Wed Sep 24 13:25:17 CEST 2003
(Quickly googles L4 and finds an implementation of Linux running on an
L4 microkernel. Cool!)
Now that I'm a bit better informed, I don't think the underlying
microkernel should inhibit anyone from having a stab at it. The kernel
primitives ought to be sufficiently equivalent to enable replacing one
with the other without much impact.
Slow at this stage is not be my primary concern. The practicability
question is more interesting to me. After that, faster microkernels may
be tried to suit real time applications.
Vlad Dumitrescu wrote:
>>Why not start from the outset with the Mach kernel, as are the Gnu-HURD
>>team, and build a native Core-Erlang system on top of that?
> I'd first go with making Erlang a virtual OS, seems easier.
> [That's a relative measure anyway... Just trying to split apart parts of ERTS is
> a lot of work. I tried without success the naive solution of just removing
> selected applications from the lib directory :-) ]
>>Mach provides the process scheduling and message passing primitives.
> Are they fast enough? I feel L4 matches better...
More information about the erlang-questions