OS design...

Eugene Zaikonnikov viking@cit.org.by
Sun, 18 Oct 1998 14:19:09 +0200

Martin B. Pomije wrote:
>Francois-Rene Rideau wrote:
>Has anyone on this list investigated the techniques used in Kali Scheme?
>Could this sort of technique be used for a practical OS?

Yeah, it's quite neat, although Kali's distributed GC can't deal with
network failures. But it worth to look at: both procedures and continuations
(threads) can migrate between so-called 'aspaces', and aspaces may lay on a
same machine, or on a different CPU's of multiprocessor, or on a different
network nodes. And all that stuff relies on Scheme constructions (e.g.
continuations) in a natural way. There's no new special syntax  introduced
and all control stuff quite easy and high-level, so anyone could cope with
it in a couple of days.