Newsgroups: comp.lang.scheme
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!uhog.mit.edu!sgiblab!spool.mu.edu!torn!newshub.ccs.yorku.ca!nexus.yorku.ca!oz
From: oz@nexus.yorku.ca (ozan s. yigit)
Subject: Re: The expense of call#cc (was R4RS)
Message-ID: <oz.782533000@nexus.yorku.ca>
Sender: news@newshub.ccs.yorku.ca (USENET News System)
Organization: York University
References: <19941003T142911Z.enag@gyda.ifi.uio.no> <19941004T194114Z.enag@hnoss.ifi.uio.no> <mhamburg-051094101457@macb022.mv.us.adobe.com> <36ut56$835@larry.rice.edu> <380o10$p4f@info.epfl.ch>
Date: Wed, 19 Oct 1994 02:16:40 GMT
Lines: 21

i have not yet had a chance to read the study in question,
but i take it that any reference to heap allocation is in fact a
reference to "naive" heap allocation, or what Clinger et al. [1] refers
to as the "garbage collection" strategy.. there is another method also
referred to as the "heap" strategy, which marks the captured frames, and
uncaptured frames are recycled without the intervention of the garbage
collection machinery. I have never seen this implemented anywhere
except in psi, and my performance tests showed that it is superior
to "garbage collection" strategy, and to naive "stack" strategy
where the entire stack is copied and restored.

heap strategy is probably worth considering for systems
that use fixed-size frames. more information is in Clinger et al. [1].

oz
---
[1] William D. Clinger, Anne H. Hartheimer and Eric M. Ost
    Implementation strategies for continuations
    Conference Record of the 1988 ACM Conference on
    Lisp and Functional Programming, 124-131, Aug. 1988.

