[8779] in www-talk@info.cern.ch
Re: State and tracking need to be separated (Re: session-id redux)
daemon@ATHENA.MIT.EDU (Paul Burchard)
Sat Jul 29 09:53:32 1995
From: Paul Burchard <burchard@horizon.math.utah.edu>
Date: Sat, 29 Jul 95 07:50:30 -0600
To: www-talk@w3.org
Reply-To: burchard@geom.umn.edu
Resent-From: www-talk@w3.org
I said:
> the current agent-level model of the Web is rather vague
> about state
It would be irresponsible for me to say this without also referring
you to Dan Connolly's extremely non-vague formalism for the Web:
<URL:http://www.w3.org/hypertext/WWW/People/Connolly/drafts/formalism.html>
However, I believe this model needs some refinement. For instance
-- somewhat surprisingly -- it does not define any concept of
linking, only of objects ("entities") and object addresses ("URIs").
Also, the only stateful (time-dependent) part of this model is the
"representation" of URIs by entities, and even that is strictly
under the control of timed expiration. Thus, the entities
themselves are specifically _not_ stateful (probably a good choice).
--------------------------------------------------------------------
Paul Burchard <burchard@math.utah.edu>
``I'm still learning how to count backwards from infinity...''
--------------------------------------------------------------------
