Re: Wanting opinions... -- widespread lack of UNIX history

["Bryan J. Smith <b.j.smith@ieee.org>" <thebs413@earthlink.net>]

NZG wrote:
>   "De Raadt says BSD could have become the world's most popular
>    open source operating system, except that a lawsuit over BSD
>    scared away developers."
> LOL! And the Linux community hasn't been threatened? hello, SCO...
> I have never even heard of a lawsuit against BSD.

This is what SCO counts on, people who are completely oblivious
to the '70s and the post-AT&T breakup by the US federal government
that allowed AT&T to make UNIX a product.  Step 2 was to sue
UCB over BSD UNIX.

That lack of knowledge is why people think Linux is based on UNIX,
when Linux itself is a "clean room" UNIX-clone.  Prior to the SCO
lawsuit, I have _long_advocated_ that we technically refer to Linux
as a "GNU System" because of this -- both for technical and historical
reasons.  I could really _care_less_ if people call it GNU/Linux, but
from both a developer _and_ a historical perspective, I have
advocated it be refered to as a GNU System.

BSD (as well as SunOS/Solaris) is a GNU compatible system, but not
a GNU system by design.  In the absence of the Linux kernel and
adoption in the early '90s, SunOS was the preferred platform for
GNU.

> We don't use it because it doesn't have nearly the hardware support,
> community, documentation, or features of Linux.

That's post-facto!  At one time, 386BSD kicked the living crap out of
Linux's support -- and Linux heavily "borrowed" from BSD.  Even today,
there are large sections of a Linux system that have lineage to BSD,
although they are all (AFAIK) 4.4BSDLite which is the legally "clean
BSD" per the terms of the AT&T v. UCB settlement.

The 4.4BSDLite derrivatives are still very good for many applications.
So don't sell them short.  You're comments are completely post-facto.
By the time AT&T v. UCB ended, there was a leading GNU platform
in Linux.  And many companies like to leverage the GPL/LGPL and other
copyleft licenses because they mean they cannot be "leeched" into
a proprietary product, resulting in more forking.

People complain about forking in the Linux world, but that's largely
just marketing.  If you consider all the countless BSD codebases
that have been forked into dozens or even hundreds of splinter
commercial products -- it quickly shows the problem.  E.g., the
database world is largely either UCB Ingress or Postgres derrived.

And there are _more_ BSD UNIX variants than just the 3 Freedomware
ones. 


From: Gary Smithe <gary.smithe@gmail.com>
> NZG,
> I thought the BSD lawsuit was common knowledge:
> http://en.wikipedia.org/wiki/Berkeley_Software_Distribution

Nope.  Most people using Linux are oblivious to it.  Hence why much
of the media has ignorantly given SCO some credit because they
believe Linux is based on UNIX, when it's completely either GNU or
other clean-room efforts or 4.4BSDLite in a few cases.

> The lawsuit slowed development of the free-software descendants
> of BSD for nearly two years while their legal status was in question,
> and as a result systems based on the Linux kernel, which did not
> have such legal ambiguity, gained greater support.

The AT&T v. UCB lawsuit, combined with the AT&T Standardization
efforts around System V from 1986 on-ward really _stalled_ a lot of
contributions to the BSD codebase.

Many vendors started getting behind MIT projects and GNU, including
Digital, Sun, etc...  Some came up with their own, BSD-like (MIT
licensed) clean-room efforts, while others supported GNU.  Probably
the single biggest mover in the GNU world was Michael Tiemann
and his resulting Cygnus** commercial effort.

It wasn't until the settlement and resulting 4.4BSDLite codebase
that BSD finally got more interest.  But by then, it was 1994, and
Linux was already very capable and mature as an Internet platform.

[ **NOTE:  Although Cygnus was always a privately held company
so we'll never know, at its time of purchase by Red Hat (a move
that basically bloated Red Hat 3x in employees), Cygnus was
rumored to have more annual revenues than _all_ Linux companies
combined, and a healthy profit margin that has only been recently
exceeded by RHEL subscriptions.  I.e., the main reason Red Hat
went from deep red to black was because of their Cygnus purchase,
until they finally started offering RHEL some 2 years after the
acquition completed. ]

> Linux and 386BSD began development at about the same time,

You have to actually think back to before that (as I did above).

But yes, 386BSD started up as the lawsuit was clearly looking
like it was going to be settled as of 1991-1992 (AT&T was looking
to get out from under UNIX).  But it wasn't until the actual
settlement and finalization of the 4.4BSDLite codebase that
corporations could shed the indemification issue.

Again, by then, Linux was already a viable Internet platform.

> and Linus Torvalds has said that if there had been a free
> Unix-like operating system on the 386 at the time, he likely
> would not have created Linux.

Yep, and I believe him too.  GNU was entangled in a CS ideal
of a microkernel, and Linus believed a monolithic kernel was the
way to go, hence why BSD would have been a natural choice.

But instead, Linux leveraged his new "target" for the i386+MMU
by using a combionation of Minix, Solaris and GNU tools.  The
result, despite a chastizing by Tannebaum** for designing a
monolithic kernel that required a MMU, is history.

[ **NOTE:  Tannebaum is a greatly respected CS authority, and
his comments made sense from a viewpoint.  His works are
required study for both CS and ECE majors alike.  But like many
CS ideals that are innovative -- Intel's EPIC and Predication in
the IA-64 is my favorite of recent -- they don't always work out
as well in the "real world."  RT/Linux is more of a microkernel
approach when you need response time, and I will never
believe a "pure" microkernel is ideal for a general OS.  Much
like Intel's compiler-focused EPIC/Predication will never due
for superscalar architectures either. ]



--
Bryan J. Smith   mailto:b.j.smith@ieee.org


-
To unsubscribe, send email to majordomo@silug.org with
"unsubscribe silug-discuss" in the body.