[plt-scheme] xxx chooses MzScheme as preferred language

From: Paulo J. Matos (pocmatos at gmail.com)
Date: Wed May 31 03:33:50 EDT 2006

On 31/05/06, Brent Fulgham <bfulg at pacbell.net> wrote:
> Hash: SHA1
> On May 30, 2006, at 6:53 AM, Matthias Felleisen wrote:
> >
> > On May 29, 2006, at 6:40 PM, Paulo J. Matos wrote:
> >
> >> - unless we could compile all the stuff and make it quick, the stuff
> >> would be slow, still that would not make users away (see Gentoo and
> >> python);
> >
> > Are you sure that the scripts in Python are faster than equivalent
> > scripts in MzScheme?
> At the risk of a bit of self-promotion, I can point you at a few
> examples that seem to indicate
> that Python is a bit faster than MzScheme.  In fact, it's a bit worse
> than I show, since these
> data are for Mzc-compiled routines, not the interpreter.  Run using
> mzc version 301:
> http://shootout.alioth.debian.org/sandbox/benchmark.php?
> test=all&lang=mzc&lang2=python

Wow, I must say the interface is very nice. It is sad, however, that
the results for MZC Scheme are not nice. Which is one of my worse
fears... Scheme is nice but when it comes to speed, you just can't
compare it with C/C++.

And sometimes you do need speed. For example, in my line of work, SAT,
we have been developing SAT solvers for research and then we just send
them into competition. Most popularity come from your papers and your
solvers results in competition. You win/lose by a few seconds.
Initially I proposed the development of an extension of SAT (SMT)
solvers to the group in Scheme which was neglected, and worse people
made fun of me for a week, since C++ is the way to go, they say. Now,
I fear that it was for the better. Most probably I would have had
terrible results, still I may have had a faster development but that's
not really what's important. I also tried at the time to use CMUCL
since it is fast and somewhat well-known but I talked with António
Leitão, which works with Lisp for 'centuries' and he didn't seem that
confident that it could be competitive.

However, I must say that other than this kind of development with
severe efficiency needs, PLT Scheme is the way to go.


Paulo Matos

> - -Brent
> Version: GnuPG v1.4.2.2 (Darwin)
> iD8DBQFEfS8zzGDdrzfvUpURAkTOAJ4mUySodP56Ai/xSWukCE5Prm2biwCdHd5E
> 8JuwGjPai/2tKTr9amh8rO8=
> =Bhlf

Paulo Jorge Matos - pocm at sat inesc-id pt
Web: http://sat.inesc-id.pt/~pocm
Computer and Software Engineering

Posted on the users mailing list.