Fri, 02 May 1997 18:39:01 -0700
Tim Pierce wrote:
> Here is a possible breakdown of the LispOS group into working
> groups. Comments welcome, but I would like to see these projects
> start to crystallize quickly.
> * Ground-up kernel design.
> Hack a Lisp to operate as the kernel for a system
> toolkit (or e.g. a process server for a microkernel).
> The most popular option seems to have been using the
> Flux OS kit and linking it with a Lisp or Scheme to
> create a minimal kernel, but this could also be the
> design of a Lisp personality for the Hurd as some have
> suggested. It depends on how much labor and resources
> are available, and the skills and enthusiasm of the
> people involved.
> * A hacked Unix kernel (Linux, FreeBSD, or what have
> Begin with a minimally running Unix system (as opposed
> to a non-self-hosting toolkit or microkernel). Link the
> kernel with a Lisp and begin throwing away undesired
> Unix facilities, rewriting them in Lisp (or modifying
> them to use Lisp facilities) along the way. This system
> differs from the first in that it would probably start
> booting immediately, but some are concerned about the
> ultimate system suffering from too many inherited Unix
> * Lisp environment on top of a complete Unix system.
> This approach would make minimal changes to the Unix
> kernel and/or tools and would focus instead by providing
> a complete Lisp application and development environment
> to run entirely on top of an existing Unix system. The
> Lisp world could even conceivably run entirely in user
> space, and merely translate user requests for services
> into the underlying Unix calls and processes. It might
> dominate the system (requiring a lot of concessions to
> get Lisp to run efficiently) but may also permit
> Lisp-Unix cooperation.
> These three projects roughly encompass the different perspectives
> expressed on the LispOS list recently. I don't think that there
> are any significant omissions, but may be wrong.
> Can we have volunteers for project coordinators? I will volunteer
> for the first if no one else does -- someone with more system
> hacking experience should probably take the lead, but it is
> important for someone to step in.
I do not know about coordinating, but I can provide a lot
of real life experience on project #3 since my current job
is very similar to this (except that instead of a LispEnv
we do have a LispGraphicalEnv). I wonder if #3 shouldn't
be more cross-platform than just target UNIX.
Luca Pisati Voice: (310) 577-0518
Nichimen Graphics Fax: (310) 577-0577
12555 W. Jefferson #285 EMail: email@example.com
Los Angeles, CA 90066 Web: http://www.nichimen.com