SCO invokes DMCA, announces earnings [LWN.net]

SCO has sent out a pair of press releases. the first release says that it has started some new "intellectual property enforcement" initiatives which are sure to be a hit with its customers: "Under the terms of SCO's System V UNIX contracts, the company has commenced issuing written notice to thousands of licensees requiring each licensee to provide written certification that it is in full compliance with their UNIX source code agreement, including certification that such licensee is not using proprietary UNIX code in Linux..." The PR also says that the company has sent out DMCA notices to "select Fortune 1000 Linux end users." The claim is "Certain copyrighted application binary interfaces have been copied verbatim from the UNIX System V code base and contributed to Linux without proper authorization and without copyright attribution."

Thanks to Avi Dines and Don Marti, here is a copy of the SCO letter. The interface in question would appear to be a bunch of error number and ioctl() code definitions. The letter is also available in the original MS Word format.

The second PR is the quarterly results. There's $10 million in licensing (from Microsoft and Sun only), and a net loss of $1.6 million. The inevitable conference call is coming; more afterward.

(Log in to post comments)

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 8:17 UTC (Mon) by danw6144 (guest, #14336) [Link]

Any Linux kernel developer who contributes code
under the unsigned, nonexclusive GPL license
retains the original copyright ownership. If the
developer subsequently legally transfers his
copyrights to a new owner, the code released under
the GPL is no longer protected by the GPL
from infringement claims by the new owner.

TITLE 17 - COPYRIGHTS
CHAPTER 2 - COPYRIGHT OWNERSHIP AND TRANSFER
Sec. 205. Recordation of transfers and other
documents
(e) Priority Between Conflicting Transfer
of Ownership and Nonexclusive License. - A
nonexclusive license, whetherrecorded or not,
prevails over a conflicting transfer of copyright
ownership IF THE LICENSE IS EVIDENCED BY A
WRITTEN INSTRUMENT SIGNED BY THE OWNER of the
rights licensed or such owner's duly authorized
agent, and if -
(1) the license was taken before
execution of the transfer; or
(2) the license was taken in good faith before
recordation ofthe transfer and without notice of
it.

Caldera contributed copyrighted source code files
to Linux under the nonexclusive UNSIGNED GPL.
Caldera sold to Novell who sold to SCO.

The old contributions under the GPL, because it's
unsigned, infringes on SCO's (the new owners)
copyrights.

The GPL is a binding legal agreement under a
public covenant though. That's the same thing as a
"contract license held in public trust". See IBM's
Amended Counterclaims. The GPL is valid and
binding under contract law. See Procd, Inc v.
Zeidenberg.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 8:33 UTC (Mon) by fandom (subscriber, #4028) [Link]

What did Caldera sell to Novell?

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 9:05 UTC (Mon) by danw6144 (guest, #14336) [Link]

They are reversed.

SCO invokes DMCA, announces earnings

Posted Dec 23, 2003 17:33 UTC (Tue) by bojan (subscriber, #14302) [Link]

You're forgetting that Caldera *is* the new SCO. They are one and the same - Caldera simply changed their name to The SCO Group. The old SCO is a different company now.

Troll alert

Posted Dec 22, 2003 9:26 UTC (Mon) by proski (subscriber, #104) [Link]

It's a troll. It's obviously wrong. It was clearly posted to attract attention and cause many replies. There have been similar posts in other stories, also first posts and also irrelevant.

Reading LWN isn't that fun anymore when such trolls are posted and there is nothing I can do about it. It remains the first post no matter how many readers point to the flaws in it. Sure, the articles are still interesting, but reading the comments leaves a bitter aftertaste. That's not something I pay money for.

I'm not suggesting to go the Slashdot way. There are other systems, such as Scoop which work much better because "mod down" has no cost except the one you have paid already, i.e. the time spent on reading the comment.

The current approach is blocking trolls by the editors. Unfortunately, as the trolls become less obvious, they become harder to identify by the editors.

(JOKE) alert

Posted Dec 22, 2003 9:56 UTC (Mon) by mmarq (guest, #2332) [Link]

C'mon you dont want to close yourself in a "all well" cristal dom!...
Even if possible the world would lose much of its fun...

Troll alert

Posted Dec 22, 2003 10:25 UTC (Mon) by TimCunningham (guest, #10316) [Link]

Yeah. Here's another vote for only allowing paid users to comment. I don't mind if they troll if they're paying LWN for the right... :)

Troll alert

Posted Dec 22, 2003 10:26 UTC (Mon) by vblum (subscriber, #1151) [Link]

I don't think this could count as a troll; posting the same comment in more than one place is still not a crime, although bad style. Yet, it's somewhat subjective to disqualify someone as a "less obvious" troll. Simply don't reply if you think they are, but sometimes people may make mistakes without being a troll.

You are right though that danw is off, whether deliberate or not:
Caldera did not "sell" to SCO. Caldera IS SCO. Renamed. Even if that claim about the unsigned GPL were true (which it is not - I am sure writing the GPL on one of the sold! OpenLinux Boxes counts as much as a signature), I cannot change my name and then say hey, all our previous implicit agreements are now invalid.

Copyrighting error and signal numbers?

Posted Dec 22, 2003 8:59 UTC (Mon) by eru (subscriber, #2753) [Link]

They are rather critical header files, which partly define the ABI. In a sane world such things should not be copyrightable, since they just define a technical interface, like the mapping of errno numbers to constants like EDOM etc. and there is really only one way to write these files (modulo some layout and commenting variations) for a given mapping.

The big question in my mind is: is it conceivable that SCO could succesfully claim to own a certain numbering of error codes or signals? (assuming the numbering used by Linux really is the same as in the historical unix - i have no idea if this is true). If so, this could really drag Linux back to square one.

It would also make it rather impossible in general to make free operating systems or emulators that would be binary compatible with some proprietary operating system.

Set against this the fact that clone systems have been historically tolerated. For example, an IBM compatible PC BIOS must implement the same numbering for software interrupts as the original. Did IBM ever succesfully sue on this basis?

Copyrighting error and signal numbers?

Posted Dec 22, 2003 10:08 UTC (Mon) by abhishek (guest, #10716) [Link]

There are no commenting variations either ;-)
http://www.opengroup.org/onlinepubs/007904975/basedefs/errno.h.html

Copyrighting error and signal numbers?

Posted Dec 22, 2003 10:11 UTC (Mon) by ken (subscriber, #625) [Link]

As you yourself point out a large number of projects would be illegal if SCO was right. Every x86 clone ever made and every BIOS and the wine project could just stop as everything they do is reimplementation of an API including header files.

It's simply no way SCO could convince a judge they are right. They can't possibly beleive in this them self and at the same time claim to have even a shred of intelligence.

Copyrighting error and signal numbers?

Posted Dec 22, 2003 10:40 UTC (Mon) by ccchips (guest, #3222) [Link]

They are not trying to convice any judge. They are trying to convince clueless readers of various sycofant press outlets, and just maybe possibly a jury.

Since McBride and friends took over, greed has been their single motivation, to the exclusion of all that is good about the sharing of computer software and technologies. It is truly sad that much of the press can't (or won't) see this.

Copyrighting error and signal numbers?

Posted Dec 22, 2003 10:26 UTC (Mon) by wweber (guest, #11678) [Link]

This proves it is no longer about the hard-won Intellectual Property at all. It is more as if they want to force all of GNU/Linux to rebuilt from the beginning by laying claim to something arbitrary like this.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 9:07 UTC (Mon) by rgmoore (subscriber, #75) [Link]

What's more ridiculous is that their claims are utter nonsense, and for two reasons as I see it. The more obvious one is that these files very easily could be the result of a clean room reimplimentation. Things like error numbers and codes are well documented, so there're really a very limited number of ways that you even could code them. More importantly, their claims under the USL/BSD lawsuit are ridiculous. They're claiming that the copied files were required, under the settlement, to have USL copyrights added. But A) the judge ruled that USL was unlikely to be able to claim copyright because of delay in adding the required notices and B) the Linux developers are not parties to that agreement. Thus the files are likely not to be under copyright anyway and there's no legal requirement for copyright notices.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 8:58 UTC (Mon) by dps (subscriber, #5725) [Link]

I know LWN is in the US but perhaps it is worth mentioning the EU position: APIs are not subject to copyright, period. IANAL but this was stated to me by somebody who should know.

Expressions of an API, for example vendor header files, might be subject to copyright but the API is *definitely* not. Implementations of APIs are subject to copyright (just like the sound of a Harley-Davison motorbike, an area of germany, etc[*]).

Somehow I think SCO would face really serious problem in the EU. IMHO suing
the the fortune 1000 is a bad move, because they have deep pockets and can afford good lawyers. I doubt the IBM "do you want us to search our patents for ones you really have voilated?" tactics work well against the fortune 1000.

[*] I think I heard six similar items, incuding those two. I got the impression these were selected highlights of a much longer list.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 9:32 UTC (Mon) by proski (subscriber, #104) [Link]

I don't think they plan to sue Fortune 1000 companies, but if they scare just 0.1% of them, they could get some money.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 9:36 UTC (Mon) by jwb (subscriber, #15467) [Link]

Eh? Sue 1 company, to get a lot of money?

SCO invokes DMCA, announces earnings

Posted Dec 23, 2003 1:04 UTC (Tue) by zakaelri (guest, #17928) [Link]

And then use the money from the last lawsuit to fund the next one, of course.

errno.h is part of ISO C and Posix

Posted Dec 22, 2003 12:14 UTC (Mon) by vonbrand (subscriber, #4458) [Link]

IOCTLs are idiosyncratic calls. The Linux IOCTL numbers are (supposed to be) defined in a quite convoluted (Linux-native, as far as I know) way to catch mistakes "and programs compiled with old versions" (see include/asm-i386/ioctl.h, dated 1993). This is extremely unlikely to come (directly) from anywhere else. Verbatim copying is total nonsense.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 12:43 UTC (Mon) by vonbrand (subscriber, #4458) [Link]

Right. But they are claiming that it was included in Linux without permission. If so, this is easy to fix: The Caldera licence for ancient Unix (if there are similarities, they probably stop there) is BSD-like, and allows any use (including modification, binary derived works, etc) as long as the copyright notices are kept intact.

OTOH, the first versions of those files probably came from Minix...

IANAL, anyway.

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 15:44 UTC (Mon) by vblum (subscriber, #1151) [Link]

Hm. I may be missing a subtlety here, so I'll ask this:

Heise claims the code in question as published under the GPL, officially downloadable for 2.4.6 from Caldera. Is Heise wrong in claiming that the GPL was used? If so, how could the linux-abi kit have been linked into the kernel?

SCO invokes DMCA, announces earnings

Posted Dec 22, 2003 16:16 UTC (Mon) by ajs318 (guest, #18043) [Link]

Is it possible that two people working independently of one another could write what amounted to the same header file?

Let's examine what is in a header file. It is a list of the names of the variables, constants and functions used in the main programme source code. The names of these entities are chosen at the whim of the programmer, but good practice teaches that names should be short and meaningful; this is done for the benefit of future developers. Two programmers who learned from similar texts might develop a particular style. Of course, some names might be influenced by specific extant personality traits. If psychometric testing of the two programmers revealed significant differences that might influence choice of variable names, then we might be surer that the two sets of header files are derived from a common ancestor.

Then there is the question of ordering. Convention dictates that the declarations be split into groups by type, and ordered either alphabetically or as they appear in the programme proper. If both programmers have used alphabetical ordering then this is not an issue, but given some automatic latitude by virtue of mathematical equivalency in the order of statements, it would be a greater coincidence for two programmers to code their functions, make use of constants and initialise their variables in the same order.

Of course, if the code in question is trivial, the circumstances are such that names are effectively prescribed {for instance, "port_addr" might well be a popular choice for a variable specifying a port address; and "init_printer()" is a logical choice of a name for a function to initialise a printer}, and/or it acts in certain very specific ways that depend on an exact sequence of instructions {thereby negating the "automatic latitude" proposed above} then there is more likelihood that two programmers would produce identical code when working independently to perform the same function.

The greatest degree of freedom comes in the addition of comments to code. An individual will inevitably develop a particular style in writing comments. Even this might be influenced by mood. Anything beyond the mere explanatory {for instance, "Jeff was here 9T6!"} is as individual as a fingerprint. However, coding convention in certain environments {a corporation keen to project a "professional" image, for instance; or a fluid international collaborative effort} might demand that comments are confined to the minimum necessary to explain the functionality of the code, with correspondingly less scope for personal expression.

In the light of the fact that we are talking about short programmes to accomplish specific simple tasks, in an environment which necessarily minimises an individual's latitude to leave a personal stamp on their code, it becomes more likely that two independent developers could indeed produce identical code. Alternatively, the existence of ancestral code on which Linux could legally have been based would cast reasonable doubt on SCO's assertions that portions of Linux were copied from SCO Unix.