Testing the waters.

[William A. Barnett-Lewis]

ET wrote:
> 
> Suppose my company were interested in developing LispOS in
> this manner:
>    Native i386 code,

No doubt. 

>    no underlying unix (lisp down to the metal),

I like, but I think that the flux toolkit is more likely to get quick
results w/out being tied into unix-isms. This could also allow better
integration of the VM/GC bits of memory management.

>    start with Scheme as basis language,

Prefer Common Lisp instead. It's big & ugly, but until the R5 or R6
report standardizes a scheme library, it does work that scheme can only
dream about. Plus CLOS ties into any kind of persistant object system
much better. 

>    design with support for persistent object system,

Have you looked at SHORE? http://www.cs.wisc.edu/shore/ is it's home
page. It's designed as a file system replacement and would work well
with CLOS.

>    implementation tools running on '95/NT,

Do you mean delievery tools (ie I can make a win32 .exe) or do you mean
a version that runs under win32? The former is fine, use cygnus' gcc as
a base. The later would be ugly.

>    free software.
> 
> How many people would be interested in actually investing
> time and energy on this?  If there is a critical mass, I'll
> start the project.

In principle, hell yes, I'm ready to go. OTOH, there are still some
questions that need answers before even basic design can proceed.

William Barnett-Lewis
wlewis@mailbag.com