LispOS compiler question

[Fare Rideau]

>: Emergent Technologies

> What would be a good compiler to use for writing a LispOS?
>
The LISP FAQ is meant to list all possibilities.
My page Tunes/Review/Languages.html#LISP tries to focus on interesting
freeish systems (not all are compilers (GUILE) or really free (ACL)).

> I think the following goals would be nice:
> 1.  Free
> 2.  Fairly easy to hack (so the underlying lisp architecture can
> change).
> 3.  Object code portable to Linux and Win32 (for bootstrapping
> reasons).
> 4.  Support proper tail recursion (not strictly necessary for Lisp,
> but required for Scheme).
> I don't think it would be necessary to have a highly optimizing
> compiler.
>
Again, lots of choice depend on language choice.
The only currently active cooperative effort for a free ANSI CL compiler
with enough system support to make a LispOS is CMUCL (www.cons.org/cmucl).
The only <etc> for Scheme is RScheme (www.rscheme.org)
Now, if you want to start your own dialect and implementation,
you're free, but there's the need to define a dialect, first!

(2) depends on the notion of "fairly easy". To most people,
 this seems to preclude ANSI CL (CL has many advantages, tho).
(3) (portability to Linux and Win32) should be made easier
 thanks to Cygwin32.
(4) Mostly restricts you to Scheme systems, hence RScheme.


> What do others think?  Would anyone be able to take some spare time
> and write a wimp-ass compiler?
>
I think the only conclusion from former discussions was that
(1) many of us joined with completely different goals, so that
 there was no possible common decision.
(2) no clear subgroup of same-minded people even formed, so that
 everyone has the entire freedom and responsibility to head one's efforts
 toward the direction that one thinks the best,
(3) until some new event can make our interests converge, we cannot
 and must not rely on other LispOS to make things actually better,
 including making such an event happen. We're on our own.

Of course, I'm sure all LispOS subscribees have been studying and
experimenting their own approach to the problem; just don't expect
that the approach you'd prefer will have silently progressed in your back.


-- #f