Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!samsung!think.com!barmar
From: barmar@think.com (Barry Margolin)
Newsgroups: comp.lang.lisp
Subject: Re: Memory Management in Lisp?
Message-ID: <1991Feb15.223520.17267@Think.COM>
Date: 15 Feb 91 22:35:20 GMT
References: <1991Feb15.191259.20090@aero.org>
Sender: news@Think.COM
Organization: Thinking Machines Corporation, Cambridge MA, USA
Lines: 33

In article <1991Feb15.191259.20090@aero.org> srt@aero.org (Scott "TCB" Turner) writes:
>Will weak reference be added to Common Lisp?  

Not in the current round.  Only a few Common Lisp implementations currently
have such mechanisms, and there isn't yet a concensus on the right
interface to them.  Maybe a future, followon standard (Common Lisp 2000?)
will have them, because many of us acknowledge that they are useful.

>					       Or even better, will some
>kind of user memory management be added?

There have been no proposals made to X3J13 for any such mechanism, and it's
too late now for new features.  I'm not even sure how good a portable
memory management mechanism could be in a Lisp environment.  Could you
describe the kind of interface you're hoping for?  Something reasonably
simple like resources (pools of similar objects, useful when you're
repeatedly allocating and deallocating objects, as you reuse objects rather
than letting them become garbage), or something more complicated like areas
(sections of memory with different garbage collection parameters).

Note that Common Lisp currently specifies virtually nothing about memory
management.  CLtL doesn't use the term "garbage collection" at all, and
CLtL2 only mentions it in passing (in the description of the DYNAMIC-EXTENT
declaration).  An implementation without a garbage collector would conform
to the spec; in fact, for several years most MIT Lisp Machine users
disabled the garbage collector because the original implementation was very
buggy -- they simply rebooted when they ran out of memory.

--
Barry Margolin, Thinking Machines Corp.

barmar@think.com
{uunet,harvard}!think!barmar