[e-lang] __reactToLostClient and distributed confinement

[Kevin Reid]

On Mar 1, 2006, at 2:48, Mark S. Miller wrote:
> Kevin Reid wrote:
>> Has the effect of the information provided by  
>> __reactToLostClient's  parameter on distributed confinement been  
>> considered?
>
> No. What problem do you anticipate?

I'm not sure. I once noticed that it supplies information that is not  
explicitly passed by the object's 'normal' clients, and so I wonder  
whether it reveals something.

I now suspect that this isn't a problem for distributed confinement,  
as in that case information is already leaked in by the /timing/ - or  
just occurrence - of deaths of vats making up the box. But are there  
any other situations?

-- 
Kevin Reid                            <http://homepage.mac.com/kpreid/>