Kernel Lisp Definition?

Dwight Hughes dhughes@intellinet.com
Wed, 14 May 1997 11:15:55 -0500


| From: Marc Wachowitz <mw@ipx2.rz.uni-mannheim.de>
| 
| "Dwight Hughes" <dhughes@intellinet.com> wrote:
| > From initial browsing I think the EuLisp Definition version 0.99 at:
| >        <ftp://ftp.maths.bath.ac.uk/pub/eulisp/definition/>
| > would be a good place to start. It is very clean with quite
| > distinct Level-0 and Level-1 definitions.
| 
| For a more modular application-level Lisp it's a nice start, except for
| missing parts, like a useful macro system or declarations for efficiency,
| and such nonsense as the inability to do general character comparisons.
| For a Kernel Lisp, I think it's too high-level, and doesn't leave much
| room e.g. for the recently discussed immutability and freedom to share
| or clone some immutable data.
|  
| -- Marc Wachowitz <mw@ipx2.rz.uni-mannheim.de>

I agree. The reason I presented it as a possible starting point is that
its definition is one of the least baroque I've seen, therefore, changing
it to suit our purposes should not require tearing down the walls of
Jericho like trying to start with, say, the CL spec. I didn't mean that
we should necessarily build on top of it. Its modules are not quite
what I would want either and have limitations as specified.

Its implementation puzzles me - for nearly every class of object defined
there is an associated *.h,*.c, and *.em file. I guess they preferred
this to defining an integrated VM with the same support.

-- Dwight