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

Re: Poison


I think Marcel's latest note makes it clear that we should 
stick with standard occam channels and processes - this is
getting way too complex and various for a canned solution.

A static examination of the code can generate a dependency
tree that can show all POSSIBLE parallel processes, though
it may not be possible for them actually to exist at the
same time. Could something like this be used as a tool in
a two-stage link? In my experience, real-life programs are
not terribly complex when examined this way.


>From owner-occam-com-out@xxxxxxxxx Fri Jul 20 05:19:37 2001
From: M_Boosten <mboosten@xxxxxxxxxxxxxxxxxxx>
Date: Fri, 20 Jul 2001 14:17:45 +0200 (MEST)
To: java-threads@xxxxxxxxx, occam-com@xxxxxxxxx
Subject: Re: Poison
X-Sun-Charset: US-ASCII
Precedence: bulk


I do very much like Gerald's pragmatic viewpoint on the use of exceptions
to distribute "poison" in the Java-based CSP solutions.

I'm wondering about terminating subsystems.

Suppose you have got 5 central servers, and you want to shut one down
for maintenance reasons, so you poison it.  In that case, you do not
want to poison the whole network; you just want to shutdown part of the

In general: How do I poison the right subsystem?

Maybe there are more than just the two tokens: POISON and RESET.
Maybe there is also POISON_KENT, and RESET_TWENTE, to avoid them to
either poison or reset eachother via the java-threads mailing
channel. :-)

Maybe poison should have a scope, a certain process for example...?

What do you think?