4 Jun
2022
4 Jun
'22
5:48 a.m.
On Fri, Jun 03, 2022 at 04:52:52PM -0600, Warner Losh wrote:
Um, so there were 3: 386, Net and Free. That's already 2 too many.
So I had a successful commercial product that ran on all of those variants
without issue. I supported linux on everything from ARM to IBM's z-system
mainframes and all the arches inbetween. I think I have one #ifdef SPARC
in there because there was a cache flush bug but that was a hardware issue,
not a software issue.
I also supported {Free,Net,Open}BSD and I had way more problems with them
than I did with Linux.
The problem is
there was {386,Net,Free,Open,DragonFly}BSD where there
should have just been "BSD". One, not a bunch.
Except from 1993-1996 there were only two of those BSDs. NetBSD and FreeBSD
forked in 1993 due to the inability of the patchkit to adequately cover the
problems
in 386BSD governance. Where do you
think Linux would be if there was {A,B,C,D,E,F,G}Linux?
There is one kernel. One and only one. With everyone working on that
one kernel.
Except there never really was only one kernel. There have been hundreds
of forks of the Linux kernel over the years. Most of them have been
commercial
of some flavor (Redhat, Debian, OpenSUSE, MontaVista, WindRiver, Android
etc)
had hundreds or thousands of patches on the base Linux kernel for a long
time
and trying to move from one to another if you also had patches was a
nightmare. Kernel.org has kept going, and many of the chanages
from these systems were
lost.
Some were not as good as what came in upstream, while others were encumbered
by commercial contracts that made them unappealing to upstream. True, many
of
them did wind up in kernel.org, but to say there aren't forks in Linux is
stretching
reality a bit...
There is one kernel development stream that matters. RedHat knows that
if they don't get their stuff into Linus' tree, they have a nightmare
on their hands. That's why RedHat paid so many of the kernel developers.
Sure, there are forks, but there is one tree that matters, and that is
Linus' tree. You can't say that about BSD and that is the problem in
it's entirety. If I want to change BSD, which one?
Even today, with the benefit of hindsight, it's
hard to pin which of these
facts on the ground was the biggest driver for most people...
For me, I gave up when there was no longer one BSD, there was one Linux.
--lm