[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Priority modeling with CSP?

Jim Davies wrote:

Adrian says:

I forgot to add that I think the above identification of priority and time only applies in an interleaving semantics. Of course, original pure CSP is based on an interleaving semantics. Indeed, that is one gloss we can give nowadays to the word "sequential" in the name which we may otherwise think to be an anachronism.

But when we are modelling a synchronous circuit in full detail, interleaving semantics is inadequate. But it is possible to add true concurrency into CSP with very little change including retention of traces.

And, (genuinely!) without wishing to start a debate, but simply
wishing to indicate a divergence in views, I would disagree.
I think that there is good reason for wishing to distinguish such a
notation.  Calling it CSP suggests that it has the same (fundamental)
notion of event as CSP, which it most certainly does not.

I take that point. And have sometimes worried about it as well. I call it HCSP. It look and feels like CSP in almost all respects. And with true concurrency removed, or parked, it then reverts to CSPP which is very close to a superset of CSP.

I agree that the extensions are disturbing to someone schooled in conventional CSP. But the ideas of CSP are remarkably robust with respect to larger interpretation.

Any suggestions for another name? Deviant_CSP perhaps? :-)

Having said all that, I am not so sure as you that the event model is *really* different. My events always involve the exchange, and thus the communication, of information. The fact that some processes are passive can just be regarded as a choice not to block which, as far as the model is concerned, they might have done. They are still regarded as having synchronised - very quickly.

I too don't want to get into an extended debate :-) Also genuinely.

Dr A E Lawrence