RE: XML Public Indentifier

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

It's a warping definition of "need".  It 
seems you are saying this is all application 
dependent.  I tend to agree with that but add 
that preserving the option is good given that 
one might want to strip the URLs, replace them 
with URNs, and archive the puppy.  I am thinking 
about applications that have to purge information 
but might at a later time have to retrieve it 
given a change of policy.
 
That is painful with a relational database given a 
lot of keyed relationships. 

RDDL is handy.  So is a catalog.  In what situation 
is one to be preferred over the other?

I understand the namespace density problem 
and always have.  But we can't spend two 
years telling the world a namespace is just a 
label then tell them it can also be a URL 
in every sense of the definition.  It shocks 
the monkey tree.

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

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


-----Original Message-----
From: Tim Bray [mailto:tbray@textuality.com]


At 01:56 PM 05/09/01 -0500, Bullard, Claude L (Len) wrote:
>If the author of the namespace spec can say 
>on one hand that it is just a label, and on 
>the other hand can say he won't use a URN 
>that can't be dereferenced,  isn't that a 
>contradiction?

They don't *need* to point to anything to do the job
for which they were defined.  They are just labels.

Labels that can be dereferenced to get something to
tell you something about the namespace, in the case
where you don't know anything, are better than those 
that can't [in some applications at least]. 

I didn't realize this when we were doing the namespace
work, and at that time would have been friendlier to 
URNs than I am now.  But RDDL [or something like it]
seems awfully useful in a namespace-dense world, which
seems to be the one we're getting. -Tim