[cap-talk] Is "Authority" Subjective?

[Toby Murray]

On Fri, 2007-06-22 at 17:26 +0200, Pierre THIERRY wrote:
> Scribit Toby Murray dies 22/06/2007 hora 16:12:
> > I don't understand. Alice and Bob both have permission to invoke
> > Carol.  She can't refuse invocations from either of them. This is
> > represented by the possibility of both aliceInvokesCarol and
> > bobInvokesCarol initially.  If Alice invokes Carol first, (ie.
> > aliceInvokesCarol occurs), then Bob has to wait until the invocation
> > finishes (ie. until carolRespondsToAlice occurs) until he can invoke
> > Carol himself by performing bobInvokesCarol.
> 
> I'm not sure that this CSP models effectively any used ocaps system. It
> seems to me that most of them, if not all, are concurrent. That is,
> while Alice waits for Carol's response, Bob can still send Carol a
> request.

Yes but Carol can't act on that request until she's finished with Alice.
So she doesn't see it until after then. Hence, modelling it in this way
doesn't seem a problem to me. 

While not advantageous, there are plenty of synchronous concurrent
systems out there. Consider models multiple threads trying to access a
"synchronised" object in Java, for example.


> > What I'm after is that if Alice can /ever/ cause Carol to be invoked,
> > then we consider that Alice has authority to invoke Carol.
> 
> That seems the way authority is described here so far. It's pretty
> conservative but you better be exagerating the authority of some
> subjects than underestimating it.
> 
Precisely. I'm glad we agree.

Cheers

Toby