To: davidw@debian.org
Cc: netbsd-advocacy@netbsd.org
Subject: Re: NetBSD/Linux 'distribution'
References: <19990219211709.A25512@debian.org>
From: cgd@netbsd.org (Chris G. Demetriou)
Date: 19 Feb 1999 20:06:12 -0800
In-Reply-To: David Welton's message of Fri, 19 Feb 1999 21:17:09 -0600
Message-ID: <87g1811zyj.fsf@redmail.redback.com>
Lines: 85

> I hope this list is the best forum, but since the question at hand seems
> more hypothetical and 'political', than technical, in nature, it
> seemed appropriate.

Heh.

My (and I repeat "my" -- this is filled with my own personal
assumptions and biases, and it's also worth noting that I know
approximately nothing about Debian or what, exactly, it ships 8-) take
on this:

* This is free software, if you think it'd be a good thing to do, you
can do it.  As noted, you might annoy some people (and, indeed, you
might not be able to call the fruit of your work "NetBSD," since that
means what the NetBSD project wants it to mean), but if you think
there's a market for it i'd rather see wider distribution of NetBSD
than be absolutely dogmatically correct about it.  8-)

* Linux really was/is just a kernel, and so's HURD.  NetBSD is a whole
system.  When you say "Linux x.y.x," you (or at least I 8-) mean a
kernel.  However, when I say "NetBSD x.y.z" i mean the whole system.
I think this generally applies to much of the public that knows about
these things, not just me, but i have nothing to back that
assertion/assumption.  8-) Therefore, by doing what you suggest you
might create serious confusion about what it is that you're actually
selling.  Certainly, The NetBSD Project wouldn't want to be getting
but reports about Debian utilities.

* On a more technical level (who'd have thought?! 8-), "What's the
kernel?"  By that I mean: do you also bring along NetBSD library or
libraries (C library, others?), support code (shared loader, for
instance), kernel-related utilities (ps et al.), etc.?  I don't think
you'll have that easy a time plopping a NetBSD kernel underneath a
non-NetBSD C library, at least not if you want the interfaces provided
to be "NetBSD-ish".  For HURD that's a bit more reasonable,
e.g. because it, like linux is often content to use a glibc as the
library.  You could probably even go that route with NetBSD, but i'm
not convinced that it'd be easy or that the results would be
considered "very useful."


When it comes with down to it, if you're providing the same APIs,
etc., what differentiates the NetBSD kernel from the Linux kernel from
HURD from any number of other systems?  I mean, sure, there are likely
to be things like "performance," and "stability," and "hardware
support" (and i'm not claiming any advantage or disadvantage by
mentioning any of those, just saying that they're the differentiators
8-).  There are also things like "license" (BSD-ish vs. GNU-ish), etc.
However, the public has shown that for the most part they don't give a
good g*d d*mn about those things; note Windows' success.  (actually,
it wins in exactly one category: hardware support, which is probably
the only category that most users truly care about.)


So, all that having been said, my "in a nutshell" opinion:

* you providing a distribution and perhaps support for a NetBSD
  distribution _plus_ many of your tools would be an unqualified-good
  thing.

* you providing a distribution and perhaps support for a NetBSD system
  with many of the NetBSD tools replaced by your tools might be a good
  thing from the perspective of getting more BSD code out into the
  world (and in terms of making money for you folks 8-), but has the
  potentially serious side effect of muddying the water around "What
  is NetBSD," and things like that.

* i don't really see you providing a NetBSD kernel plus all of your
  user-land would be beneficial for anybody.  I think that down
  that road lies a whole lotta pain for everybody involved.  8-)

* certainly NetBSD, even just the kernel, could use more hands working
  on it, writing drivers, improving stability and performance, etc.
  If your efforts have that effect, then in at least that way they're
  positive for the project.

That having been said, don't let me dissuade you from trying to do any
of the above if you're truly interested.  8-)



cgd
-- 
Chris Demetriou - cgd@netbsd.org - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.