Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: Notesfiles $Revision: 1.7.0.10 $; site uokvax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxn!ihnp4!inuxc!pur-ee!uiucdcs!okstate.UUCP!uokvax.UUCP!emjej
From: emjej@uokvax.UUCP
Newsgroups: net.micro.6809
Subject: Re: os9fork() and chain()
Message-ID: <3500133@uokvax.UUCP>
Date: Mon, 10-Feb-86 08:03:00 EST
Article-I.D.: uokvax.3500133
Posted: Mon Feb 10 08:03:00 1986
Date-Received: Wed, 12-Feb-86 08:25:11 EST
References: <304@chinet.UUCP>
Lines: 25
Nf-ID: #R:chinet.UUCP:304:uokvax.UUCP:3500133:000:1065
Nf-From: uokvax.UUCP!emjej    Feb 10 07:03:00 1986



You can't clone the Unix fork() with os9fork(); even if you fork a process
with the same primary module, its state isn't duplicated from that of the
parent.  Frankly, this is a good thing; the Unix fork() sucks--it requires
considerable hardware assist to avoid massive overhead (ask someone the
story behind Berserkeley's vfork() call), and is practically never used
save in a form that emulates os9fork(), i.e.

		if (fork(.) says I'm the child)
			execl(what I wanted to run in the first place);

instead of the code in the parent

		os9fork(what the child ought to run);

As for what parameters it and chain() ought to have, well, I'm afraid you
have to either know, or you can link to the module and take a peek. It 
would have been nice if the table mapping module types to how they should
be run when forked or chained to were in the kernel and tweakable, rather
than in the shell.  (While OS-9 is internally much cleaner than Unix, it's
not perfect; I'd like to think that recognition of this distinguishes us
from the Unix-worshippers.)

					James Jones