RE: Is a namesapce a resource? - was: duck

:> But more than that: the question if you take a step back is 
:clearly absurd:
:> "why can't a namespace be a mailbox?".  Just in natural langugae you
:> can see that they are completely diffeernt concepts.
:
:Well, I agree.  A namespace is not a piece of hypertext, or a mailbox,
:or an interactive connection to an Internet host.  Yet an 
:http: or mail:
:or telnet: URI is syntactically valid as a namespace name.  

URIs _are_ syntatically valid as strings that are understood to be
namespaces. That doesn't imply that they are semantically valid. What does
the conflation win us in order to justify the confusion? Because the essence
of this is that we are conflating strings for names with strings for things.
The logical dismabiguation of Names and Things has been one the great
intellectual leaps of this century and  the last. Conflation seems like a
return to the kinds of confusions the Scholastic philosophers had to put up
with. 

Mike Kay said it best on xml-dev:

"
Except the question "why?". I still find it quite inexplicable that any sane
XML standards committee could have recommended using names beginning with
"http:" for things that were not accessible via HTTP, thus confusing vast
numbers of people into this very misunderstanding.
"


:I have no problem with this, as I don't identify "name" with "URI" as 
:properties, any more than I think Greg Bear (the writer) is really a bear.

Well surely there _is_ a problem. You had to tell us that "Greg Bear" points
to (as an act of naming) the writer whose name is Greg Bear, who is not a
Bear. Using "Greg Bear" alone isn't sufficient as an identifier for Greg
Bear (the writer) to disambiguate a Writer from a Bear. What about Greg Bear
(the bear not the writer)? Are we even sure that we are talking about the
same Greg Bear (the writer) when we use the string "Greg Bear". The author
of the books which have the isbn which is identified in turn by the string
I'll mention as ""057560266X""? Same guy? Are we sure? Will machines be?
It's fun when Douglas Hofstadter writes about such matters, it's less fun
perhaps trying to build interop systems over them.

Now if we can't easily pass simple names about, or indeed accept that the
act of naming alone is not fully consistently understood (it's certainly an
open philosophical and liguistic issue), how on earth can we expect people,
including the particulary bright and eminent people on this list such as
yourself and as importantly, people like me, as well as machines (lets not
forget them), not to get mixed up over the conflation of namespace and URI
strings? 

I'm baffled: why can't we just say that namespace strings that incidently
look like URI's aren't and _never_ should be treated as URIs, it just that
URIs are handy way of generating unique strings, and all go home? If that's
not enough and namespaces _are_ that important to the future of the web,
then push for a component that can be inserted into a URI, to identify that
URI as a namespace name as opposed to a possible resource.


-Bill

Received on Thursday, 8 June 2000 09:45:32 UTC