[Oz] Re: Tapping a watcher on the shoulder

[Peter Van Roy]

> Ok, I see your point. But this is somewhat different from what you started with.
> My impression was that the watchers did not work for you at all.

So far we find that the watchers work fine.

> I assume that perm is what you're looking for and that you excluded temp
> form your scenario. How should the interface look like. I assume that you
> want to "ping" a specific entity and its watcher:
> 
> {Fault.triggerWatcher <Entity>}
> 
> Fault.triggerWatcher is then assyncrounous and issues a message to be
> sent, thus triggering the fault mechanism.

Yes, that's the idea.  I would rather call it 'nudgeWatcher', though,
because it may not actually cause the watcher to be triggered (if the
<Entity> is ok, nothing happens).  It should be an optimization, i.e.,
programs should still be correct if it is omitted (but maybe slower).

> > > Could you give me an example that gives this behavior you describe, so we
> > > can have a look at it. I assume that you are using the latest version of Mozart.
> > >
> > > /Erik
> >
> > Yes, Metwally is using the latest version.  He can give you more details.
> 
> Please give me all the details. I realy want to understand the scenario you are
> targeting.

Metwally and I will send you a detailed explanation.

> Nice to see that you're using the Distributed Subsystem. I'm impressed of its
> stability.
> You have not reported any bugs yet....
> 
> /Erik

So far we found no bugs in the subsystem, only in our fault tolerance 
abstractions :-).
Metwally is going over Ilies Alouini's GlobalStore with a fine-toothed comb, 
and he
has fixed and improved many things.  In particular, the old version had a bug 
in the
recovery process that made it take very long to recover from a server crash.  
Our
current prototype recovers much quicker.

Peter

-
Please send submissions to users at mozart-oz.org
and administriva mail to users-request at mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.