On Thu, 2003-02-20 at 15:54, Stavros Macrakis wrote:
> > To my mind, there are six lisp implementations that are logical
> > possibilities for maxima: GCL, Clisp, CMUCL, ACL, SBCL and OpenMCL.
>
> Supporting multiple Lisp implementations is nice, but it is also work.
> I admit I am not up on all these systems; could someone summarize what
> the rationale is for each of them? HW/OS platform support?
> Performance? Development environment? Other facilities (graphics?, RPC
> support, ...)?
This is off the cuff, so don't take the details too seriously:
---------------------------------------------
GCL is fairly fast and works on a wide variety of platforms. As of GCL
2.4.x, it was not very close to the ANSI standard, although that
situation is constantly improving.
Clisp is available on a wide variety of platforms. It is very close to
the ANSI standard.
CMUCL is available on a limited number of platforms, but it is often
substantially faster than the lisp implementations above. It is
reasonably ANSI-compliant.
ACL is polished, fast, ANSI compliant and, unlike CMUCL, available on
Windows (and other platforms). It is the only commercial product on this
list.
SBCL is a fork of CMUCL. I won't try to describe the differences.
OpenMCL is a fast ANSI lisp implementation available on MacOS X.
---------------------------------------------
No one lisp is completely superior to all the others.
I have found the ability to compare maxima on multiple lisp
implementations invaluable, particularly when trying to decide the
difference between a Maxima bug, a Lisp bug and a standard
misinterpretation. Adding more lisps to the list isn't really that hard.
Furthermore, other people are doing it, not me. (!)
--Jim