We would like to hear some experiences from Lisp users who worked with AKCL
(good and *especially* bad news).
So far we know AKCL generates C-code from lisp sources which should make it
very easy to make ports for different processors. Since we consider
a port to the unusual Intel i860 this is possibly the right choice ?
Can anyone give us a hint what effort is necessary for such a port ?
Are there any significant drawbacks of AKCL compared to other implementations
(not full CL support, inefficiency of the code,...) ?
Could it be the manual conceals some points ?
Up to now, we prefered CMU-Lisp for our purpose which is fairly huge in
comparison with AKCL. (Typically, 15-20 MByte image files )
Therefore it is hard to imagine that AKCL will give the full functionality.
Any hints are greatly appreciated !
In case I get enough answers I'll post a summary.
-------------------------------------------------------------------------
... always look on the bright side of life ... (Monty Python)
-------------------------------------------------------------------------
Wolfgang Koehler ····@first.gmd.de
GMD-FIRST an der TU Berlin German National Research Centre
Tel. (Berlin 030/049) 6704-2650 for Computer Science
--
-------------------------------------------------------------------------
... always look on the bright side of life ... (Monty Python)
-------------------------------------------------------------------------
Wolfgang Koehler ····@first.gmd.de
GMD-FIRST an der TU Berlin German National Research Centre
Tel. (Berlin 030/049) 6704-2650 for Computer Science
From: Jeff Dalton
Subject: Re: Experiences with AKCL ?
Date:
Message-ID: <8401@skye.ed.ac.uk>
In article <·················@gimli.first.gmd.de> ····@first.gmd.de writes:
>
>We would like to hear some experiences from Lisp users who worked with AKCL
>(good and *especially* bad news).
>
>So far we know AKCL generates C-code from lisp sources which should make it
>very easy to make ports for different processors. Since we consider
>a port to the unusual Intel i860 this is possibly the right choice ?
>
>Can anyone give us a hint what effort is necessary for such a port ?
I've been involved in a couple of ports, including some moderatley
odd Gould machines. I would say it's very portable. Very little
(if any) of the compiler has to change, and most of the other changes
are thinks like integer-decode-float, loading compiled code, and
saving an executable image. If you look at the AKCL sources, you'll
see what things typically differ between machines and operating
systems. I don't know anything about the i860, though, so I can't
say what special problems you'll encounter.
>Are there any significant drawbacks of AKCL compared to other implementations
>(not full CL support, inefficiency of the code,...) ?
>Could it be the manual conceals some points ?
>
>Up to now, we prefered CMU-Lisp for our purpose which is fairly huge in
>comparison with AKCL. (Typically, 15-20 MByte image files )
>
>Therefore it is hard to imagine that AKCL will give the full functionality.
It's a complete implementation of the first edition of CLtL.
Compilation is slow because C compilers take a lot of time.
Some of the built-in functions haven't been implemented as efficiently
as they could be. The sequence functions aren't optimized very much
(if at all) and may be recursive when they could be iterative. This
is part of the price you pay for having a smaller system.
Compiled code is reasonably fast.
Cons cells take up more space than they should: 3 pointers worth
instead of 2 (this is for the GC). But I haven't had any problems
because of this.
The backtrace information for compiled code is fairly minimal.
More recent versions than the one I'm using (which is 1.505)
may be better.
PCL works in AKCL if you want CLOS, but the run-time compilations
are a big pain (due to slow compilation -- see above).
I've been using KCL since 1985 or so and have used it more than any
other Common Lisp, so I guess I'm reasonably happy with it. On the
other hand, if I used lots of things from CLtL II I might feel
differently.
-- jeff
In article <····@skye.ed.ac.uk> ····@aiai.ed.ac.uk (Jeff Dalton) writes:
>In article <·················@gimli.first.gmd.de> ····@first.gmd.de writes:
>>
>>We would like to hear some experiences from Lisp users who worked with AKCL
>>(good and *especially* bad news).
>>
>>So far we know AKCL generates C-code from lisp sources which should make it
>>very easy to make ports for different processors. Since we consider
>>a port to the unusual Intel i860 this is possibly the right choice ?
>>
>>Can anyone give us a hint what effort is necessary for such a port ?
>
>I've been involved in a couple of ports, including some moderatley
>odd Gould machines. I would say it's very portable. Very little
>(if any) of the compiler has to change, and most of the other changes
>are thinks like integer-decode-float, loading compiled code, and
>saving an executable image. If you look at the AKCL sources, you'll
>see what things typically differ between machines and operating
>systems. I don't know anything about the i860, though, so I can't
>say what special problems you'll encounter.
>
On the subject of porting, has anyone got AKCL up and running on a
Solaris 2.X machine? I'd be more than pleased to avoid the effort myself
if somebody else has managed it.
Thanks
Matt