SCO would like to redepose Otis Wilson, and they will. No doubt they are hoping to undermine his prior testimony on AT&T/USL policies on methods and concepts. However, we've found a document dating from 1995 that supports Mr. Wilson's, and IBM's, position. We find our old friend errno.h in this paper too, along with all its cousins, publicly on display, and here's the kicker: the copyright page says it was published with the permission of USL.
The paper, SYSTEM V APPLICATION BINARY INTERFACE, PowerPC Processor Supplement, by
Steve Zucker of Sun and Kari Karhi of IBM, interestingly enough, was published in September of 1995, Part No: 802-3334-10 Revision A, with an earlier 1993 IBM copyright. The website's description of the document goes like this:
This document describes the PowerPC Application Binary Interface for SVR4 Unix. It details parameter passing conventions, the OS interface,
object file format, debugging format... etc. Unfortunately, it seems
that the pages are in a backwards order... so the first page is page
150, the 150th page is page 1.
Well, well, indeed it do, as they say. (Update: Here's a version [PDF] in the right order.) So we have a document that precedes IBM's contributions to Linux by a mile that is pertinent to both the PowerPC processor and the SVR4 ABI internals with detailed ABI information that even has a Sun part number, publicly available since at least 1995. And here's the copyright page's wording:
© 1995 Sun Microsystems, Inc. All rights reserved.
© 1995 IBM Corporation. All rights reserved.
This specification includes material copyrighted by UNIX System Laboratories, Inc., which is reproduced with permission.
What does it all mean? I asked Dr. Stupid to explain it all to me. Also SCO might like to know, so its lawyers don't torment Mr. Wilson with questions that this paper does away with. Earth to SCO: USL let the dogs out in the mid-90s.
Here's what Dr. Stupid explained to me about this paper: This is the detailed processor-specific supplement for UNIX (i.e.
System V) on PowerPC chips. As mentioned previously, including by SCO itself, the System V ABI
has two parts: a common part (the "gABI") and the part that is
slightly different for each processor architecture. This document is
Now, it is important to note, he pointed out to me, that oldSCO, Santa Cruz Operation, dealt in UNIX on Intel x86
chips, for which there was a separate document, which Groklaw published already. So there might be information given in the x86 document that
isn't given in this (PowerPC) one. That said, the differences in
information would not be differences in copyrightable code, but in
low-level "agreed ways of doing things" such as what registers to use
(e.g. look at page 3-14 in the PowerPC document), information that I believe
would fall under trade secret protection, if anything. However, as you can see, both the x86 and PowerPC supplements were openly
published, and even that assumes that the "way of doing things" was
devised by USL -- they may have been devised or
co-devised by the chip manufacturers (e.g. Intel and Motorola). Now, if
it is a given that the objective of publishing these specs was to define
"this is how you do a System V ABI," then wouldn't you, by definition, end up
with only one way of doing something, with the result that the merger doctrine kicks in
even for those small bits of source code (header files) that are in
the ABI documents?
Note that errno.h appears again on page 6-11 - our friend sure gets around.
Possibly the most significant point about the whole document is that
USL granted permission for this material to be openly published. It
shows clearly that at the time, USL was not interested in
keeping this material secret. And of course, any material USL
published (or allowed to be published) would then presumably no longer be covered
by any non-disclosure clauses in UNIX licensee's contracts. IBM's contract specifically freed them from any secrecy requirements if material was made public by anyone other than IBM.
I thought it'd be nice to have this handy before the deposition, knowing that SCO reads Groklaw religiously, with the hope that poor Mr. Wilson will not have to be toasted and roasted over the fire at least regarding materials publicly revealed so long ago in this document, made public with the express permission of USL to IBM itself in 1995, and from the copyright page some portion at least by 1993. I'd call it a methods and concepts doomsday document.