SCO tool chain USWAG


July 19, 2006

This is a dumb but curious question regarding the SCO-Linux tool chain correllation.

Yesterday, an anonymous poster on Groklaw, at 01:26 PM EDT, asked about a comment in Appendix A

a. Misappropriation in the form of line-for-line code copied from System
V into the Linux kernel (Item Nos. 183-185, 205-231) as well as line-for-line
code copied from System V into the Linux tool chain used to compile and operate
Linux (Item No. 272) and line-for-line code copied from System V into STREAMS
modules used by, among others, enterprise Linux customers to operate
"Carrier Grade Linux" (Item Nos. 150-164);

nixaware, on Yahoo!SCOX at 06:15 PM, noted that adherence to standards could cause this similarity.

I made a subpost, and observed that SCO products often appear to be borrowed from Open Source sources.

My subpost motivated nixaware to do some research, and come up with this, posted at 1:50 AM:
>If they compiled their kernel with GCC...<

Okay, you made me look.

"This tool kit includes the GNU Tool Chain including a set of compilers, debugger, object tools, libraries and other open source tools that can be used to develop, enhance and compile UNIX® programs..."

I must be missing something.

These people are not really complaining that they appropriated the GCC compiler and tool chain, that the compiled kernel from SCO now looks like the Linux kernel because of use of H.J.L.'s binutils, therefore there is literal copying?
< EOM >

10:20:41 AM

Re: SCO tool chain USWAG



July 19, 2006

<< These people are not really complaining that they appropriated the GCC compiler and tool chain, that the compiled kernel from SCO now looks like the Linux kernel because of use of H.J.L.'s binutils, therefore there is literal copying? >>

No they aren't.

The 'tool chain' meme comes from what 724 Appx A says about Item 272: "Line-for-line code copied from System V into the Linux tool chain used to compile and operate Linux". (page 2). Note the words *operate* *Linux* which is completely inapplicable to any appropriation of GCC into Unixware.

There's more. Look at what Page 3 says about Item 272: "SCO's proprietary System V code was simply copied and pasted into the Linux kernel or associated libraries that were then included in a Red Hat distribution. (Item Nos. 183, 184, 272.)" Again, "Linux kernel or associated libraries" and "in a Red Hat distribution" would be completely inapplicable to an appropriation of code from GCC into Unixware.

So we can rule your Unixware theory right out. Sure, they *distributed* GCC in *aggregation* with their own toolchain, but that's grist to the estoppel and counterclaim mill, not to the 294 allegations.

Now look at document 707, page 13: "Item 171 of the December Submission disclosed that the entirety of ELF, including the ELF specification, is being challenged. (December Submission at Item 171, Ex. 15 ("Use of SVR4 ABI as source reference in Linux programming, revealing details of SVR4 ELF/ABI specification and use of SVR4 ELF/ABI specification to develop Linux."); see also December Submission at Items 272-75, Ex. 15)."

It's clear they are challenging everything in the Linux ecosystem that implements ELF - kernel, dynamic linker, and, yes, GCC and binutils (both of which have IIRC featured in SCO filings many moons ago) - both as Methods/Concepts, and as code where snippets reminiscent of the published spec can be discerned.

Finally, there's as yet no indication that GCC and binutils were explicitly fingered in the December Submission. If I was to wild-ass guess the items Rochkind says he kyboshed, GCC, binutils and sysVinit would be top of the list.

< EOM >

10:47:42 AM

Re: SCO tool chain USWAG


August 4, 2006

I went back and looked at my parent post, and re-read PJ's Groklaw article "SCO is Distributing ELF Under the GPL Still. Yes. Now. Today." from Monday 03:56 PM.

I think I got it right the first time.

If I am following Hatch's argument correctly, and am reading his Exhibit "G" correctly, and am correct in my assumption that the binutils file from "The SCO Group" website remained essentially unchanged from 2001, then my conclusion follows.

To-wit: SCO added works copyrighted by the FSF (Free Software Foundation), namely the binutils, to their distribution, long ago. Because they long ago incorporated those outside works into their systems, they have collectively forgotten the actual source. Because they do not recall the actual sources, they have presumed that pieces like binutils are properly part of UnixSysV. When they compare current Linux distributions with their in-house source, there are very predictable matches. In particular, the kernel is strikingly similar, particularly in the ELF areas.

If I were a judge, the following, as the only copyright notice within a document, would be proof that "The SCO Group" *is not* the owner of the copyright on binutils:

/* ELF support for BFD.
Copyright (C) 1991, 92, 93, 95, 1997 Free Software Foundation, Inc.
Written by Fred Fish @ Cygnus Support, from information published
in "UNIX System V Release 4, Programmers Guide: ANSI C and
Programming Support Tools".

5:48:28 PM

Re: SCO tool chain USWAG


August 4, 2006

Sorry AP, you are off base here. ELF (extensible linker format) is a successor to COFF (Common Object file Format) and both were developed by AT&T and friends, presumably including, but not restricted to SCO. In order to allow GCC to create object files that were linkable with the ELF based object files on Unix boxes, GCC needed the ability to generate ELF files. This all predates Linux. At some point, Linux acquired the ability to load ELF files in addiion to a.out files. I don't know when this happened.

In any event, this is all ancient history, and the entire purpose of ELF format was to allow third party compiler and library developers to generate object files that could be linked together. There was never any intent that I am awae of to restrict the use of ELF to SVR4 licencees. SCOX seems to be operating on the theory that since AT&T never explicitly gave permission to anybody to implement ELF, any implementation other than AT&T's is some sort of license violation. The entire notion is idiotic and if upheld would effectively make reverse engineering of anything illegal.

6:03:50 PM

Re: SCO tool chain USWAG


August 5, 2006

monsieur_bobo, I have long appreciated your observations. Thank you for taking the time to reply to my earlier comment.

In your message, you suggest that I am off base in my conclusions about this source of ELF and binutils. You are probably correct, and this is not the first time that I will have been off base. It won't even be the first time today.

However, when I re-read the history of ELF 1.2, this is what I find:

1991 – AT&T splits off USL as a separate company.

Pre-1993 - USL grants ELF license to the TISC (Tool Interface Standards Committee), as a precondition to revamping it for use as an industry-wide standard.

The TISC members, including IBM and USL, jointly author the new ELF format from the old USL ELF format.
(Is this where the Extensible Linker Format became the Executable and Linkable Format?)

1993 - TISC grants license to use their new product: ELF 1.1. Linus Torvalds conceives Linux. Novell buys USL. Novell crows about their contribution of ELF to the public.

1995 – TISC grants license to use their next new product, ELF 1.2, saying in part:
"The TIS Committee grants you a non-exclusive, worldwide, royalty-free license to use the information disclosed in this Specification to make your software TIS-compliant."

Later that year, Novell and the Santa Cruz Operation conclude their APA, which includes an appendix entry specifically denying that any copyrights are transferred.

I, myself, have never personally talked to anyone directly involved in these creative works. All I know is what I read on the Internet, and I do not believe a great deal of what I read.

Presumably, the TISC files have a copy of the USL-TISC license somewhere. The license says, essentially, USL grants the TISC a non-exclusive, worldwide, royalty-free license to use the ELF specification, and to re-license ELF and any TISC modification to the specification to anyone else under a non-exclusive, worldwide, royalty-free license. If “The SCO Group” possesses the business records from USL, then they would be presumed to possess the USL copy of the document granting the license. Because there is a conflict of interest for “The SCO Group” to release a document inconsistent with their current legal position, courts are sympathetic to introduction of collateral evidence to prove the existence of the license and its terms. Similarly worded contemporaneous licenses to TISC from other committee members, and in particular IBM, would be strong evidence that that this was standard conduct. Testimony from any of the actual individual participants, saying that was the policy, would be persuasive, especially when backed up by copies of contemporaneous licenses.

Also persuasive would be an "admission against interest" by "The SCO Group" or one of its predecessors-in-interest. Novell bragging to the press about their contribution of ELF to the public is nicely sufficient, IMHO.

The USL representative is presumed to be an authorized agent of USL. As such, he has the apparent power to contribute certain USL copyright IP to the TISC. In particular, the consequences of joint authorship apply under U.S. copyright code.

Among others, IBM and USL were members of TISC. If the license described above cannot be found, then IBM is a joint author of the work along with the other committee members. IBM would be a co-owner of an indivisible interest in the work. As a joint author, if the initial license cannot be located, IBM may license the work independently of the other authors. A license for exclusive use is the one bug in the ointment. None of the original committee members may independently sell an exclusive license to use the work. Unless they all do it together, it doesn’t happen.

The TISC members, acting jointly when exercising their individual authority, could have licensed ELF 1.2. Because they had the authority, the license is still valid, and was neither void nor voidable.

UnixSysV did not use ELF 1.2 in version 4, initially. It could not have. ELF 1.2 had not yet been created.

The author of binutils was apparently Fred Fish @ Cygnus Support, with initial versions starting in 1991, also pre-ELF 1.2.

I maintain my belief that they appropriated the GCC compiler and tool chain, that the compiled kernel from SCO now looks like the Linux kernel because of use of H.J.L.'s binutils, and therefore they conclude there is literal copying.

I also maintain my belief that Bill Gates does not possess enough money to persuade me to want to trade places with the hapless attorneys for "The SCO Group" that will have to continue to appear in court in this wretched case.
AllParadox - Retired Attorney, no legal opinions, just my opinion.

Correction: Re: SCO tool chain USWAG


August 5, 2006


Wait just a minute, now...


If you intend to stand forth as the universal authority so many people seem to accept you as, please be accurate in fundamental facts.



10:14:50 AM

Re: Correction: Re: SCO tool chain USWAG


August 5, 2006

OMG. I've been "outed".

All right. I'll 'fess up. I have clay feet all the way up to my waist.

(sigh) It was great while it lasted.
AllParadox - Retired Attorney, no legal opinions, sometimes not even good ones of my own.

10:24:02 AM

Source: Investor Village SCO Board [ ]

Copyright 2006