Doing it better 1999-08-02 c
###

From Reuben.

> ...which I'd like you to write up.

Perhaps while I'm in France. Perhaps next month...

> That sounds like one idea. I'm beginning to think, based on my recent
> work, that we should do something rather wierd.

In that case, we should also do something rather normal. One approach that
attracts me is to have a normal Forth environment in parallel with the
rest, built on my Forth-like base. This is a nice standard thing, and also
is really easy to build (even better, I've done it before) so shouldn't
distract too much from the wierdness. Rather than compiling Forth into
your world, it would diverge at the VM level, but my framework for VMs
is plenty open enough to allow this, either as two separate VMs or as the
same one with two sets of instructions.

> Implementing it is much easier than implementing a Mite-like system,

How's that? Is it just that nodes are really easy to compile, along the
lines you gave before? Seems like it, from what you say below.

> Linux is an excellent base. I advocate abusing the Linux kernal to steal
> its device drivers and market penetration, but to ignore the scheduler and
> memory protection and stuff which we'd prefer to write in Tau. We
> certainly want to diverge before the login prompt.

Exactly. Use the kernel for hardware support, and rip out the rest. The
other nice thing about this is that we can develop on top of Linux,
emulating the memory/processor stuff, and then dive underneath and
implement it. Along the lines of the new Amiga, I think we should pick a
standard, simple hardware configuration (something like Pentium
instruction set, PCI bus, IDE disk, VESA graphics, PS/2 keyboard and
mouse) which applies to almost everyone's computer but lets us strip out
90% of the drivers.

> Quite. Java proves that evan an "open" system can be defended vigorously
> if it becomes necessary.

Yeah; it's easy anyway, as it's our property, and we are not trying to
control it in the sense of foist our version on everyone else. I suggest
we GPL the whole lot; if anyone gets interested and wants to get involved,
fine; if they don't like the way we're doing things, fine, they can do it
their own way. Obviously the ideal thing would be for our version of Tau
to catch on, but even if someone else's does, that doesn't matter, and
will probably only help us in the long run.

--
http://sc3d.org/rrt/ | certain, a.  insufficiently analysed