On Fri, Jun 3, 2022 at 3:32 PM Larry McVoy <lm@mcvoy.com> wrote:

On Fri, Jun 03, 2022 at 11:20:58PM +0200, Tom Ivar Helbekkmo via TUHS wrote:
> Clem Cole <clemc@ccc.com> writes:
>
> > Some of us on this list remember the original BDSi fight, the 386BSD
> > to FreeBSD, then NetBSD and OpenBSD (I was friends with both sides of
> > many of these wars).
>
> Irrelevant to the topic, I know, but I'd just like to point out, since
> you call these things "wars", that NetBSD grew out of 386bsd in a quiet,
> friendly fashion, and then FreeBSD out of NetBSD just as quietly. (BSDi
> growing out of 386bsd was a completely separate affair that I know very
> little about, and the OpenBSD fork from NetBSD was mostly just a
> personal animosity thing, Theo de Raadt having made enemies in both the
> NetBSD and FreeBSD camps -- but it has left no bad blood behind it.)
>
> In other words, no wars that I know of.

Umm, were you there? I was a BSD guy before I turned to Linux and I
turned to Linux because of those wars. There is no good reason to have
{386,Free,Net,Open,DragonFly}bsd other than, as Linus stated, "Nobody
could decide who would drive the big red fire truck so now they each
have their own toy fire truck that they drive around".

BSD would have won if there was a Linus for BSD. There was not so you
got all this replicated effort, the BSD community effectively divided
and conquered themselves.

It was, and is, a train wreck. It's the poster child for how not to
manage a project.

I did BitKeeper for Linus because he refused to use any crappy source
management solution and people like Dave Miller were threatening to
fork just so they had some solution. I did that because a forked Linux
would turn into the same mess of {386,Free,Net,Open,DragonFly}bsd which
is obviously not remotely close to ideal. Far from it.

386BSD died because its founder couldn't deal with collaboration. He tried to

be dictator and that failed because he didn't accept other people's collaboration

out of worries he couldn't sell 386BSD. NetBSD and FreeBSD took up the charge

for a free and open system. I'll agree it was unfortunate that there was a split

since NetBSD focused on portability and FreeBSD focused on fastest possible

i386/i486 code. I'd suggest, though, that the USL lawsuit cast a huge pall on

things and introduced enough uncertainty to further derail things. Had it not

been for that additional blow, things would have turned out differently.

OpenBSD and Dragonfly BSD didn't split until years later and also

represented differences of opinion on where to take the focus of the

system (OpenBSD thought the NetBSD folks didn't take security seriously

enough and the DFBSD folks thought the efforts to make a parallel kernel

in FreeBSD were off track and should be done completely differently).

I lived through all of that, I was an active kernel developer at Sun,
SGI and elsewhere. I would have loved to have seen the SunOS VM system
ported to 4.x BSD and that been the default answer for a kernel. Instead
we got Linux, which has it's positive points for sure, but it also has
decided to let every feature imaginable into the kernel.

We wound up with MACH in BSD because when Sun tried to donate their

VM code to Berkeley, the corporate lawyers said no. It was giving away too

much shareholder value, and would result in a huge write-off which would,

one would presume, negatively affect the stock price. Had this donation

actually transpired, 386BSD would have had a bigger advantage from the

get go... Oh well

Warner