----- Original Message -----
From: "Hilarie Orman" <HORMAN(_at_)volera(_dot_)com>
To: <ietf-openproxy(_at_)imc(_dot_)org>
Sent: Friday, June 08, 2001 12:15 PM
Subject: Re: Efficacy of rule specification, processing
While there might be systems with no requirements for dynamic
rule uploading, there are others that simply cannot be managed
without that capability.
I'd like to distinguish some types of rule dynamicism, though.
The simple one is "active/deactivate rule". The other is
"incorporate new rule." And, at the workshop, we noted
that changing the state of a variable in a rule activation
can be a substitute for adding a rule. I.e., if we have
the variable "list of banned users", then the action
"add Alice to the list of banned users" and the
rule "if current_user is a member of banned_users then
deny access" is equivalent to the rule
"if user is Alice then deny access".
This hints at why I believe that we need to define
some common set of state variables for actions to
manipulate.
I aggree. I guess the question would be then how and when those variables will
be communicated.
In streaming case, as talked on the workshop, there might be no need for
streaming rules (since static user profile will be stored somewhere else).
But if there will be dynamic state variables - and efficient way of
communicating them to the opes client- then a rule can for example use those to
say something like that: "if nw bw is x (or has changed to y) stream with
format/bit rate A/or B" or "if client has resolution x (or has changed to y)..."
Gamze Seckin, Ph.D.
Luxxon Corporation
www.luxxon.com