From owner-cvs-all  Tue Mar 13 13:32:40 2001
Delivered-To: cvs-all@freebsd.org
Received: from winston.osd.bsdi.com (winston.osd.bsdi.com [204.216.27.229])
	by hub.freebsd.org (Postfix) with ESMTP
	id EAFF437B71A; Tue, 13 Mar 2001 13:32:22 -0800 (PST)
	(envelope-from jkh@osd.bsdi.com)
Received: from localhost (jkh@localhost [127.0.0.1])
	by winston.osd.bsdi.com (8.11.2/8.11.1) with ESMTP id f2DLV8H48106;
	Tue, 13 Mar 2001 13:31:08 -0800 (PST)
	(envelope-from jkh@osd.bsdi.com)
To: asmodai@wxs.nl
Cc: phk@critter.freebsd.dk, grog@lemis.com, rwatson@FreeBSD.org,
	imp@harmony.village.org, cvs-committers@FreeBSD.org,
	cvs-all@FreeBSD.org
Subject: Re: Core's function (was: The Project and onward [was: Re: cvs
 commit: src/sys/netinet ip_output.c])
In-Reply-To: <20010313104930.C60817@daemon.ninth-circle.org>
References: <20010313121002.F59348@wantadilla.lemis.com>
	<3794.984471257@critter>
	<20010313104930.C60817@daemon.ninth-circle.org>
X-Mailer: Mew version 1.94.1 on Emacs 20.7 / Mule 4.0 (HANANOEN)
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-Id: <20010313133108S.jkh@osd.bsdi.com>
Date: Tue, 13 Mar 2001 13:31:08 -0800
From: Jordan Hubbard <jkh@osd.bsdi.com>
X-Dispatcher: imput version 20000228(IM140)
Lines: 36
Sender: owner-cvs-all@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

> And unfortunately, most, if not all, of the discussions I have seen
> raised are short-term based.  We fail to work towards goals on the
> longer term.  Dare I name clustering yet again, yes I do.

If you dare to raise clustering then you should also dare to consider
taking on the design work involved and releasing some white papers on
potential approaches to solving the problem in FreeBSD.  To do
otherwise is to simply throw out a buzzword ("clustering!") and say
that somebody, somewhere, should work on it because everyone says it's
important.  That's not architectural discussion, that's a bullet point
on a marketing slide.

The biggest reason I can see for things like clustering not being
addressed, and here I'm going to reveal a basic secret of the universe
so watch carefully, is that it's HARD. :-) I'll bet you cringed when I
told you to write a white paper because if you know anything about CS
at all, you know that I didn't just ask you to do something trivial
that you'll be able to pump out in an hour or two.

That's why so much of this kind of work goes on inside of
universities.  They have the kind of time and personnel resources to
write white papers which give a programmer the kind of outline they
can work from in writing some actual implementation code, and that's
usually hardly trivial either.  Some of the most complex work to enter
FreeBSD in the last couple of years didn't come out of discussions in
-arch, in fact, they came out of white papers like Ganger-Patt's
"CSE-TR-254-95" which Kirk followed in writing the softupdates code.
Even just the implementation work for that was so hard that he didn't
turn around and give that code away, either, he made money on it for
over a year before he was happy enough with the return on all the time
he'd invested to give it away.

This is why I think a lot of your and phk's recent hand-waving on this
topic may be "right" in principle but just not very realistic either.

- Jordan

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message