In Judge Dale Kimball's July 2008 order in SCO v. Novell, he described OpenServer like this:
E. OpenServer I think that isn't completely accurate. SCO began declaring their full ownership, but it's not that simple a story. Santa Cruz was, after all, a licensee, using AT&T's -- later Novell's -- UNIX System V Release 3 code. So obviously not all the code in OpenServer would belong, in the copyright sense, to SCO. In fact, we would expect that we'd find copyrights belonging to AT&T and Novell. And we do. Let me show you.
OpenServer is the brand name for the release of UNIX System V, Release 3 that Santa Cruz developed in the 1980s. Novell never owned, or had any license to, OpenServer. OpenServer was Santa Cruz's flagship product through the 1990s. OpenServer produces two-thirds of SCO's UNIX revenue and has thousands of customers, including small to mid-sized businesses and large corporations, such as McDonald's.
Here's what the output of the "copyrights" command under OpenServer 5.0.5
gave, and I've highlighted all the owners that I found particularly interesting, in addition to Novell and AT&T:
This SCO software includes software that is protected by
(C) 1983-1996 The Santa Cruz Operation, Inc.
1989-1994 Acer Incorporated
(C) 1989-1994 Acer America Corporation
1990-1994 Adaptec, Inc.
(C) 1993-1994 Advanced Micro Devices, Inc.
1990 Altos Computer Systems
(C) 1992-1994 American Power Conversion,
(C) 1988 Archive Corporation
(C) 1990 ATI Technologies,
(C) 1976-1992 AT&T
(C) 1992-1994 AT&T Global Information
(C) 1993 Berkeley Network Software Consortium
1985-1986 Bigelow & Holmes
(C) 1988-1991 Carnegie Mellon
(C) 1989-1990 Cipher Data Products, Inc.
(C) 1985-1992 Compaq
(C) 1986-1987 Convergent Technologies, Inc.
1990-1993 Cornell University
(C) 1985-1994 Corollary, Inc.
(C) 1994 Dell
(C) 1988-1993 Digital Equipment Corporation
1990-1994 Distributed Processing Technology
(C) 1991 D.L.S.
(C) 1990 Free Software Foundation, Inc.
(C) 1989-1991 Future
(C) 1994 Gradient Technologies, Inc.
(C) 1994-1995 IBM Corporation
(C) 1989 Irwin Magnetic Systems, Inc.
(C) 1988-1991 JSB Computer Systems Ltd.
(C) 1989-1994 Dirk
(C) 1987-1994 Legent Corporation
1988-1994 Locus Computing Corporation
(C) 1989-1991 Massachusetts Institute
(C) 1985-1992 Metagraphics Software Corporation
1980-1994 Microsoft Corporation
(C) 1984-1989 Mouse Systems
(C) 1989 Multi-Tech Systems, Inc.
(C) 1991-1995 National
(C) 1990 NEC Technologies, Inc.
(C) 1989 Ing. C. Olivetti & C. SpA
(C) 1989-1994 Open
Software Foundation, Inc.
(C) 1993-1994 Programmed Logic Corporation
1989-1995 Racal InterLan, Inc.
(C) 1990-1992 RSA Data Security, Inc.
1987-1994 Secureware, Inc.
(C) 1990 Siemens Nixdorf Informationssysteme
(C) 1991-1992 Silicon Graphics, Inc.
(C) 1987-1991 SMNP Research,
(C) 1987-1995 Standard Microsystems Corporation
(C) 1984-1994 Sun
(C) 1987 Tandy Corporation
(C) 1992-1994 3COM
(C) 1987 United States Army
(C) 1979-1993 Regents of the
University of California
(C) 1993 Board of Trustees of the University of
(C) 1989-1991 University of Maryland
(C) 1986 University of
(C) 1976-1994 UNIX System Laboratories, Inc.
(C) 1988 Wyse
(C) 1994 X Consortium
(C) 1992-1994 Xware
Eric P. Allman
(C) 1987-1989 Jeffery D. Case and Kenneth W. Key
(C) 1989 Mark H. Colburn
(C) 1993 Michael A.
(C) 1982 Pavel Curtis
(C) 1987 Owen DeLong
(C) 1993 Carlos Leandro and Rui Salgueiro
(C) 1992 David L. Mills
(C) 1992 Ranier Pruy
1986-1988 Larry Wall
(C) 1992 Q. Frank Xia.
Now, we're guessing that most of these copyright
attributions were actually for added-on packages (Larry Wall being a good
example) rather than copyright attributions for actual OpenServer code. But not the copyrights belonging to Berkeley, AT&T, Novell, USL (which Novell bought), or the Regents of California, for example.
Isn't that interesting? So just because code is in OpenServer, it doesn't mean it necessarily belongs to SCO. It will have to provide evidence that any code it alleges is infringed actually does belong to it. If it's AT&T's UNIX code, it obviously does not, since Novell didn't pass along any UNIX copyrights to SCO in the 1995 deal, as the Utah District Court ruled.
As it happens, in the older OpenServer, there was a developers kit that you got separately, as you can see in this announcement [PDF]:
The SCO OpenServer Development system is comprised of a set of state-of-the-art compilers, debuggers, application programming interfaces (APIs), and libraries needed to develop applications. The SCO OpenServer Development System can also be augmented by over 200 third-party development tools to create the most robust and efficient development environment. Maybe all of that belongs to SCO and maybe that is where the code in the AutoZone case comes from?
Not so fast. Notice this posting from 2000 to comp.unix.sco.misc, and please in particular notice the copyright notices he finds in the UnixWare/OpenServer Development kit:
So even in this development kit, Novell owns copyrights on some of the contents, it appears all of it up to 1995, judging from this posting. So, the question in AutoZone will be, who owns the CompX and DecompX and other allegedly infringed code? Did the licenses restrict use to only OpenServer? We're researching that very thing, actually, at Groklaw, and we'll share whatever we find. So far, SCO's hoped for allegations in the proposed amended complaint include this list:
Subject: Attempting to install Openserver Development kit
X-Http-Proxy: 1.0 x32.deja.com:80 (Squid/1.1.22) for client [redacted ip address]
Organization: Deja.com - Before you buy.
X-Article-Creation-Date: Fri May 12 16:10:16 2000 GMT
X-Http-User-Agent: Mozilla/4.72 [en] (Win98; U)
I am hoping to install Openserver development kit on this PC
Unfortuneately, there the instructions on the back of the attached
paper are woefully incorrect as there is no '/cdrom/info/install.htm'
Anyway I blundered through all the files and figured this was how
to do it, obviously I was wrong, or am i doing something wrong.
I have seen there is some patch to install but this is a fresh install
of OSR5.0.5Eb ....
UnixWare/OpenServer Development Kit
Version 7.1.1 (IA32)
(C) Copyright 1996-1999 The Santa Cruz Operation, Inc. All rights
Copyright 1984-1995 Novell, Inc. All Rights Reserved.
33. To illustrate, SCO's limited discovery to date confirms that AutoZone engaged in the following activities, among others, in violation of SCO's contract rights and copyrights:
AutoZone has debunked much of this already, years ago when SCO first started talking about it, and its primary points include the fact that it had a license to OpenServer. And it had a license from a third party, as SCO acknowledges, to use CompX and DeCompX. So what is the beef about that, one wonders? Of course SCO's initial claims are always as big as the sky, and by the time trial comes around, they get smaller and smaller until, at least in the Novell case, they disappear almost entirely.
AutoZone developers copied 1,681 separate COFF files onto at least 387 AutoZone store machines located throughout the United States.
AutoZone developers copied 28 COFF files consisting of sort files, help utilities, and other miscellaneous files, onto all its machines in 3,500 AutoZone stores in the United States and Mexico.
AutoZone copied two COFF files, Compx and Decompx, onto the machines located in those stores. These files were programs that AutoZone had licensed from a third party which contained proprietary SCO code. AutoZone used these files at least from January 2000 until it deleted them during the Court-ordered discovery process. When AutoZone deleted Compx and Decompx from its Linux servers, the replenishment system used by AutoZone to replace inventory from its warehouses failed on approximately 650 machines.
AutoZone's machine load computer was found to contain a program entitled dexpand.x that was compiled under SCO's proprietary OpenServer operating system.
AutoZone copied over 4,500 programs that were compiled to run on OpenServer onto AutoZone's "Spirit Server" which was used to store AutoZone's source code in its headquarters. The vast majority of these programs contain some portion of SCO's proprietary static libraries. AutoZone has admitted to copying on Spirit at least 1,130 programs compiled to run on OpenServer.
In addition, with the aid of a software tool written by SCO's technical consultant, AutoZone discovered an additional fifteen SCO Extensible Linking Format
("ELF") and Xenix files which were also compiled to work on SCO proprietary operating systems (earlier versions of OpenServer that were licensed by AutoZone). AutoZone admitted that those files "likely also exist on all 3500 AutoZone store servers."
AutoZone copied approximately 370 programs onto its Linux development machine known as "Wrangler." The majority of these programs appear to contain some portion of SCO's proprietary static libraries.
AutoZone developers copied numerous SCO files, the precise number of which has not been disclosed in discovery, onto AutoZone's "Vision" server which was used in part by AutoZone to compare the output of programs that it was porting from OpenServer to Linux, to ensure that the output was identical.
SCO mentions COFF files, and in fact when it sent the letter to Linux users [PDF] about alleged infringement years ago, it listed AT&T code it claimed to own that SCO alleged Linux wasn't allowed to use according to the terms of the then-secret
settlement agreement, including the file arch/mips/boot/ecoff.h.
Well, as we learned later, SCO doesn't own any of those files, because they didn't get those UNIX copyrights from Novell in 1995. Does that impact the AutoZone claims? What if the COFF files SCO is now accusing AutoZone of copying actually stem from that older UNIX code SCO doesn't own? I believe this is the right question if the case is permitted to go forward.
SCO now wishes to amend its complaint to sue AutoZone over copyright third party files Compx and DeCompx. It says it believes there is proprietary code of SCO's inside those two files. It "believes" it? Remember these are binary files, which is one reason Autozone had no possible way to know what was in them. It's comparable, in my mind, to getting sued for copyright infringement by a third party for using Microsoft Word, because of allegations some third party code is in Word. Why wouldn't you sue Microsoft instead of me? What is SCO's basis for its belief it has copyright ownership of anything in CompX, etc.? I think this PDF is pertinent. You'll find both compx and decompx showing up in the installation log in the instructions from the US Department of Education on page 15, but first, this is what the paper is about:
Welcome to the U.S. Department of Eduction's Federal Student Aid (FSA) Student Aid Internet Gateway (SAIG) that offer Title IV-eligible post-secondary institutions, third-party servicers, state agencies, lenders and guarantors, a secure, Internet-based method of exchanging Title IV data with the FSA Applications Systems. The SAIG replaces what was formerly known as "TIV WAN" by moving Title IV transmissions from the General Electric (GEIS) value-added network to the Internet. And here's what you find on page 15:
Installing TDAccess in directory /home/jtest/TDAccess2.2... So, if this is the same code SCO is referring to, the next correct question would have to be, what were the license terms for using Compx and Decompx?
Decompressing TDAccess 2.2 Installation file (this may take a minute or two ...)
TDCompress Build 0465 (master, triple DES) (c) Copyright 1990-2003
DECOMPRESS STARTED - Tue May 20 11:52:00 2003
So, there's the map with the X on it. If you'd like to help us dig, please do share whatever you find. If you weren't with Groklaw back in 2005 when SCO first raised some of these allegations, you might like to read the article we published, SCO v. AutoZone - What Are Statically Linked Libraries, Anyway?.
Update: Here's an article in Unix Guardian, SCO OpenServer 6 Launches with Unix SVR5 Kernel, from June 2005, in Volume 2, Number 24, by Timothy Prickett Morgan, that talks about the kernel in OpenServer in the version that AutoZone might have licensed, if not an earlier one, replaced only in 2005 by a SVR5:
While OpenServer 5.0.7 is notable in that it provided some limited support for UnixWare 7 applications,OpenServer 5 was based on the Unix System V Release 3.2 kernel, which is very old and has some pretty severe limitations in terms of threading, main memory, and file system support. That's why SCO bought UnixWare and the rights to the Unix operating system created by AT&T from Novell to have a more scalable Unix than OpenServer. To preserve backward compatibility with the large installed base of OpenServer customers--there could be as many as 1 million servers installed in the world that are running OpenServer and UnixWare--SCO has not messed with that kernel, even as Unix System V was updated to Release 4 and then Release 5. With Legend, that changes, and OpenServer now uses the SVR5 kernel while maintaining backward compatibility with all prior generations of OpenServer, Unix, and Xenix Unixes from SCO. Yes, you can still run Xenix 286 binaries developed in 1986 for the 16-bit 80286 processor on today's 64-bit Xeon processors from Intel on top of OpenServer 6, bragged Sandy Gupta, SCO's vide president of development, at the announcement this week. He said, in fact, that some game developers had tested Xenix 286 games on OpenServer 6 to make sure they still ran. Obviously, SCO does not own the copyrights on Unix System V Release 3.2.
Novell does [PDF]. Although SCO tried to claim it [PDF] too, the Utah District Court has ruled it remained with Novell and did not pass to SCO in 1995 under the APA. So in the AutoZone case now going forward in Nevada, the judge there has said to go forward with the Utah decision, and in that context, SCO doesn't own the code at issue, from all that I can see. You might notice that in both attempted copyright registrations, SCO and Novell agree on one thing: the date of the OpenServer System V release 3.2 code is 1988.
Incidentally, if you are interested in COFF files, here's a good place to start, a SCO manual for SCO OpenServer Release 5.0.7 dated February 11, 2003, which is smack dab in the middle of SCOsource, while it was still ambulatory, and notice this page tells us who developed ld:
ld(CP) See what I mean? Those of you who are programmers may find even more helpful info in that manual than I would.
ld -- invokes the link editor
ld [options] filename [[options] filename]
The ld(CP) command combines several object files into one, performs relocation, resolves external symbols, and supports symbol table information for symbolic debugging. Both ELF and COFF file formats are supported (see ``Generating COFF vs. ELF binaries'') and ld can operate in either static or dynamic modes (see -d). ...
Generating COFF vs. ELF binaries
When ld is called, it scans all object files and library files (if any) given through filename arguments on the command line. It determines whether an object file is in COFF or ELF format and executes the appropriate binaries in each case. The file format of a library is treated as the same as the file format of the first module in that library file. ld defaults to ELF format if none of the user specified object files are in ELF or COFF format. If at least one of the user specified object files is in ELF format, all COFF modules in the group will be converted to ELF and the resulting binary will be in ELF format. If none of the user specified object files are in ELF format and there is at least one user specified COFF object, then the resulting binary will be in COFF format. To force the resulting binary to a particular format, use, for example, the option -b elf or -b coff. Since this utility relies on the files in /usr/ccs/bin/coff, /usr/ccs/lib/coff, /usr/ccs/bin/elf and /usr/ccs/lib/elf, changing or moving any files in those directories results in an error....
ld is not part of any currently supported standard; it was developed by UNIX System Laboratories, Inc. and is used by permission.