Re: Research questions WRT SCO's complaint

On Sat, Mar 08, 2003 at 09:26:50PM +0000, Karsten M. Self wrote:
> I'm doing some factual research on the SCO issue.  In particular there
> appear to be numerous factual errors in the complaint.
> 
> Anyone with answers to the following:
> 
>   - What x86 Unices have there been?  When did they emerge?  I'm aware
>     of Xenix (which became SCO), Solaris X86, BSDi, FreeBSD, OpenBSD,
>     NetBSD, AIX/86, Dynix, and (for a broad definition of Intel
>     architechtures), QNX.  I'm not clear of timelines.  And any others
>     would be welcomed.

On Sat, Mar 08, 2003 at 01:39:38PM -0800, Rich Persaud wrote:
> Any others can also augment  http://www.levenez.com/unix/history.html .

A few points to note.

- SunOS 1.0 was a 4.1BSD derivative.  It gradually picked up pieces 
  of SVR4 over the years and became Solaris.  By the time it became
  Solaris x86, it was a SVR4 platform; Sun had licensed the right to use
  UNIX for Solaris in perpetuity from USL by this point.  (One of the
  reasons why Sun is saying that their OS is safe from lawsuits today.)

- Both 386BSD and BSD/386 were initially released in Feburary 1992,
  contemporary with the early Linux 0.x kernels.  Both of these BSDs
  are 4.3BSD Lite (BSD Net/2) derivatives, which was an attempt to 
  unencumber the BSD sources from AT&T/USL licensing.

- 386BSD petered out and forked into FreeBSD.  Many companies
  considered FreeBSD to be "enterprise grade" since at least the
  2.x days in the mid 90s (Yahoo!, Hotmail, the former Walnut Creek
  CDROM to name just three).  

  (cdrom.com boasted the single-server single-day bandwidth record
  for quite a while, running apache and ftp services from a single
  Pentium II FreeBSD box.  I think they stopped making press releases
  after they hit the 1TB/day mark.)

- BSD/386 was a commercial operating system delivered by BSDi, and 
  widely regarded (er, marketed) as an "industrial strength UNIX" for
  the x86 architecture.

- 4.3BSD Lite eventually led to NetBSD, and OpenBSD through forking.
  Neither of these are considered to have the "enterprise grade
  features" that Linux or even FreeBSD have today.  NetBSD focuses on
  extreme portability, while OpenBSD focuses on portability and
  security.  (Journaling filesystems or running Oracle / DB2 is
  not a priority for the OpenBSD folks.)

- Other UNIX architectures that have appeared on x86 include Mach,
  NeXTSTEP 3.3, and standard unadorned 4.xBSD.  Darwin is the
  modern derivative of Mach/NeXTSTEP, and also runs on x86 today.

- Many of the "enterprise grade" features that SCO alleges IBM added
  to Linux still do not exist for x86 BSD.  SMP beyond two processors
  is a good example.  *BSD does not have a standard journaling
  filesystem, but one is in progress for the upcoming FreeBSD 5.0
  release, which borrows heavily from the work done on journaling
  filesystems for Linux, including those that have no contact with
  SCO/IBM.  (reiserfs and SGI's xfs come to mind)

  Usually, "enterprise grade" in the context of *BSD has been
  shorthand for stability/uptime, throughput and performance, not 
  OS partitioning, clustering (esp. failover), 4-way SMP or greater.

- QNX is not a UNIX per se.  It started in 1981, and appears to
  have been influenced by the 4.4BSD sources around 1993, before
  the release of QNX 4.1.  It's also targeted at embedded systems, 
  not "enterprise grade" UNIX.

- Depending on how broad your definition of "UNIX" is, you may also
  want to include Minix, Coherent and Plan 9.
 
>   - What's the timeline of Project Monterey?  Best I can find, it was
>     launched 1998.  That's seven years after the first release of
>     the Linux kernel. 

Yes, but Monterey wasn't created as a completely new OS project.
It was chartered as a single multi-vendor UNIX port to bring SVR4
to ia64.  According to this news story, it sounds like Monterey
was still a multi-company management vision in September 1998, but
the vision dates back to 1996:

	http://news.com.com/2100-1001-215643.html?tag=rn

	Apparently, the idea is not new to Intel. According to Oracle's
	Larry Ellison, now-retired CEO Andy Grove was the driving force
	behind earlier efforts to amalgamate Unix operating system
	variants into a single version. Grove was involved in these
	efforts as recently as two years ago, News.com previously
	reported.

Implementation of Monterey appears to have been spearheaded by the 
SCO-IBM-Sequent triad in 1999, when other vendors were still planning on
porting their own UNIXes to Merced/Itanium.  (Remember that IBM bought
Sequent).

	http://news.com.com/2100-1001-229834.html

	Monterey-64 is one of several operating systems that Intel is
	helping to prepare for the arrival of Merced. Others include
	Sun's Solaris, Hewlett-Packard's HP-UX, Compaq's Tru64 Unix,
	Microsoft's Windows 2000, and Linux. IBM and SCO are hoping to
	make Monterey-64 sell in the highest volumes.

With this perspective, you could argue that SCO is angry that IBM is
actively killing Monterey by making high end UNIX features available for
Linux on (1) ia64 and (2) x86.  A more realistic perspective is that SCO
is that the market doesn't care about either Monterey or ia64 *at* *all*,
and is blaming IBM for the lack of customer adoption.

Z.