[e-lang] [Fwd: Re: inter-vat promises]

[Kevin Reid]

(MarkM: you might want to forward this if Tony Finch isn't subscribed  
and you haven't said this yourself)

> Date: Tue, 8 May 2007 21:10:50 +0100
> From: Tony Finch
>
> Yes it does, though it makes me wonder what happens to an attempt  
> to resolve a promise that is already resolved.

The resolve/1 operation throws; the resolveRace/1 operation returns a  
boolean indicating whether it succeeded.

http://wiki.erights.org/wiki/Resolver

> This usually happens when a turn completes and the only result of  
> the turn is to resolve the promise, so the error can't be reported  
> to the resolver.

No; for the promises which are implicitly created by eventual sends,  
the resolver is held only by, and not revealed by, the internal  
eventual-send subsystem, which would have a bug if it tried to reuse  
a resolver.

> It also can't be reported to the holder of the promise because  
> resolved promises can only be broken by network partition. So is  
> the result from the loser of the race just discarded?

Only if the sender of the resolve message ignores the result.

> Also, I'm not entirely sure what the LocalPromise <-> RemotePromise  
> transitions in the reference state transition diagram refer to. How  
> are remote promises different from a promise with one end passed by  
> proxy to another vat?

A remote promise is a promise which forwards its messages to another  
vat for pipelining. Remote promises result from passing *promises*  
over a comm system; passing a resolver results in a simple far ref to  
that resolver.

http://wiki.erights.org/wiki/Remote_promise
http://wiki.erights.org/wiki/Proxy

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