W3C home > Mailing lists > Public > www-tag@w3.org > July 2002

RE: TB16 Re: Comments on arch doc draft

From: Bullard, Claude L (Len) <clbullar@ingr.com>
Date: Tue, 2 Jul 2002 10:57:52 -0500
Message-ID: <2C61CCE8A870D211A523080009B94E430752B62C@HQ5>
To: "'Jonathan Borden'" <jonathan@openhealth.org>, www-tag@w3.org

I don't disagree with that.  It should not be a 
problem to go from a system of names that maintain 
identity within a local system to a nameset that 
enables identity and locatability in a different 
system.  That is a boundary issue and not that 
complicated given a recognizable boundary.

Identity is an emergent phenomenon arising from 
the process of identification, not the other way 
around.  It depends on the system the 
observer is situated in as to whether or not 
a name is sufficiently unique to establish 
identity.  The recent snafu with the federal 
databases using arabic names to request processes 
from the Saudi banks is a good example of why 
global names are desirable for systems that 
request services of each other for common 
entities.  Saying and doing are the rut there.

So yes, I understand why a URI is desirable. 
I also understand when it is not. Systems that 
insist on global names in all cases become 
non-cooperative and are usually devolving 
with respect to their environment.  Trying 
to modify the environment to behave in 
accordance with the viewpoint of any single 
entity is how reciprocal modification works, 
but don't expect cooperation from all of the 
other entities trying to do the same unless 
they have and recognize common needs, and 
agree on the common boundaries.  Sometimes, 
all that is necessary for cooperation to 
be sustained is a set of rules for crossing 
a boundary.


From: Jonathan Borden [mailto:jonathan@openhealth.org]

If you ever want to use a document on the Web (i.e. give it a URI), then I
agree strongly with Tim that the namespace names should (and would support
must) be dereferencable.

For the vast majority of applications using "http:" based URIs is the
simplest way to go. There are situations when a "non-Web" naming system,
such as ISBN, needs to be integrated into the Web, i.e. a URI needs to be
generated for otherwise non-web based resources. In such cases, relatively
few, and hopefully fewer in the future, URNs do seem applicable.
Received on Tuesday, 2 July 2002 11:58:30 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:55:52 UTC