Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!gorodish!guy
From: guy%gorodish@Sun.COM (Guy Harris)
Newsgroups: comp.unix.questions
Subject: Re: tty watcher
Message-ID: <28747@sun.uucp>
Date: Tue, 22-Sep-87 14:05:24 EDT
Article-I.D.: sun.28747
Posted: Tue Sep 22 14:05:24 1987
Date-Received: Thu, 24-Sep-87 06:34:15 EDT
References: <4263@ozdaltx.UUCP> <15136@hi.UUCP> <1178@mtune.ATT.COM> <1310@mtune.ATT.COM>
Sender: news@sun.uucp
Lines: 41

> >this means that *everything* going up and down the stream gets sent upstream
> 
> Well, the original requirement was that *everything* the user typed or
> was sent be seen on the monitoring terminal!

When I said *everything*, I meant *everything*, including e.g. "ioctl"
messages.  This was not part of the original requirement, and naively dumping
the bytes of such a message on the monitoring terminal is not likely to be
useful.  If the person sitting at the monitoring terminal isn't familiar with
UNIX systems calls, even dumping the "ioctl" information in a nice
programmer-readable form isn't likely to be useful....

> >I could imagine some packaged systems supplying a prebuilt kernel or
> >kernels.
> 
> I can imagine this as well although I understood that the SVR3 licence
> forbad this.

If this is true, this would be an incredibly stupid move, as it would interfere
with people using UNIX as e.g. a hidden component in an embedded system.
> 
> >Umm, no, it doesn't.  I said "consist(s) *solely* of a user-mode program
> >that drops on top of a *non-hacked* UNIX system."
> 
> You're right. I withdraw the comment. Although I personally believe that
> any streams modules *ought* to be user-level code, and that an
> experimental streams module shouldn't wreck the rest of the kernel when it
> falls over. Of course this would mean that you'd need more than two
> mutually protected address spaces. (Does this violate POSIX?)
> 
> How about agreeing on 'come the revolution, this *will* be easy'?

Yes, I agree on that one; were it possible to put large chunks of what is now
kernel code into user mode (not just streams modules, but device drivers, file
systems, etc., etc.), it would be a lot easier to add this sort of extension.
Some people from Apollo gave a paper at the last Usenix on their user-mode
streams implementation, which admittedly sits on top of something other than
(and more modern than) a UNIX kernel.
	Guy Harris
	{ihnp4, decvax, seismo, decwrl, ...}!sun!guy
	guy@sun.com