coordinating efforts..
Maneesh Yadav
cj@utpulse.com
Mon, 25 Jan 1999 16:45:14 -0500
>
> I believe the main point is to get a few design principles for
> something *much less ambitious* than TUNES, but eventually able to
> bootstrap TUNES. In my opinion (but of course, I do expect other
> people to try something really different) this means a 32 bit
> kernel with a very few syscalls, capable of supporting a persistent
> object system. This also means defining a middle level langage. (I
> am not sure that R4RS Scheme is the way to go!)
>
All I am going to say is that making a kernel is just not a good idea right
now...not at all. Think about all the issues we face:
distrbuted computing
multiprocessers (SMP, NUMA etc.)
rewriting of device drivers
making a model that elegantly supports all these things
making a model that isn't hooked to a particular system to allow multiplatform
capability.
These issues IMHO, are not what TUNES is about. We are working on a
metaprogramming system no? Why not design the meat parts of the OS with the
meta programming system we are aspiring to make?
Persistence is definately going to be a part of Tunes, but why not use an
existing OS's file system initially for that, THEN we can bootstrap it.