First off, this issue may be of concern to users of many ports besides
i386 -- Linux is being developed for ARM and Alpha and others, FreeBSD
is branching out as well, and OpenBSD appears to be supporting as many
ports as they have resources for. There's also Mac and SPARC port users.

I'm not sure at what point it would be appropriate to drag in other ports.
Maybe this just belongs on netbsd-advocacy?

> I've just recently returned from Comdex in Las Vegas.  While I was there,
> I conducted a number of interviews, with a number of organizations and
> individuals for Internet Paper.  Based on the responses I have received,
> as well as information from other sources including the web, mailing
> lists, news sources, and other publications, it would appear that Linux
> (which is already the most popular ix86 Unix OS) is gaining in some of
> its growth at the expensive of BSD based Unixes, including NetBSD.  More
> alarming, this trend appears to be predominate among new Unix adoptees.

Yes. This is because compared to Linux, the *BSD development process looks
so much like a non-profit Cathedral it isn't funny. Thus, every kid who
wants to dabble in the free O/S racket starts fooling around with Linux,
either because it's the first free O/S he heard about (this has incumbent
effects of its own), or because he tried to get into the *BSD's and found
that if he tried too hard to help out, many people would treat him like an
irritating pest.

> As someone who has personally chosen to support NetBSD for many years,
> and used NetBSD as much as possible in the home, work, and for software
> development, I find this trend alarming.  It is true that NetBSD does
> benefit from the talent pools from other Unix operating systems,
> including Linux, FreeBSD, and others, especially through its binary
> emulation of applications and porting of non NetBSD based source code.
> However, it has become clear that this approach is not enough to prevent
> NetBSD's gradual erosion.  I'm sure may of us NetBSD supporters envy
> the increasing user base and increasing "native" software base that
> Linux is receiving.

Yes. I think there is going to be little we can do to stop this, and for
quite some time. Most of the growth for Linux has its roots in the concept
of "taking a stand against Microsoft". Everyone is behind the penguin because
they realize that to support Free O/S' in general leads to fragmentation and
they are very afraid of the inevitable FUD and divide-and-conquer tactics
from Microsoft. You can see this in Eric Raymond's article "Open Source's
First Six Months" which contains the following paragraph:

= There's a problem with this, however, that's almost serious enough to make
= me wish Samba didn't exist. While stealthing open-source boxes will solve
= a lot of individual problems, it won't give us what we need to counteract
= the attack marketing and FUD-mongering that we are going to start seeing
= big-time (count on it) as soon as Microsoft wakes up to the magnitude of the
= threat we actually pose. It won't be enough to have a presence; we'll need
= a visible presence, visibly succeeding.

I disagreed so strongly with the paranoia in this paragraph that I worked up
the guts to send him an email. My position is that we want as many "shadow
ships" as possible to be buried in the corporate world, doing real work and
outperforming NT crates, waiting for the day the FUD finally hits the fan.

> Part of the problem with NetBSD is that it is one of several "forks" or
> splits from BSD, which also include FreeBSD, OpenBSD, BSDI, etc.  This
> splitting up of BSD into the different forks has divided up the talent
> pool of BSD developers, benefiting non-forked operating system like
> Linux.

Sort of. The non-forkedness mostly applies to the kernel. The rest of a
typical Linux distribution is obtained from the FSF or public FTP sites,
or where-ever.

If anything, the great triumph of Linux is that Linus is a Zen style leader.
He spends most of his time making sure everyone else gets along, choosing
his lieutenants carefully, and above all TRUSTING them to DTRT, or at least
something close enough to the right thing that it can be fixed. After all,
that's where their favorite quote "debugging is parallelizable" came from.

> NetBSD, FreeBSD, and OpenBSD are all open source, and each of them have
> their own advantages over the other.  Now is the time that people put
> their egos aside and perhaps at least talk about merging some components
> of these BSD operating systems, including kernels, drivers, etc., taking
> the best features from each.  Only then can we establish a BSD based OS
> as the real non-Linux Unix alternative; something that Sun Solaris, SCO,
> OSF, etc. are also trying to do.

While I would certainly cheer this on as much as many other users, trying
to explain *BSD to my co-workers has forced me to realize _why_ they forked
off in the first place:

    FreeBSD wants to be the "people's BSD", doing what pleases users most.
    NetBSD wants to be the "guru's BSD", with the most bogon-free system.
    OpenBSD wants to be the "non-USA-encumbered BSD", for the security types.

These are genuinely conflicting goals, and I don't think reconciling them
is a simple matter of people checking their egos at the door for a change.

> I know that merging the BSD OS forks back together is a huge undertaking,
> technically and politically.  But if we all plan to continue using NetBSD
> (or its successor) as a viable and robust operating system in the many
> years to come, I can not see any other alternative.

Well, that's the question, isn't it. Is Linux going to smother us, or is
it going to educate and then disillusion a generation of defectors for us
to recruit?

> As a software developer and system administrator, I love NetBSD, I would
> hate to have to force myself to switch to another non-BSD based operating
> system in order to keep from sacrificing performance, robustness, or
> application/driver/support diversity.

Admittedly I feel some of this myself. It does not help that so far, what I
have learned from these email lists and my own experiences tracking -current
is that the NetBSD project appears to be running very close to survival mode
in terms of growth -- core seems to be burning out about as fast as it is
taking on new people. What pisses me off is that this leaves the PRIMARY
NetBSD advantage, superior code leverage as a result of meticulously clean
design, tragically underexploited.

> I would be most interested in hearing from other NetBSD users about the
> idea of possible merging the BSD OS forks back together, especially from
> those of you who are actively involved in NetBSD OS development.

So would I. Especially if anyone has comments/flames/etc on my analysis.

Todd Whitesel
toddpw @ best.com