RE: now://example.org/car (was lack of consensus on httpRange-14)

If we keep to this path, that dereferenceability 
is the critical property, then URIs are URLs and 
are always capable of being dereferenced.  That 
being the system definition, then this debate is 
simply going to yield a tautology as has been 
said over and over and over again.  To me, this 
validates what Bray and so many before him have 
tried to say:  it isn't an architectural issue 
because the system works the way it works if the 
system, ie the web, is defined as a URI-based 
set of dereferenceable resources that return 
representations or act as control arguments 
to a resource (for the case of robots).

It is a social issue:  put SOMETHING at the location 
the URI points to and be responsible for it.  Because 
it comes down to responsibility, if the owner wants 
to declare a name and use a name, they are responsible 
for the problem of changing that over to a proper 
location-based address.   All that is needed ane 
really, all that is really sensible is 
what is there now: strong advice to use an http address 
as a future-proofing of the name.

len

-----Original Message-----
From: Paul Prescod [mailto:paul@prescod.net]

> Why "never"? (except for the problem that Microsoft is using unregistered
> URN schemes...)

I don't believe that universal URN registration schemes will be widely 
deployed. I've been wrong before, but even if I am, wouldn't it be funny 
if people flee away from HTTP to schemes that "do not imply dereference" 
and then sometime in the future they imply dereference.

Received on Thursday, 10 October 2002 13:37:47 UTC