Newbie comment on URNs and URLs- (RE: Paul has volunteered (was Re: Overloaded URIs must GO!))

RE6 Computer Services re6 at
Mon Jun 14 10:26:34 BST 1999

Hash: SHA1

>-----Original Message-----
>Didier says:
>Here we are. I remember a good discussion (within IETF groups) we got
>URIs and a lack of a RFC explaining the role of each. We didn't
>consensus and the result today... we are right in the problem :-))
>I remember that the main point is the name. Read attentively the
>name: U_niversal R_esource L_ocator
>The name itself self describe in the sense that it is used to express
>location, a "place".
>At contrario, a: U_niversal R_esource N_ame
>Is a "name".
>The former explicitely contains the meaning of a location. The latter
>explicitely the meaning of a "name" and not of a location. For
instance, I
>may use the URN convention to express the inventory record
identifiers or as
>identifier for some collection organized in a typical name space.


>But, if we want to express a location with a URL, its made for this.
So, if
>the name space reference uses a URL and that no document is at the
>end, it is like a 404 error (in the case of the HTTP protocol). Or
>having a rendez-vous but nobody came to it :-)
>Didier PH Martin
>mailto:martind at

Pardon for the interruption, but I've been following this thread, and
it seems to me that Didier has said it succinctly:  Use the URLs to
point to a URN.  The question of how to identify a document uniquely
is another matter, and one that is to be addressed by the URN, and not
the URL, correct?

Also, much ado has been made about the seeming requirement to have one
and only one "object" on a network.  In small to medium-sized
networks, such as a local network or a corporate network, that makes
sense.  In larger networks it doesn't make sense.  It is impractical
to jump 73 times to retrieve a document 12 Meg in size at the same
time 120,000 other people are trying to retrieve it.  Nor does it make
sense to make 73 jumps to retrieve an object that is only 18k in size.
 Notice the popularity of "mirrors".  On a network the size of the
Internet, one instance of an object is clearly not the best approach.
The key, though, is that each instance has a unique identity combined
with its URL.

An analogy:
URNs and URLs are related as telephone lines are related to entities:
I can have several telephone numbers, each one pointing to a unique
individual (me).  I would be considered a URN, and each of the phone
lines would be considered the URLs.  However, because I am unique,
that does not mean I have a unique name.  It is only unique when it is
combined with a unique identifier, such as a URL (phone number).
Obviously, URLs can only point to one unique object, period.  The only
reason that it isn't a good unique identifier is if the object's
*location* changes.  If I move to Florida, USA, then my phone numbers
change and I can no longer be uniquely identified by the old ones.
Nor does it help anyone trying to contact the prior owners of the
telephone numbers I inherit in Florida.

In a DNS, the domain name is coupled with an IP address.  HTTP
utilizes the IP addresses to find its way on the network; which leads
me to wonder if the DNS can also affiliate a domain name AND a URN to
a location instead of having another dedicated server provide the same
function? Therefore, I don't see a problem using HTTP to specify a URN
to retrieve rather than the URL, which changes just as long as the DNS
is updated with the new location.  Sort of like using "call
forwarding" on my old telephone lines to ensure that I continue to
receive phone calls to my old numbers (URLs).  After a specified
period of time, that service is discontinued; anyone not getting the
change information would have to relocate my telephone numbers which
is usually accomplished by using my name and other personal
information (URN).  Of course, one of the great things about IP
addresses and domain names is that it is much easier to change where a
URL points, especially if the hyper link uses a domain name rather
than specifying IP addresses.  After all, wasn't that one of the
intents behind creating domain name systems?

In other words, although a URL is a unique identifier, it should not
be used as part of the URN, a truly unique identification for that
object.  One last question: If an object is copied, is the URN
supposed to change, or is it supposed to remain the same?

My apologies if I have inadvertently offended someone's intellect, or
if I have violated any rules of conduct, or performed any other type
of rude error.

Robert Reese~
RE6 Computer Services
re6 at

"You can't strengthen the weak by weakening the strong." - Unknown
Version: PGP 6.0.2
Comment: Outrageous!


xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at
Archived as: and on CD-ROM/ISBN 981-02-3594-1
To (un)subscribe, mailto:majordomo at the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at

More information about the Xml-dev mailing list