Re: Efficacy of rule specification, processing

I am not so sure it makes much sense to add environment variables like
local system load, # of client connections etc. to IRML. Consider for
example an IRML rule module authored by a content provider or a client.
Why should a content provider or a client have to worry about the system
load of a specific OPES device when authoring IRML rule modules? Also,
wouldn't an IRML rule evaluating the system load be OPES
device-specific? How can this work if the same rule modules are to be
distributed to different OPES devices?

Therefore I think we should not add any OPES device-specific conditions
like system load to IRML. If we allow late service binding in IRML, then
the OPES device itself can handle things like load balancing. 

-Andre

Gamze Seckin wrote:
> I guess streaming people already know that dynamic adjustments can be done 
> through the streaming server and there is actually no need for an 
> intermediary. However if there are rules that make life easier and standards 
> based, why not?
>
> Anyway regarding state parameters may be we can discuss possibilities:
> 1- local system load (i guess many agree on that)
> 2- call-out server load ?
> 3- end-device-to- local system bw ?
> 4- local system -to- call-out server bw ?
> 5- ?
>
> Gamze Seckin, Ph.D.
> Luxxon Corporation
> www.luxxon.com
>
> ----- Original Message -----
> From: "Markus Hofmann" <hofmann(_at_)bell-labs(_dot_)com>
> To: "Yang, Lily L" <lily(_dot_)l(_dot_)yang(_at_)intel(_dot_)com>
> Cc: "Maciocco, Christian" <christian(_dot_)maciocco(_at_)intel(_dot_)com>; 
> "'Gamze Seckin'" <Gamze(_dot_)Seckin(_at_)luxxon(_dot_)com>; 
> <ietf-openproxy(_at_)imc(_dot_)org>
> Sent: Tuesday, June 12, 2001 4:21 PM
> Subject: Re: Efficacy of rule specification, processing
>
> > "Yang, Lily L" wrote:
> >
> > > Many of the dynamic behavior can be achieved with a set of carefully 
> > > written
> > > rules -- the rules are static but the run time matching will result in
> > > dynamic behavior accordingly.
> > And this run time matching requires a set of state variables - the
> > question was which and how many. For example, should "local system
> > load" be a run time variable for the rule language?
> >
> > -Markus