[TUHS] Why BSD didn't catch on more, and Linux did

[tytso@mit.edu (Theodore Ts'o)]

On Tue, Feb 06, 2018 at 02:13:51PM -0800, Dan Stromberg wrote:
> 2) I think the main reason BSD nearly died, was the AT&T lawsuit.  At
> the time, Linux appeared to be a safer bet legally.

At the time of the AT&T lawsuit, most of the people who would be
interested in using a Un*x-like system on their personal x86 systems
probably wouldn't have been worried about their own personal legal
liability.  The decision of corporations to use Linux was well *after*
the AT&T lawsuit was resolved.

The lawsuit have suppressed the willingness of the 386BSD's to
advertise what they had --- I had no idea 386BSD was far as advanced
as it was, because I didn't happen to have been at that magic BOF
where word was apparently passed around on the down low.  But that's a
little bit different, and more subtle, than just saying "safer bet
legally".

> 3) Linux got a reputation as an OS you had to be an expert to install,
> so lots of people started it to install it to "prove themselves".
> This was sort of true back when Linux came as 2 floppy images, but
> didn't remain true for very long.

Something to remember is that in early 90's, floppy disks was the only
affordable way hobbiists to get OS's installed on x86 systems.  Even
OS/2 as distributed from IBM / Microsoft came on 30+ floppy disks.  In
1990, CD-R recording system cost $35,000 (and dollars were bigger back
then).  In 1992, the price had dropped to $10-12k, and it wasn't until
1995 that he first CD-R system under $1000 was available.

So I would argue that Linux was *easier* to bootstrap than
NetBSD/FreeBSD during that era.  The fact that we could shrink a
kernel and a root file system down to two 1.44 MiB floppy disks
required an on-trivial engineering effort, and it meant that all you
had to was to download and write half-dozen to a dozen flopy disks,
and then it was *trivial*.

In contrast, bootstrapping a BSD system if you didn't have a
quarter-inch tape drive (which was $$$) was non-trivial.  So I would
argue the reverse; the fact that Linux was easier to install may have
helped it.

> 5) I think FreeBSD's ports and similar huge-source-tree approaches
> didn't work out as well Linux developers contributing their changes
> upstream.

I'd frame this slightly differently.  The fact that we had multiple
Linux distributions meant that we had competition to make a better,
easier-to-install userspace, while keeping a common kernel.  Also,
distributions cooperated with each other in a very surprising way.
One archetypal story was one where at a Linux meet-up, Bob Young, who
was one of the founders of Red Hat, was helping to hand out Slackware
CD's because that was what was available.  Bob's philosophy was that
growing the pie was way more important that fighting over the share of
the pie.

In contrast, during that era, NetBSD and FreeBSD were busily
quarrelling with each others, with politics and ill-will due to people
being ejected from the core team which caused the various BSD forks.
I can't imagine this being helpful in the long term....

In particular, the kernel engineers who were hired by the distribution
vendors were working together on a common kernel, and on low-level
userspace subsystems (glibc, PAM, etc.) were also done with a huge
amount of cooperation.

						- Ted