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

Re: continuations over the network

On Tue, Jun 10, 2003 at 03:21:49PM +0200, Michael Sperber [Mr. Preprocessor] wrote:
> Francois-Rene> Problem is, Kali did so much for you that you had
> Francois-Rene> little control.  If you want a system where the
> Francois-Rene> programmer can explicitly reify continuations and
> Francois-Rene> migrate them over the network, try David Alan Halls'
> Francois-Rene> Tube.
> Huh?  (Not that this is pertinent to Scott's question.)  Reification
> and migration are explicit operations in Kali as well.  (Reification
> is an explicit operation in Scheme per se.)

Well, I meant marshalling/serializing/whatever in a user-visible way.
IIRC, Kali considered all nodes as part of a same uniform pool of processors,
and would do its own background load-balancing of processes. The user-visible
migration primitive would advise the system to move a few stack frames worth
of computation to the destination processor, with no guarantee as to where
any specific computation would actually take place.
In the Tube, you control exactly what happens where, and you can explicitly
migrate the current thread's continuation from one place to another one,
with all its internal state.

[ François-René ÐVB Rideau | Reflection&Cybernethics | http://fare.tunes.org ]
[  TUNES project for a Free Reflective Computing System  | http://tunes.org  ]
(define (cthulhu_fhthagn)          ;   Iver Odin Kvello, iverk@hfstud.uio.no
   (call/cthulhu  (lambda (destroy) ;   Lambda, the Ultimate Horror
      (if (stars-are-right? (now)) (destroy 'everything) (cthulhu_fhtagn)))))