Re: One Party Consent Model

At 20:46 14/07/04, Markus Hofmann wrote:
> See your point, but the example you outline above violates the OPES 
> tracing requirement, which says that Markus (or B in the generic example) 
> must be able to trace the OPES service. I.e. I would be aware of the OPES 
> service that removed me.
> Now, I'm not yet sure how to achieve this and whether this is the 
> full/correct answer...
This is excactly my point and that responds Alex's question over my more 
general comment (Alex thinks "how". I think "what").
- I am not sure how to achieve this - IMHO this cannot be achieved by an OPES
- is what we say a correct answer ? because we debate within an 
"end-to-end, dumb network/smart host, protocol on the wire" vision of a 
network system, which architecturally totally opposes our target of 
external-change-in-multiple-alternative-routes
This is why I quoted the Judge who had a parallel problem and who gave it a 
solution I can architecturally accept : considering the mail when 
temporarily stored at an Agent. There is no difference with HTTP (and Alex 
there is right): in both cases what we discuss is changing the header and 
the payload when stored on a host. The particulars of HTTP blurred that 
debate (my initial discussion) because it permitted to consider the stream 
as a virtual host. The particulars of "one to interacting many" of mail 
services do not permit that.
I fully agree with Alex: there will be no drastic change in OCP and P etc. 
But this will not be OPES and there will be impossible problems of 
interoperation with the interconnected interapplications - as mentionned 
above. Authorizations, Error reporting, cancellations, priorities, delays, 
orders, authentication, application interface, LHS acceptation, security, 
authoritative domains, relation with the DNS, resends, archiving, 
inter-OPES operations, server failures, spam, etc.
Just one small question. Let assume a change is conditional to time. What 
is the time we chose as a reference. The time the mail was entered, the 
time it was sent, the time it reached each OPES filter, the time it reached 
every OPES filters, do we take into account the time the first mail or the 
last mail reached the filter? Let assume a very common case : authority 
changes at a given time (a different watch, a justice decision, etc.) and 
also the management rules.
jfc