me.  If I was going to describe my namespace in RDDL, I would have no trouble, philosophically, using a URL for my namespace name.  Well, maybe a little trouble, since I would still be a little uncomfortable with the tying of name and RDDL document location...  (From a puritanical standpoint, I would probably prefer some other kind of attribute that points to an RDDL document.)

David

-----Original Message-----
From: Bullard, Claude L (Len) [mailto:clbullar@ingr.com]
Sent: Thursday, August 30, 2001 10:59 AM
To: Roger L. Costello; xml-dev@lists.xml.org
Subject: RE: Namespace name: better to use URN or URL?

Philosophically, I prefer a URN.  As you say, it
is clear what it is, and practically, the systems
I work with don't color it blue and make it clickable;
that is, it isn't a hyperlink.  Practically, some
of us are tool thralls and the tool vendors
make that choice for us.

We've debated it endlessly, but I think using a
URL binds it to the web by using a system
specific location identifier in the namespace,
and that pollutes the namespace in situations
where the information could be reused on non-Internet
systems.  One has to decide if the information
is ever to be repurposed off the internet. 
When we had this argument some time ago, it
came down to, do you ever plan to resolve it?
If it is *really just a label*, no winking,
no nudging but honest clean design, one has
to choose for themselves and decide if

http://anything

really is a name or a location.  Identity
is not the horse you want to ride for that
argument.  Systems sit inside systems and
that is the problem and solution to describing
a namespace.

Len
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

--------------InterScan_NT_MIME_Boundary--