Newsgroups: comp.lang.dylan
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!hudson.lm.com!news.pop.psu.edu!news.cac.psu.edu!howland.reston.ans.net!news.sprintlink.net!hookup!news.mathworks.com!news.kei.com!world!edwards
From: edwards@world.std.com (Jonathan Edwards)
Subject: Re: Multithreading & multiprocessing
Message-ID: <D20quE.Bx3@world.std.com>
Organization: IntraNet, Inc.
References: <D1ypys.8G5@world.std.com> <3ejvjt$1as@cantaloupe.srv.cs.cmu.edu> <D2013s.I71@world.std.com> <3eksi5$cai@jive.cs.utexas.edu>
Date: Sat, 7 Jan 1995 04:51:50 GMT
Lines: 17

In article <3eksi5$cai@jive.cs.utexas.edu>,
Paul Wilson <wilson@cs.utexas.edu> wrote:
>Do you just mean a "real time" GC that never stops an application for
>a significant period of time, or one that never blocks the application
>thread because of a page fault by the collector?

I want guaranteed small latencies for high-priority threads. You need that
to build database and networking software. In C++ you would use a separate
heap that is pinned in memory. But I imagine Dylan like Lisp will tend to
create many short-lived objects that need to be GC'd sooner or later.
If the GC is interuptable with a millisecond latency that is fine, but
preferably it can run entirely in parallel as a lower-priority thread.
One could impose special constraints on "real-time" objects to acheive this.
-- 
Jonathan Edwards				edwards@intranet.com
IntraNet, Inc					617-527-7020
One Gateway Center				FAX: 617-527-6779
