Subject: Re: Scheme implementations - comparisions?
From: rpw3@rigden.engr.sgi.com (Rob Warnock)
Date: 1997/12/17
Newsgroups: comp.lang.scheme
Message-ID: <677rcf$5jq4n@fido.asd.sgi.com>

Ketil Z Malde  <ketil@ii.uib.no> wrote:
+---------------
| Is there a resource that compares the various Scheme implementations?
| There seems to be an abundance of implementations, probably due to the
| simplicity of the language.  Any reason to pick one over the others?
+---------------

Part of the problem is that because there *are* so many, doing any kind
of reasonably complete comparison is very time-consuming.

It's also very application-specific. What matters to your application
might not matter at all to mine. By what metric does one compare two
implementations?  Start-up time?  Run-time for massive floating-point
calculations? Time to do a byte-at-a-time scan of an input file read
from disk? Memory needed to get it to run at all? Memory used after
the 10,000th garbage collection? Time to append two lists? Size of
the source code in lines of the Scheme system itself? [Hey, useful
for students!]

Finally, precisely *because* Scheme is so small, portability tends to
be rather brittle with respect to added or missing or oddly-implemented
features (e.g., maros, structures, object system, libraraies). That is,
any application of size will tend to exploit to the limit whatever useful
extensions are available in the implementation it's originally developed
in, and thus have trouble being ported to another implementation. This
even if the second Scheme implementation is a lot "better" than the first
(by any reasonable measure -- but see above).

So... Good luck.


-Rob

-----
Rob Warnock, 7L-551		rpw3@sgi.com   http://reality.sgi.com/rpw3/
Silicon Graphics, Inc.		Phone: 650-933-1673 [New area code!]
2011 N. Shoreline Blvd.		FAX: 650-933-4392
Mountain View, CA  94043	PP-ASEL-IA