Dear all,
Getting a Lisp + Emacs + Slime to run on a Mac Pro seems to be more
troublesome than we hoped for. We spent all day getting it to work and
although it runs now, it doesn't really meet all of our expectations.
This résumé might help some other people who are trying to achieve
the same goal and maybe someone could give us some insights on how we
might get it to work better.
The way we got it working now is by using Carbon Emacs and the Slime it
is shipped with (November 2006 edition,
http://homepage.mac.com/zenitani/emacs-e.html) and OpenMCL
(openmcl-darwinx8664-snapshot-061110, ftp://clozure.com/pub/testing/).
This set-up works fine, but only if you choose to use only one
processor core instead of the four the Mac Pro is shipped with. If we
choose more than one processor the lisp process stops responding at a
seemingly random point in the program without giving any feedback.
After performing some tests we found out that it has something to do
with the interaction between OpenMCL and Emacs as running OpenMCL from
the command-line works like a charm. We stumbled upon a post
(http://clozure.com/pipermail/openmcl-devel/2006-August/003526.html)
which stated that there might be some problems with the treads. Alas,
changing our swank-openmcl.lisp as stated in that old post didn't solve
any of our problems.
Did anyone of you have more success bringing a Lisp to the Mac Pro?
Another sidequestion: at this moment we are only considering OpenMCL as
this implementation is the only one which has some support for the
processors inside a Mac Pro next to Lispworks (which at this moment
seems too expensive). Does anyone know another implementation we might
consider?
Many thanks in advance,
Pieter & Joris
··········@gmail.com wrote:
> Dear all,
>
> Getting a Lisp + Emacs + Slime to run on a Mac Pro seems to be more
> troublesome than we hoped for. We spent all day getting it to work and
> although it runs now, it doesn't really meet all of our expectations.
> This r�sum� might help some other people who are trying to achieve
> the same goal and maybe someone could give us some insights on how we
> might get it to work better.
>
> The way we got it working now is by using Carbon Emacs and the Slime it
> is shipped with (November 2006 edition,
> http://homepage.mac.com/zenitani/emacs-e.html) and OpenMCL
> (openmcl-darwinx8664-snapshot-061110, ftp://clozure.com/pub/testing/).
> This set-up works fine, but only if you choose to use only one
> processor core instead of the four the Mac Pro is shipped with. If we
> choose more than one processor the lisp process stops responding at a
> seemingly random point in the program without giving any feedback.
>
> After performing some tests we found out that it has something to do
> with the interaction between OpenMCL and Emacs as running OpenMCL from
> the command-line works like a charm. We stumbled upon a post
> (http://clozure.com/pipermail/openmcl-devel/2006-August/003526.html)
> which stated that there might be some problems with the treads. Alas,
> changing our swank-openmcl.lisp as stated in that old post didn't solve
> any of our problems.
>
> Did anyone of you have more success bringing a Lisp to the Mac Pro?
>
> Another sidequestion: at this moment we are only considering OpenMCL as
> this implementation is the only one which has some support for the
> processors inside a Mac Pro next to Lispworks (which at this moment
> seems too expensive). Does anyone know another implementation we might
> consider?
Yes, I do (Where is the AFLAC duck when I need him?)
LISPWORKS!!!!!!!!!!!!!!!!!!!!!!!!!!!
<sigh>
kt
ps. No, I did not pay all these people to show up on cue talking about
the tens of hours spent on avoiding cost just to win the SBCL thread.
(But, thanks, Ian.)
k
--
Algebra: http://www.tilton-technology.com/LispNycAlgebra1.htm
"Well, I've wrestled with reality for thirty-five
years, Doctor, and I'm happy to state I finally
won out over it." -- Elwood P. Dowd
"I'll say I'm losing my grip, and it feels terrific."
-- Smiling husband to scowling wife, New Yorker cartoon
Ken Tilton wrote:
> ps. No, I did not pay all these people to show up on cue talking about
> the tens of hours spent on avoiding cost just to win the SBCL thread.
> (But, thanks, Ian.)
Win the SBCL thread? When did threads become contests and what's the
prize? I was reading these threads for useful information, which
generally means skipping your posts. You waste your time moaning about
other people wasting effort on OSS lisp implementations. If you haven't
figured it out, OSS lisp implementations are like gateway drugs.
OSS implementations grow the user base in ways that pay-to-play
implementations don't. People can cut their teeth for free and then
move on the Lispworks or ACL when they've developed their code beyond
what the OSS implementations support and are ready for the big time.
The bigger the pool of people using lisp, the better for everyone,
regardless of how they are using it.
So, Grouchasaurus Rex, please restrain yourself to posting pearls of
lisp wisdom. You need to maximize your knowledge transfer because that
great big fireball you see in the sky is an OSS asteroid, so you don't
have much time left.
Long time lurker,
Tom
Ok, it turned out that we were quite close to the solution of our
problem. The mailing list included some new posts on this issue. You
can find the post that did the trick for us on:
http://clozure.com/pipermail/openmcl-devel/2006-August/003528.html. I
didn't want to mention the war ... ;)
Joris