Such a move wouldn't be very smart, even if it was technically possible. SCO could easily argue that "those evil linux people removed it because they knew it was infringing code".
Wrong.
Possibly infringing code is removed immediately, and remains out of the kernel until the legal status is clear.
For example ate_alloc got removed: it was written by AT&T and was accidentially submitted by SGI. Probably it's legal, because the copyright on ate_alloc expired, but the legal status is open, thus it was removed. There are further examples, you'd have to search through the linux-kernel mailing list archives.
But in this case, the legal status is clear: The code is written by IBM, IBM submitted it an placed it under GPL. No need to remove it.
If Kazaa loses, wouldn't it be funny if this case establishes that the terms of EULAs are worthless and unenforceable?
There are two types of EULAs:
You pay money, buy a box, and then inside is an eula. There is a good chance that such EULAs are unenforcable: It's a restriction that is placed on the buyer after the actual sale. The courts that accepted them did it only on special conditions, usually money back guarantee.
You download something from the internet, and then an EULA appears. No prior sale, thus you cannot argue that the licence agreement is invalid because it's not part of the sales agreement. I'm fairly certain that these restrictions are enforcable.
It's already a zombie, and zombies don't die quickly: Santa Cruz Operations left the Unix buisness years ago, probably because they noticed that they couldn't compete with BSD or Linux. What happened is that a group of lawyers got the ~30k contracts with ~6k companies. Now they try to squeeze money out of the rotting corpse.
Expect them to stay around for quite a long time.
What's the moderation for "Wrong"? A quick check on the wayback machine shows that the name was used at least since August 2002, fo0bar corrected his initial statement, but the correction has score 2 and the wrong statement score 5.
Yup. And they build a storage server without ECC memory. We had that - by chance we noticed that a memory chip generated single bit corruptions. We were busy for days comparing every file with older backups. And that was around 50 GB data, not 1 TB.
A password protected area is not a GPL violation: The GPL requests that you make the source available for everyone who got a binary from you, and that you give him all rights that are granted by the GPL. It's not necessary that you set up a public server.
The interesting question is the content of the SRPMS: Did SCO remove the COPYING file, or add a second file that tries to limit the rights? Such a sublicensing would be a GPL violation.
Anyone around who bought SCO Linux and could check what's in the SRPMS? My fear is that SCO has 0 customers, and thus they aren't obliged to distribute the sources.
If SGI put SCO code into Linux, and then individuals with no connection to SGI used that code, [snip]
You can stop now. The intersection of individuals with no connection to SGI that used ate_alloc is empty: I heard that the only systems that used the offending lines were beta hardware that was never shipped to customers. Additionally, certain parts were never working, because ate_alloc assumed 1-based numbers and other parts of the implementation used 0-based numbers...
I won't lie and say I've completely kept up with this SCO thing, but I was always under the impression it was a handful of lines of code in question - like maybe IBM put them there by accident.
There are two independent things: - IBM contributed a lot to the Linux kernel. Thousands of lines of new code, among them with RCU: a new locking strategy that affects indirectly even more code, or the ports to the s390 and ppc64 architectures. SCO claims that the contracts (around 400, according to paragraph 66 of the complaint) between SCO and IBM do not allow IBM to make these contributions. I have no idea about the outcome of the lawsuit, but IMHO the managers that signed 400 contracts, supplements and amendments in ~20 years should be fired - one every 20 days! The behavior of IBM is intentional - IBM interprets the contracts diffrently than SCO. SCO asks for $3 billion. The lawsuit is a contract issue, i.e. restricted to SCO and IBM.
- SCO claims that there is a legal liability of end users that install & use Linux. It's not really clear on which legal theory this is based. It's usually assumed that SCO talks about copyright. A comparison between Linux and older Unix source code contained in a printed book revealed ~80 lines which are removed now. The code was used for beta hardware that was never shipped, i.e. not a critical component. It's not impossible that there are further blocks in the Linux kernel.
SCO usually uses clever word games to mix both cases: One line that I read several times is something like "millions of lines owned or controlled by SCO". This creates the impression that it's impossible to remove the code owned by SCO, i.e. SCO would get the $699/cpu. But in reality it's probably 10 lines owned by SCO, and millions of lines are indirectly affected by code provided by IBM. Btw, I don't think millions of lines affected by IBM is unreasonable: for example IBM contributed ports to two architectures - whole kernel now runs on these archs, and the whole kernel is millions of lines long
By the 22nd of next month, SCO will have to release to IBM the offending code as part of the discovery phase. The question is how fast it will be leaked.
The offending code for the SCO-IBM case is known, just read the current complaint: SCO claims that all IBM contributions that are based on ideas that are also in AIX or dynix/ptx are violating the contracts between SCO and IBM. IBM disagrees. I can imagine three outcomes of that lawsuit: - SCO looses big time. - SCO wins in a few points, IBM must pay a fine. - SCO wins big time, and the IBM contracts/license grants that put IBM's contributions to Linux under the GPL are declared void because IBM was not authorized to contribute to the Linux kernel.
Only the third option would endanger Linux. I don't know if this can happen - if a Microsoft employee adds GPL tags to Windows source and puts it on www.microsoft.com that wouldn't put Windows under GPL, thus it's theoretically possible.
SCO's claims against IP infringement in the Linux kernel are independant from the SCO-IBM lawsuit. Two code snippets were released so far: One snipped shows code copyrighted by BSD [licence granted by Berkeley at the end of the BSD-USL lawsuit, but embarrassing that SCO didn't notice that], one 30 year old snippet of 80 lines in SGI's support code, now removed. It's unlikely that there are any big part from SCO Unix in Linux - the kernels just differ too much.
Perhaps SCO found another snippet which they could use to escape from the mail fraud investigations.
Yes, this does not mean that they see that they are lying. Only that they did not comply with a court order to quit lying;-)
Not to quit lying, the court order is more specific: The court order is against claiming that Linux contains SCO intellectual property without presenting any evidence. SCO Germany now complies with that: www.sco.de doesn't mention IP concerns with Linux at all. They were fined 10000 Euros, because they overlooked a few files, which they removed a few days later. Only 10k Euros, because it was clearly an accident - they overlooked files accessible through https. SCO Germany is free to turn around and claim IP infringement if (and only if) they include evidence.
To sum up, this code is in 2.4.x but not 2.5.x, was also present in BSD which means its open source based
Assuming red means Linux identical to Unix, then the mutex_spinlock() proves that this is not open source: - The ancient BSD codebase didn't contain any spinlocks. - The Linux spinlock function is called "spin_lock()". - FreeBSD 5.0 contains spinlocks, but they are called mtx_lock or something like that.
I.e. _if_ it's from SCO Unix and not from AIX or IRIX, then the code in Linux really violates SCO's copyright. If they sue Linux distributors, then they have a good chance of receiving a few dollars compensation.
Copyright is about expression, not about ideas. The idea of a map allocator is old and predates SCO, but this expression could be owned by SCO.
These devices work because superconductors will no longer superconduct when the current density exceeds a certain threshold- when this happens, the superconductor becomes a regular conductor and is subject to I^2R heating.
Not really correct. The article mentions that it will be a copper oxide based superconductor - a ceramic. When thresholds are reached, then the superconductor becomes an isolator. No danger of arcing, because there is no air that could be ionized. The threshold that they want to use is not the current or heat theshold, they want to use the magnetic threshold: Strong electromagnets are placed around the superconductor, and if the magnets are being turned on, then the conductivity falls to 0.
What I don't understand is why SCO is so unwilling to show the code this is all about. If it's in the kernel everyone can already see it so why the secrecy and complicated NDA stuff?
Because it's not SCO's code: SCO Unix doesn't contain any interesting advanced features.
I'd bet that the code fragments available under NDA actually compare Linux and AIX, or Linux and IBM written code fragments in Montery.
SCO doesn't have the right to show code written by IBM to the public, thus they can't show it to anyone without an NDA.
Since the country's comm. systems were already pretty lacking, and since a presumably fair amount of said systems we're damaged/destroyed, this provides Iraq with a golden opportunity to have a rebuilt, ultra modern communications system.
Like it happened in Eastern Germany? Deutsche Telekom used fibre wires everywhere. Now they can't implement DSL, because the copper-to-optical converters are not in a central place, but burried in every street/house. And optical DSL-like solutions are too expensive, because the quantities are small.
If you can use a 2.2 kernel, you can go on with out much chance of trouble.
They still distribute 2.4.13: ftp://ftp.sco.com/pub/updates/OpenLinux/3 .1.1/Work station/current/SRPMS/linux-2.4.13-21D.src.rpm
At least they did it at Mon Jul 21 21:18:28 UTC 2003. With a COPYING file that says that "The act of running the Program is not restricted".
Are there any SCO Linux 4 customers around? Could they download the sources for the 2.4.19 based UL kernel from their servers [customer only, password protected] and put it to a sourceforge project page? I can only find the UL specific patches on the public FTP site, not the full kernel tarball.
They can either enforce their copyright, assuming that the kernel contains SCO code improperly, and stop Linux distribution, or they can sit back and do nothing. Collecting money is not a legal option.
There is a third option:
They could stop the Linux distribution and ask for money.
Since they are not distribution the source, they are not bound by the GPL. And thus they can ask for money, without infringing the copyright of the Linux kernel authors. One example is lame: the source is GPL, but you may need a patent license from Thomson. No GPL violation, because Thomson doesn't distribute lame.
What SCO cannot do is to replace the COPYING in the kernel sources with a proprietary license, or to distribute a linux kernel binary without sources.
The argument contained within this table leans heavily on the "continued distribution of GPL code after realization of the consequences thereof".
The whole block is bogus: SCO makes two independant claims: that IBM violates their contract with SCO, and that Linux contains source code from the SCO Unix code base without authorization. They began with their lawsuit against IBM on March 6th, and at that time SCO board members said that this is about SCO-IBM, not about Linux. The claims that the kernel contains copyright violations came later, I don't remember the exact date. Additionally, SCO distributes a version of the 2.4 kernel, and that it's possible that the infringing code is only in 2.5.
_If_ there is a chance to claim that SCO placed their IP under the GPL by knowingly distributing, then it is the support program for their current customers: SCO said that they will continue to support them, and on the Usenix conference session about SCO-Linux one of the speakers mentioned that the source is still avaliable from their support download site.
Are there any SCO Linux customers around? Can anyone download the kernel sources and check if COPYING is still inside?
Another potential use for nanotubes are the traces on semiconductors: I've seen a presentation from Infineon about using carbon nanotubes instead of copper for the vias in copper - time frame for production 3-5 years.
The main part of the GPL is valid in Germany: The rule that any derived work must be placed under GPL. [D II 2. e (2) (a), page 47]. It's safe to assume that this part works worldwide: the right to create a derived work is an exclusive right of the copyright holder, and he can grant that right only if arbitrary conditions are met.
One problem is the no-warrenty clause - such clauses are invalid under some circumstances, for example when the loss if life was caused. I've seen open source software from US companies with an explicit line that use for medical application or life support is not permitted - perhaps there are similar dangers in the US.
Everything else is legalese - which type of contract is the GPL, who are the parties in the contract [does the user have a contract with all authors together, or with each author individually, i.e. thousands of seperate contracts, etc.]? Is an English contract enforcable?
One interesting point is that if someone violates the GPL, then it might be difficult to sue for damages: It may be necessary to name all coauthors for such a lawsuit. But since an individual author can ask for a restraining order, which is sufficient to enforce the GPL, this is not a critical problem.
So if there is violating code, SCO has no right to "bless" redistribution unless they agree to non-GPL licensing terms with EVERY SINGLE KERNEL DEVELOPER who contributed non-offending source.
I think blessing is possible: Party A distributes a GPL program and mentions that you need an additional license to actually run it. Party B sells the license. One example is LAME - souce is LGPL, but you need a patent license from Fraunhofer/Thompson to use it. SCO (==party B) could request a license fee from customers of Redhat (==party A) for running Linux. SCO would not violate the GPL in doing so - they are not distributing the Linux kernel, thus they are not bound by the GPL.
The block allocator that is part of ext2. GPL source, in the non-free directory of FreeBSD.
Hmm. Has anyone tried to disassemble the Linux Kernel Personality of Unixware? It supports ext2, and AFAIK there is no free (as in beer) ext2 implementation.
According to this website this comment is in the Sixth Edition Unix source:
*
* You are not expected to understand this.
*/
And in linux-1.2.13/mm/swap.c:
/*
* Buddy system. Hairy. You really aren't expected to understand this
*/
The same sentence in Sixth Edition Unix and the Linux source code - an obvious case of malicious infringement of a valuable trade secret owned by SCO in the Linux kernel. The only change is "You really aren't" instead of "You are not" - additionally evidence this is willful infringement: The infringer was aware of what he was doing and did a minor change, to hide the evil theft.
That comes out to 60-100 million dollars per line of code.
And this explains why they cannot disclose where the 50 odd lines are. They would be rewritten overnight, and the 1 billion $ lawsuit would collapse to a 1000 $ lawsuit. Even a non-tech savy judge would understand that something that was replaced overnight can't be worth 1 billion $.
Slashdot Mirror
Wrong.
Possibly infringing code is removed immediately, and remains out of the kernel until the legal status is clear.
For example ate_alloc got removed: it was written by AT&T and was accidentially submitted by SGI. Probably it's legal, because the copyright on ate_alloc expired, but the legal status is open, thus it was removed. There are further examples, you'd have to search through the linux-kernel mailing list archives.
But in this case, the legal status is clear: The code is written by IBM, IBM submitted it an placed it under GPL. No need to remove it.
There are two types of EULAs:
Santa Cruz Operations left the Unix buisness years ago, probably because they noticed that they couldn't compete with BSD or Linux. What happened is that a group of lawyers got the ~30k contracts with ~6k companies. Now they try to squeeze money out of the rotting corpse.
Expect them to stay around for quite a long time.
What's the moderation for "Wrong"?
A quick check on the wayback machine shows that the name was used at least since August 2002, fo0bar corrected his initial statement, but the correction has score 2 and the wrong statement score 5.
Yup.
And they build a storage server without ECC memory. We had that - by chance we noticed that a memory chip generated single bit corruptions. We were busy for days comparing every file with older backups. And that was around 50 GB data, not 1 TB.
A password protected area is not a GPL violation: The GPL requests that you make the source available for everyone who got a binary from you, and that you give him all rights that are granted by the GPL.
It's not necessary that you set up a public server.
The interesting question is the content of the SRPMS: Did SCO remove the COPYING file, or add a second file that tries to limit the rights? Such a sublicensing would be a GPL violation.
Anyone around who bought SCO Linux and could check what's in the SRPMS?
My fear is that SCO has 0 customers, and thus they aren't obliged to distribute the sources.
You can stop now. The intersection of individuals with no connection to SGI that used ate_alloc is empty: I heard that the only systems that used the offending lines were beta hardware that was never shipped to customers. Additionally, certain parts were never working, because ate_alloc assumed 1-based numbers and other parts of the implementation used 0-based numbers...
There are two independent things:
- IBM contributed a lot to the Linux kernel. Thousands of lines of new code, among them with RCU: a new locking strategy that affects indirectly even more code, or the ports to the s390 and ppc64 architectures. SCO claims that the contracts (around 400, according to paragraph 66 of the complaint) between SCO and IBM do not allow IBM to make these contributions. I have no idea about the outcome of the lawsuit, but IMHO the managers that signed 400 contracts, supplements and amendments in ~20 years should be fired - one every 20 days! The behavior of IBM is intentional - IBM interprets the contracts diffrently than SCO. SCO asks for $3 billion. The lawsuit is a contract issue, i.e. restricted to SCO and IBM.
- SCO claims that there is a legal liability of end users that install & use Linux. It's not really clear on which legal theory this is based. It's usually assumed that SCO talks about copyright. A comparison between Linux and older Unix source code contained in a printed book revealed ~80 lines which are removed now. The code was used for beta hardware that was never shipped, i.e. not a critical component. It's not impossible that there are further blocks in the Linux kernel.
SCO usually uses clever word games to mix both cases: One line that I read several times is something like "millions of lines owned or controlled by SCO". This creates the impression that it's impossible to remove the code owned by SCO, i.e. SCO would get the $699/cpu. But in reality it's probably 10 lines owned by SCO, and millions of lines are indirectly affected by code provided by IBM.
Btw, I don't think millions of lines affected by IBM is unreasonable: for example IBM contributed ports to two architectures - whole kernel now runs on these archs, and the whole kernel is millions of lines long
The offending code for the SCO-IBM case is known, just read the current complaint: SCO claims that all IBM contributions that are based on ideas that are also in AIX or dynix/ptx are violating the contracts between SCO and IBM.
IBM disagrees.
I can imagine three outcomes of that lawsuit:
- SCO looses big time.
- SCO wins in a few points, IBM must pay a fine.
- SCO wins big time, and the IBM contracts/license grants that put IBM's contributions to Linux under the GPL are declared void because IBM was not authorized to contribute to the Linux kernel.
Only the third option would endanger Linux. I don't know if this can happen - if a Microsoft employee adds GPL tags to Windows source and puts it on www.microsoft.com that wouldn't put Windows under GPL, thus it's theoretically possible.
SCO's claims against IP infringement in the Linux kernel are independant from the SCO-IBM lawsuit. Two code snippets were released so far: One snipped shows code copyrighted by BSD [licence granted by Berkeley at the end of the BSD-USL lawsuit, but embarrassing that SCO didn't notice that], one 30 year old snippet of 80 lines in SGI's support code, now removed.
It's unlikely that there are any big part from SCO Unix in Linux - the kernels just differ too much.
Perhaps SCO found another snippet which they could use to escape from the mail fraud investigations.
Not to quit lying, the court order is more specific:
The court order is against claiming that Linux contains SCO intellectual property without presenting any evidence. SCO Germany now complies with that: www.sco.de doesn't mention IP concerns with Linux at all. They were fined 10000 Euros, because they overlooked a few files, which they removed a few days later. Only 10k Euros, because it was clearly an accident - they overlooked files accessible through https.
SCO Germany is free to turn around and claim IP infringement if (and only if) they include evidence.
Assuming red means Linux identical to Unix, then the mutex_spinlock() proves that this is not open source:
- The ancient BSD codebase didn't contain any spinlocks.
- The Linux spinlock function is called "spin_lock()".
- FreeBSD 5.0 contains spinlocks, but they are called mtx_lock or something like that.
I.e. _if_ it's from SCO Unix and not from AIX or IRIX, then the code in Linux really violates SCO's copyright. If they sue Linux distributors, then they have a good chance of receiving a few dollars compensation.
Copyright is about expression, not about ideas. The idea of a map allocator is old and predates SCO, but this expression could be owned by SCO.
Not really correct.
The article mentions that it will be a copper oxide based superconductor - a ceramic. When thresholds are reached, then the superconductor becomes an isolator. No danger of arcing, because there is no air that could be ionized.
The threshold that they want to use is not the current or heat theshold, they want to use the magnetic threshold: Strong electromagnets are placed around the superconductor, and if the magnets are being turned on, then the conductivity falls to 0.
Because it's not SCO's code: SCO Unix doesn't contain any interesting advanced features.
I'd bet that the code fragments available under NDA actually compare Linux and AIX, or Linux and IBM written code fragments in Montery.
SCO doesn't have the right to show code written by IBM to the public, thus they can't show it to anyone without an NDA.
Like it happened in Eastern Germany?
Deutsche Telekom used fibre wires everywhere. Now they can't implement DSL, because the copper-to-optical converters are not in a central place, but burried in every street/house. And optical DSL-like solutions are too expensive, because the quantities are small.
Yes, it was fixed: AGP GART pages are removed from the linear mapping with change_page_attr.
They still distribute 2.4.13:
ftp://ftp.sco.com/pub/updates/OpenLinux/
At least they did it at Mon Jul 21 21:18:28 UTC 2003. With a COPYING file that says that "The act of running the Program is not restricted".
Are there any SCO Linux 4 customers around? Could they download the sources for the 2.4.19 based UL kernel from their servers [customer only, password protected] and put it to a sourceforge project page? I can only find the UL specific patches on the public FTP site, not the full kernel tarball.
There is a third option:
They could stop the Linux distribution and ask for money.
Since they are not distribution the source, they are not bound by the GPL. And thus they can ask for money, without infringing the copyright of the Linux kernel authors. One example is lame: the source is GPL, but you may need a patent license from Thomson. No GPL violation, because Thomson doesn't distribute lame.
What SCO cannot do is to replace the COPYING in the kernel sources with a proprietary license, or to distribute a linux kernel binary without sources.
The whole block is bogus:
SCO makes two independant claims: that IBM violates their contract with SCO, and that Linux contains source code from the SCO Unix code base without authorization. They began with their lawsuit against IBM on March 6th, and at that time SCO board members said that this is about SCO-IBM, not about Linux.
The claims that the kernel contains copyright violations came later, I don't remember the exact date.
Additionally, SCO distributes a version of the 2.4 kernel, and that it's possible that the infringing code is only in 2.5.
_If_ there is a chance to claim that SCO placed their IP under the GPL by knowingly distributing, then it is the support program for their current customers: SCO said that they will continue to support them, and on the Usenix conference session about SCO-Linux one of the speakers mentioned that the source is still avaliable from their support download site.
Are there any SCO Linux customers around? Can anyone download the kernel sources and check if COPYING is still inside?
Another potential use for nanotubes are the traces on semiconductors:
I've seen a presentation from Infineon about using carbon nanotubes instead of copper for the vias in copper - time frame for production 3-5 years.
http://www.eurosime.com/bgnd.htm#es03
The main part of the GPL is valid in Germany: The rule that any derived work must be placed under GPL. [D II 2. e (2) (a), page 47].
It's safe to assume that this part works worldwide: the right to create a derived work is an exclusive right of the copyright holder, and he can grant that right only if arbitrary conditions are met.
One problem is the no-warrenty clause - such clauses are invalid under some circumstances, for example when the loss if life was caused. I've seen open source software from US companies with an explicit line that use for medical application or life support is not permitted - perhaps there are similar dangers in the US.
Everything else is legalese - which type of contract is the GPL, who are the parties in the contract [does the user have a contract with all authors together, or with each author individually, i.e. thousands of seperate contracts, etc.]? Is an English contract enforcable?
One interesting point is that if someone violates the GPL, then it might be difficult to sue for damages: It may be necessary to name all coauthors for such a lawsuit. But since an individual author can ask for a restraining order, which is sufficient to enforce the GPL, this is not a critical problem.
I think blessing is possible:
Party A distributes a GPL program and mentions that you need an additional license to actually run it. Party B sells the license.
One example is LAME - souce is LGPL, but you need a patent license from Fraunhofer/Thompson to use it.
SCO (==party B) could request a license fee from customers of Redhat (==party A) for running Linux. SCO would not violate the GPL in doing so - they are not distributing the Linux kernel, thus they are not bound by the GPL.
linux/fs/ext2/balloc.c
The block allocator that is part of ext2. GPL source, in the non-free directory of FreeBSD.
Hmm. Has anyone tried to disassemble the Linux Kernel Personality of Unixware? It supports ext2, and AFAIK there is no free (as in beer) ext2 implementation.
And in linux-1.2.13/mm/swap.c:
The same sentence in Sixth Edition Unix and the Linux source code - an obvious case of malicious infringement of a valuable trade secret owned by SCO in the Linux kernel. The only change is "You really aren't" instead of "You are not" - additionally evidence this is willful infringement: The infringer was aware of what he was doing and did a minor change, to hide the evil theft.
And this explains why they cannot disclose where the 50 odd lines are. They would be rewritten overnight, and the 1 billion $ lawsuit would collapse to a 1000 $ lawsuit.
Even a non-tech savy judge would understand that something that was replaced overnight can't be worth 1 billion $.
Just FYI - there are admins that blacklist whole countries as a spam filter - there was a lkml posting about that.