Path: utzoo!utgpu!watmath!clyde!bellcore!texbell!bigtex!milano!cs.utexas.edu!tut.cis.ohio-state.edu!bloom-beacon!bu-cs!purdue!decwrl!labrea!agate!saturn!libes@cme.nbs.gov
From: libes@cme.nbs.gov (Don Libes)
Newsgroups: comp.os.research
Subject: Re: THREADS, Light weight processes
Message-ID: <5394@saturn.ucsc.edu>
Date: 6 Nov 88 18:09:59 GMT
Sender: usenet@saturn.ucsc.edu
Organization: National Bureau of Standards, Gaithersburg, MD
Lines: 23
Approved: comp-os-research@jupiter.ucsc.edu


In article <5354@saturn.ucsc.edu> mcvax!cwi.nl!jack@uunet.UU.NET (Jack Jansen) writes:
>So far nobody seems to have mentioned that LWPs are a very nice
>mechanism to program with, much nicer than non-blocking I/O with
>completion interrupts or polling, etc.

I agree completely.  I recently wrote an article discussing exactly
this.  It showed the difficulty of writing a conceptually simple
program, and how it resolved cleanly into 3 threads, just a couple of
lines a piece.  It also discussed threads with respect to OS/2 and UNIX.

	"Unraveling Threads", Micro/Systems Journal, p. 12, Nov. 1988.

Another reference you might consider (also by me) is:

	"Multiple Programs in One UNIX Process", ;login, July/August 1987.

This paper described an easy way to do LWPs in UNIX.  The idea was
that (the conceptually elegant) XINU was a LWP scheduler and that
porting it to UNIX was easy.  It required no modifications to the
kernel.  Only 6 lines of assembler to save and restore registers.

Don Libes          libes@cme.nbs.gov      ...!uunet!cme-durer!libes