Newsgroups: comp.lang.scheme
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!yeshua.marcam.com!charnel.ecst.csuchico.edu!olivea!uunet!sytex!smcl
From: smcl@sytex.com (Scott McLoughlin)
Subject: Re: GNU Extension Language Plans
Message-ID: <iB0Ruc1w165w@sytex.com>
Sender: bbs@sytex.com
Organization: Sytex Access Ltd.
References: <38k565$7pb@boogie.cs.utexas.edu>
Date: Wed, 26 Oct 1994 03:10:05 GMT
Lines: 44

wilson@cs.utexas.edu (Paul Wilson) writes:

> 
> If you're interested in a survey that explains these things have a look
> at the GC paper repository on cs.utexas.edu in the directory pub/garbage.
> The README file explains what's there, including a short version and a
> long version of my survey on garbage collection, and a paper on our
> incremental treadmill(ish) collector.  (We have a new generational version,
> but that's not in the paper.)  There's also an old paper on my old
> opportunistic generational GC. 
> 
> There are also some notes on Scheme and Scheme compilers and interpreters
> (roughly half a book's worth of class notes) in pub/garbage/schnotes.
> Constructive comments welcome.

Howdy,
        For what it's worth, I found the survey (long version) to be
a wonderful resource. I found that a second reading, after reading
H. Baker's articles on "realtime" copying/non-copying GC (available
on ftp.netcom.com::pub/hbaker), was even better.
        Here's a constructive comment:  While it's hard to include
sniggly details in a paper, sometimes a lack of detail makes a 
paper more confusing. Case in point: only now that I've read Baker's
1978 paper describing a realtime copying collector can I start to 
"imagine" how one might _code up_ one of these fancier incremental
collectors. In that paper he addresses issues of arrays and the
program stack that are frequently not discussed.  I also like
the provided "pseudo-code".
        Anyway, for this reader, dumping a bit of advice gained
from experience implementing these things into the theoretical
discussion contributes to an understanding of the theoretical
concepts.
        Another issue that I'd like to see addressed is ease
of coding/maintenance/code generation for various schemes. For
many projects these issues will be just as important as speed
(which is, of course, the _fun_ part) since these things do
have to be shoehorned into language compilers/interpreters
where many design decisions have to be juggled simultaneously.
        Keep up the good work.

=============================================
Scott McLoughlin
Conscious Computing
=============================================
