[Shootout-list] Re: ring of processes

Aaron Denney wnoise@ofb.net
Tue, 12 Oct 2004 02:53:55 +0000 (UTC)


On 2004-10-12, Raymond Racine <rracine@adelphia.net> wrote:
> Could we clarify "send".  Let the process sending a message be the
> "sender", the process next in the chain the "receiver".
>
> Possibilities.

[sychronous / asynchronous]

I don't think it matters which the implementations use, and some
languages make it much more natural to write one or the other.

Accordingly, Bengt left it as an unspecified implementation detail.

Personally, I think it makes sense to seperate synchronization from
information transfer when possible.  Fortunately, concurrent Haskell
agrees, so using Chans or MVars will be easy.

I think this is a reasonable and simple test (though I'd make the number
of messages the source sends be an input N, and the number of processes
P be constant, as I think varying the messages is far more interesting
than varying the processes, but whatever) and should be the one used for
ring of messages, while Bengt figures out what he wants for the SYSV IPC
messages test.  If people agree, I can have the haskell implementation
out fairly quickly.

-- 
Aaron Denney
-><-