GPL Gets Its Day in Court in Israel

MadFarmAnimalz writes "In what appears to be the first court test for the GPL in the Middle East, Alexander Maryanovsky, the author of the GPL licensed Jin Chess Client is taking IchessU to court for violations of the GPL license."

Comments

Right.... bit of clarification

[LiquidCoooled]

There is an open source chess client called JIN licensed under the GPL.
This is an executable and front end chess client.

This has been extended by iChessU to support a closed source DLL which adds new functionality (notably video streaming between players).

The source code to the Expanded client is available and providing you have the closed source binary DLL, you can run the newly compiled program.

Isn't this like me releasing a GPL program which is linked to the nvidia or ATI blobs?

Hell, isn't it similar if I write a GPL application which uses the Windows API?

I personally feel as though this is an overreaction, the ichessu site does not hide the fact its based on JIN and offers sources, or am I wrong and this is infact a genuine GPL violation?

Re:Right.... bit of clarification

[oliverthered]

If they ship the dll with the application then it may be a GPL violation if it falls out of the mere aggregation clause. When you write a GPL application against the windows API you don't usually ship the application with the windows API, and there has been a case of someone shipping a distro with the binary ATI + NVidia drivers being pulled up by the FSF for violation of the GPL.

Personally I don't see how the GPL can be violated here and yet we still have PC's shipping with GPL software on them.

Re:Right.... bit of clarification

[TERdON]

GNU about the GPL: "This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Lesser General Public License instead of this License."

And in the license itself: "For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable."

IMHO, that means Windows API = ok, Java API = ok, .NET API = ok, own API or library = not okay.

And that makes sense. Otherwise I could just build all of my app in my "MainApp API" and GPL my "StartMainApp()" function call... :)

Re:Right.... bit of clarification

[bloodredsun]

Nope.

Only part of the client source code is downloadable, not the whole; this is a violation of the GPL. Also, the iChessU has an EULA which violates the GPL by placing new restrictions on how the code may be used.

The bad faith negotations accusation may be an overreaction but it's hard not to think this when iChessU initially wanted to license JIN but backed out when it appeared too expensive for them and then proceeded to use it anyway.

This is not an overreaction but a devloper fighting to prevent a third party assuming legal control of that developers work.

Re:Right.... bit of clarification

[Anonymous Coward]

From the article, it says that they are accessing GPL code through sockets and open sourced the wrapper they wrote in order to do this. The proprietary DLL obviously works on top of JIN and interacts with it over sockets. The JIN author incorrectly believes that this violates the GPL. In fact, since the server code is never distributed, it can be modified in house without violating the license. If developed correctly, iChessU could easily leverage JIN, stay closed-source, and be fully complient.

Re:Right.... bit of clarification

[Anonymous Coward]

Mind you, it would help if the source was online, http://www.ichessu.com/Files/ichessu_sources.zip gives a 404

Re:Right.... bit of clarification

[bestinshow]

The method of adding functionality to a GPL application shouldn't be the issue. Clearly the client application distributed to end users is comprised of GPL software (JIN) and proprietary extensions, and they communicate with each other.

Clearly the fact that they asked to license the software and were refused (or offered a reasonable offer which they turned down) shows they were aware that it was GPL and wanted to license it under different terms. The fact they they went ahead and tried to bypass the license but still use the code shows extreme bad faith.

This may be an issue with the GPL wording though, which leaves this loophole that can be exploited by proprietary developers. Turn the GPL software you want into a server, release the source to the modifications made here, then keep all your own proprietary stuff in the client that communicates with the GPL server, then ship both together to run on the user's system. Clearly the GPL needs some effort made to outlaw 'serverising' GPL code in the same way that dynamically linking or statically linking code is outlawed. Clearly the fact that someone picks the GPL over the LGPL is because they don't want someone using their code with non-Open Source code additions.

Of course you need to allow the situation where the server functionality is explicitly added to the code base by the developer(s), e.g., something like Apache should not be restricted from serving to IE, for example (I know Apache isn't GPL, but you know what I mean). I think that this can be allowed safely, clearly the software is a server by definition, whereas a chess client is not a server by definition, and also the client and 'server' are distributed together, require each other, and run at the same time.

Re:Right.... bit of clarification

[jrumney]

Isn't this like me releasing ... isn't it similar if I write ...

No, because in those cases you are the original author, so you can license the code how you wish. The GPL includes a specific exception for OS libraries that covers the second of these already, BTW.

Re:Right.... bit of clarification

[bWareiWare.co.uk]

The nVidia blobs and kernel stub are not GPL. The GPL Linux kernel contains no code for accessing the NVIDIA blobs.

The process of installing the nVidia Kernel stub combines GPL and non-GPL code and compiles it on your machine. The resulting binary is NOT re-distributable under any licence. This is why Linux distributions do not come with the nVidia and ATI drivers built in, but you must install them separately.

This workaround works because the GPL only comes into effect when you copy a program not when you use it, as long as you are not copying (i.e. redistributing) the results you are not bound by its terms.

iChessU could use the same trick. Download the standard JIN source, download the iChessU patch and binary and compile them yourself - noone is copying the result, so the GPL is not violated. Though the resulting program contains GPL and non-GPL code and so can never be copied under any licence.

Re:Right.... bit of clarification

[TheRaven64]

The question is slightly different when it comes to plug-ins, however.

What happens if you release a GPL'd program with a plug-in API? That's fine.

Now you release a closed-source plug-in. That's not fine, since the plug-in is a derived work of the GPL'd plug-in system; it won't work without it.

Now, what happens if you license the plug-in API as a separate module under the LGPL (for example). The closed-source plug-in only depends on the plug-in module, which is LGPL'd. It inherits the GPL when used with a GPL'd application, but you could argue that the plug-in does not require a GPL'd program.

Now, when you use the plug-in, you link it with a GPL'd application. This would be a GPL violation, except for the fact that the GPL only comes into effect when distributing (it is not a EULA).

Re:Right.... bit of clarification

[jrumney]

This may be an issue with the GPL wording though, which leaves this loophole that can be exploited by proprietary developers.

Given that only the GPL gives them the right to exploit this loophole, and that a copyright holder can choose to license their software to whomever they want, can the author just terminate their right to license the code under the GPL? It's not like they've paid for it, so the author is not under any contractual obligation to allow them to exploit loopholes in the license.

Re:Right.... bit of clarification

[Spikeles]

The Jin Chess Client code was taken and changed so that it loaded a DLL made by IChessU. Now, loading DLL's dynamically from a GPL'd program does not violate GPL if the DLL is not statically linked, if all the entry points are accessed DYNAMICALLY. They released the source to this extended version, but not the DLL ( which is their own code not under GPL ).

Where is the problem? What violation has occurred? Or am i mis-understanding something?

Re:Right.... bit of clarification

[jrumney]

You're misunderstanding the terms of the GPL as they relate to various linking technologies. There are many armchair lawyers on slashdot who like to claim that the GPL only applies to static linking, or that using TCP sockets gets around it etc. But the GPL does not contain any mention of linking technologies or what is and isn't covered, it just talks about "derived works", which is up to the courts to define. In a case where the defendant approached the plaintif about licensing their work commercially then suddenly changed their mind and wrote some dynamic linking or socket based code specifically to "get around" the GPL, I would expect the court to side with the plaintif, since the defendant has shown that they understood from the start that what they wanted to do was not allowed under the GPL, and their intention is plainly to try to circumvent copyright law through technicalities, which the judge is unlikely to approve of.

Re:Right.... bit of clarification

[Decameron81]

If someone released a GPL program in one download, and the documentation instructed that you also need to download a DLL separately for the program to run, would the DLL be subject to the GPL?

Re:Right.... bit of clarification

[CortoMaltese]

GPL FAQ on plug-ins:

If a program released under the GPL uses plug-ins, what are the requirements for the licenses of a plug-in?

It depends on how the program invokes its plug-ins. If the program uses fork and exec to invoke plug-ins, then the plug-ins are separate programs, so the license for the main program makes no requirements for them.

If the program dynamically links plug-ins, and they make function calls to each other and share data structures, we believe they form a single program, which must be treated as an extension of both the main program and the plug-ins. This means the plug-ins must be released under the GPL or a GPL-compatible free software license, and that the terms of the GPL must be followed when those plug-ins are distributed.

If the program dynamically links plug-ins, but the communication between them is limited to invoking the `main' function of the plug-in with some options and waiting for it to return, that is a borderline case.

In this case, the "A/V plug-in" is a closed source DLL, which is in violation of the GPL.

Re:Right.... bit of clarification

[SashaM]

I personally feel as though this is an overreaction, the ichessu site does not hide the fact its based on JIN and offers sources, or am I wrong and this is infact a genuine GPL violation?

Actually, there are several violations:

1. IChessU's released source code does not compile, so it can't possibly be considered the "complete corresponding source code", regardless of the status of the audio/video library (which to the best of my understanding of the GPL is also covered by it).
2. IChessU do not release their application under the GPL, but instead under a horrendous EULA. Read it, really.
3. Although they do mention Jin, they don't mention my copyright. Instead it's "Copyright (C) [2006] [unknown]", as if they don't know whose copyright Jin is.

Alexander (aka Sasha) Maryanovsky.

Re:Right.... bit of clarification

[mdwh2]

Is the latest version (or any version) of Java released with Windows and other OSes these days? I certainly remember it was common to have to download it with a browser.

Another example would be various GUI toolkits, including using IDEs like Borland Builder. Or any 3rd party library that someone might need, e.g., for example an image loading library.

I can see the reasoning for doing so, as you say, but this seems a bit of a hacky way round it. This is a particular problem for obscure non-mainstream OSs where a lot of useful additions have had to be added via 3rd party libraries. And how does it apply to operating systems themselves - couldn't Microsoft get away with keeping parts closed sourced by putting them in separate libraries, as those would be shipped with the OS?

Re:Right.... bit of clarification

[mrchaotica]

The DLL would be subject to the GPL if and only if the DLL is only useful with the GPL'd program. In other words, linking to "MyP2PLib.dll" that you really just wrote to work with your "FooP2P" application would not be okay, but linking to "msvc70.dll" (a standard Windows library) would be OK.

In technical legal terms, the test is whether the library is a "derived work" of the GPL'd program. Microsoft's dll would not ben a derived work because it wasn't designed specifically to work with FooP2P, but MyP2PLib.dll would be.

Of course, if you're the original author of a program, you're free to give it a "sort-of-like-GPL-but-not-really" license whereby you could say "This program is licensed under the same terms as the GPL except that it's allowed to link to the "MyP2PLib" proprietary library." However, you would no longer be legally allowed to claim that the result was licensed under "The GPL," because the FSF trademarked the name of the license to prevent exactly that scenario. : )

Re:Right.... bit of clarification

[msergeo1]

So, what is the difference between Microsoft'a APIs, SUN's APIs and my own API? The GPL does not say about that neither, and if it is not forbidden, its allowed! How do you distinguish which API is OK and which is not? Adobe API? By the size of the company?

Re:Right.... bit of clarification

[Spikeles]

It seems to me that better wording is required in GPL. After reading your post i did what i should have done in the first place. RTFM

http://www.gnu.org/licenses/gpl-faq.html#TOCMereAg gregation

Re:Right.... bit of clarification

[Targon]

I could be wrong, but the difference is in the distribution rights, not in the end result.

If you are required to download the ATI drivers as a seperate piece, but have the code in place to allow for third party drivers, then there is no problem with the GPL.

ATI doesn't re-distribute anything, they just make a module that can be dropped in. No changes to the software were done by ATI.

So, if IChessU made their stuff as a seperate package without distribution of JIN, that's fine. The problem is that IChessU is taking JIN, making changes, then re-distributing the modified product.

If IChessU made an installer that downloaded the JIN package, patch it, and do so automatically, that might work as long as the proper copyright notices were put in place.

JIN is NOT the work of IChessU, and as a result, IChessU doesn't have the rights to take JIN and distribute a modified version.

A great program known as BartPE works in the way I suggest IChessU should if they wanted to avoid a legal problem. It asks for the original Windows CD, and builds the resulting CD image. For legal purposes, BartPE requires that you have the rights to run WindowsXP(meaning you have a valid product key), but when it comes down to it, BartPE doesn't include any part of the Windows XP code in the distribution.

Another way to look at it is that you are allowed to make any changes you want for your own personal use as long as you don't re-distribute any modified versions without also providing the full source code. That's the short version. If you keep it in house, then it's generally not a problem. The moment you try to sell something that takes source code written by someone else, then you run into problems.

Re:Right.... bit of clarification

[sjwaste]

Actually, there are several violations:

I'm not a lawyer (yet), but I take you're party to this suit. Do you really think you ought to be discussing your claim on a public forum prior to the hearing?

Re:Right.... bit of clarification

[Kjella]

If the program dynamically links plug-ins, and they make function calls to each other and share data structures, we believe they form a single program, which must be treated as an extension of both the main program and the plug-ins. This means the plug-ins must be released under the GPL or a GPL-compatible free software license, and that the terms of the GPL must be followed when those plug-ins are distributed.

The problem is that FSF is trying to pull a SCO here. Let me try to draw it:

GPL'd application
|
v
[distribution - bound by GPL]
|
v
Combined work -> [redistribution - bound by GPL]
^
|
[distribution - bound by GPL?????]
^
|
plug-in

----------------

SCO's code
|
v
[distribution - bound by SCO's license]
|
v
AIX -> [redistribution - bound by SCO's license]
^
|
[distribution - bound by SCO's license?????]
^
|
IBM's code

In short, I don't think they can be distributed or marketed as one (as that goes beyond 'mere aggregation'), but I fail to see how the GPL could possibly bind anyone from providing closed source plug-ins to a GPL'd application.

Re:Right.... bit of clarification

[SashaM]

(according to my lawyer) all the information in a lawsuit in Israel is public knowledge - the text of any filed lawsuit is available to the public. I'm not writing here anything that isn't in the lawsuit, so I think I'm safe.

Alexander (aka Sasha) Maryanovsky.

Re:Right.... bit of clarification

[sjwaste]

Sounds good. I hope you don't think I was giving you a hard time, I wasn't!

Seemed like you were just restating what was in your claim. Hope you win this one, seems like someone is just taking advantage of open source software for their own gain.

Re:Right.... bit of clarification

[Anonymous Coward]

In this particular case I believe the term is pwned.

Re:Right.... bit of clarification

[SashaM]

There's nothing wrong in taking advantage of open source software for your own gain. I do it every day when I use Linux. But yes, I know what you meant :-)

Alexander (aka Sasha) Maryanovsky.

Re:Right.... bit of clarification

[Courageous]

Now you release a closed-source plug-in. That's not fine, since the plug-in is a derived work of the GPL'd plug-in system; it won't work without it.

What you seem to be saying is that Microsoft owns every ActiveX component ever written. ActiveX is a form of "plugin," and you are saying that plugins are derived works.

I object! And do not agree: a plugin is not a derived work of the plugin system.

Further, a point of order: the GPL does not define what it covers, really. Only the law does. The GPL covers whatever the law believes is a derived work. I don't believe that the law is going to back you up here.

C//

Re:Right.... bit of clarification

[Tim C]

own API or library = not okay

Except of course that if it is your API/library, you own the copyright and are perfectly free to release it under more than one licence. That is, you can release a GPLed version, and a non-GPLed version that you incorporate into similarly closed code.

Re:Right.... bit of clarification

[LiquidCoooled]

Sasha,
I have read most of the replies to my clarification (the slash header didn't make it very clear) and upon reflection have changed my mind; as you point out there does appear to be verious conflicts with the GPL.

I wish you the best of luck with your actions and hope you are successful in cleaning up these violations.

Gary

Re:Right.... bit of clarification

[Anonymous Coward]

Actually, there are several violations:

      1. IChessU's released source code does not compile, so it can't possibly be considered the "complete corresponding source code", regardless of the status of the audio/video library (which to the best of my understanding of the GPL is also covered by it).
      2. IChessU do not release their application under the GPL, but instead under a horrendous EULA. Read it, really.
      3. Although they do mention Jin, they don't mention my copyright. Instead it's "Copyright (C) [2006] [unknown]", as if they don't know whose copyright Jin is.

Alexander (aka Sasha) Maryanovsky.

(NB I am not a lawyer; I am just a computer scientist)

I don't want you to prejudice your case by replying to this yourself, but I wanted to borrow your quote to say I'm not as convinced as many Slashdotters that the GPL is really all that clear-cut. (Which is awkward for me, since I write a fair amount of GPL software!)

Consider point 1. They are required to provide the corresponding source code of the "derived work", but what constitutes a "derived work" and what constitutes "mere aggregation" in GPL terms is a philosophically open question. You seem to feel that a derived work has to compile; I'm not so sure. The programs I write constitute a "work" and I have copyright in them before they compile. For example, an algorithm is certainly a "work". And if I tie programs A and C together with shim B, then B is a derived work of A and C, but it's hard to argue that C is a derived work of A. But is A-B-C a derived work or the "mere aggregation" of A, B, and C in GPL terms? If B is just a communication-shim between A and C, then it perhaps it is (regardless of how many geeks grumble "not if they use static linking" it's not been legally determined what the boundaries are).

And to be frank, technical issue like linking do not seem to help us at all here -- compressing two files into the same archive could techincally create more of a derived work than writing code that uses other code. In the compressed files, there are bytes in the resulting file that depend on bytes in each of the two files -- and yet most people would consider sending a zip file with two files in it "mere aggregation" of the two files (and I can't imagine a judge disagreeing). We end up having to consider "aggregation" and "work" as philosophical or cognitive units, not technical ones. (If you send a GPL program over email using PGP, the communication is a derived work based on both the program and your PGP key - one would hardly argue that a user has a GPL-imbued right to ask you for your private key as part of the "source" of your communication though)

IchessU do provide some modified Jin sources for download, so it looks to me like they feel that constitutes the derived work, and what you download in compiled form is the aggregation of that with their non-derived-works around the video codec.

And to be honest, I personally (as a non-lawyer-comp-sci) reckon they've got a fair argument there. Software to support video communications over the internet doesn't sound to me like it's inherently Jin-derived, regardless of whether it happens to sit next to Jin in a pretty window or whether its class files are aggregated alongside some modified Jin class files.

Re:Right.... bit of clarification

[arose]

Consider point 1. They are required to provide the corresponding source code of the "derived work", but what constitutes a "derived work" and what constitutes "mere aggregation" in GPL terms is a philosophically open question.

It may be a philosophical question in some ways, but the fact that a binary compiled from GPL code is such a derived work should be pretty clear, so let's look at the "corresponding source code" side:

For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable.

As I read that it means that it should compile and if the original author can't compile the code then it sounds suspect to me.

Software to support video communications over the internet doesn't sound to me like it's inherently Jin-derived, regardless of whether it happens to sit next to Jin in a pretty window or whether its class files are aggregated alongside some modified Jin class files.

And to me it sounds that the seperate programs didn't cut it in terms of integration and they were forced to do it in one. The GPL couln't stop them from making a chess program and a video communication program that work really good together, yet they clearly didn't take this route.

Re:Right.... bit of clarification

[petrus4]

If the program needs the non-GPL binary lib in order to work at all, and the binary lib does not have available source, then yes, that is a violation. Anything being GPLed means that source must be available with binaries for the entire application, including libraries.

However, if the binary lib in question is optional, then no, it is no more a violation than people using binary video card drivers with the Linux kernel. Of course, the FSF attack bots who oppose that practice will probably want the defendant in this case to be successfully made an example of as well.

I've said it before, and I'll say it again:- The GPL v2 can be beneficial in *some* contexts. What is *not* beneficial is the autistic fanaticism of most of Stallman's followers.

Re:Right.... bit of clarification

[someone1234]

If you yourself write your application to support a closed source dll, it is definitely not a violation. As you are the one who chose to limit yourself. If someone takes YOUR source, and adds a proprietary plugin to it, then it depends on the license. If you licensed your code to him by GPL (not LGPL), then he breached the license.

Re:Right.... bit of clarification

[Anonymous Coward]

Consider point 1. They are required to provide the corresponding source code of the "derived work", but what constitutes a "derived work" and what constitutes "mere aggregation" in GPL terms is a philosophically open question. You seem to feel that a derived work has to compile; I'm not so sure. The programs I write constitute a "work" and I have copyright in them before they compile. For example, an algorithm is certainly a "work". And if I tie programs A and C together with shim B, then B is a derived work of A and C, but it's hard to argue that C is a derived work of A. But is A-B-C a derived work or the "mere aggregation" of A, B, and C in GPL terms? If B is just a communication-shim between A and C, then it perhaps it is (regardless of how many geeks grumble "not if they use static linking" it's not been legally determined what the boundaries are).

This is an interesting tangent, but we should clarify one thing first. They stripped copyright attribution and said the authorship of the code was unknown. That is clearly not allowed under any interpretation of the GPL. The only question of fact is whether there are additional ways they have violated the copyright.

Now let's go to your tangent.

I wouldn't consider the boundary between a "derived work" and "mere aggregation" to be a philosophical question. It is actually a very precise legal question. Namely, is this work legally derived under copyright law from that one? It isn't a simple one, copyright law isn't simple. But it isn't a question of abstract philosophy either. It is a question of fact of a type that courts deal with every day.

Now consider your A-B-C example. What you say is entirely true when it comes to source code. It may well be that C is not a derivative under copyright of A. In fact it was thinking about this exact situation that lead Linus Torvalds to the publically stated opinion that lead to people linking proprietary drivers against the Linux kernel could be OK. (Note that Linus has come out later and said repeatedly that many proprietary drivers being linked against the Linux kernel are probably not OK. He hasn't litigated any of them yet, but he could...)

However it does not work that way after compilation. After compilation, the resulting binary is a copyrightable work, and is subject to all of the copyrights on its component parts. This is fine as long as you do nothing subject to copyright law with the compiled work. Which people don't in the case of the Linux kernel. (Though when they combine it with various kinds of virtualization they do. Someday I expect to see this litigated.) However if you do do something, like distribute it, then you are now subject to copyright law.

And now we come to the compilation question. It is true that a derived work need not have to compile. Very frequently people derive works in such ways that the result is not compileable. (And usually figure out their typo and fix it quite quickly as well!) However if you have a derived work that is being delivered in compiled form, then it is possible to compile that work. Furthermore if you've distributed that, then you have distributed it in object code or executable form and section 3 of the GPL applies. That section is very explicit, you must offer complete source code. Complete source code is defined in such a way that if you offer something which does not compile, that is not complete source code. Therefore they are in violation of the GPL because they have not offered complete source code.

There could be legal arguments about whether it might be OK to deliver complete source code that compiles A-B-C' down to something that can compile and run, and then you copy a library for C over C' during installation so the program the end user gets is not one they can compile for themselves. Those legal arguments could be interesting, but are moot in this case because they haven't done that. They simply haven't complied with the requirement for comple

Re:Right.... bit of clarification

[Schraegstrichpunkt]

I think the legal theory is as follows:

If you take a core dump of the combined executable that is running in memory, just after the runtime dynamic linking process is complete, then the resulting core dump is a derived work of the program and all the libraries it loads.

In other words, the program that every end user will actually run is a derived work, and the dynamically linked ELF executable is just a fancy way of distributing that derived work to end users.

The idea is that distribution of copyrighted works (including derived works) is always covered by copyright law, regardless of what kind of mechanism you use. In that sense, transmitting the work using a dynamically linked binary is seen as equivalent to transmitting the work using encryption, compression, BitTorrent, or any other mechanism that ultimately results in the recipient gaining a copy of the work.

Re:Right.... bit of clarification

[49152]

I think your definition of a derived work is a bit confused.

From wikipedia: In copyright law, a derivative work is an artistic creation that includes major, basic copyrighted aspects of an original, previously created first work. The rights of the first work's originator must be granted to the secondary work for it to be rightfully called a 'derivative work'. If no copyright permission is granted from the originator, it is instead called a 'copy'.

Wether or not the NONGPL.dll was designed to work or not with the GPL.exe program is completely irrelevant. If the NONGPL.dll however uses copyrightable parts of the GPL.exe in order to work, like datastructures, classes, methods etc it would be a derived work. You basically right with the first statement about beeing useful on its own but then you seem to confuse the issue a bit.

In techincal terms this usually indicates what component links with what. A GPL program linking with a proprietary library and using it to do something useful is OK and the GPL program might actually in legal terms be considered a derived work of the proprietary library. The other way around would violate the GPL license.

Re:Right.... bit of clarification

[49152]

>IChessU's released source code does not compile, so it can't possibly be considered the "complete corresponding source code", regardless of the status
>of the audio/video library (which to the best of my understanding of the GPL is also covered by it).

Unless the audio/video library contains copyrightable parts of the GPL program it IT IS NOT COVERED BY THE GPL.

Why is it so damned hard to understand what "derived work" means.

You cannot make a non GPL library GPL merely by linking with it, because your GPL program would be the derivative work not the other way around.

I wish people would stop making stupid claims like this because it hurts the open source movement more than anything else. Why do you think some people claim the GPL is viral and dangerous. It is mostly because of people like you that dont understand copyright law or the GPL license.

Re:Right.... bit of clarification

[johansalk]

This "Copyright (C) [2006] [unknown]" alone deserve a lawsuit and a buttkicking too, imho.

Re:Right.... bit of clarification

[Brandybuck]

I wish people would stop making stupid claims like this because it hurts the open source movement more than anything else.

Actually I wish the case would move forward, so a judge can finally tell us once and for all that the GPL does not trump copyright law!

Re:Right.... bit of clarification

[49152]

Well said.

Re:Right.... bit of clarification

[nlago]

Isn't this like me releasing a GPL program which is linked to the nvidia or ATI blobs?

No. If you are the author, you can do any weird thing around the GPL and it will not be illegal; it may, however, affect other people's ability to distribute the resulting code (that is, it will be illegal for them to do it, which probably means you shouldn't have used the GPL for your code in the first place).

Now, if you meant "me DISTRIBUTING a GPL program...", then yes, this is exactly the same. And it is illegal according to the GPL. ATI, NVidia and others don't have a problem with this for their linux drivers because Linus explicitly said he thinks it is ok and will not sue (check the /Documentation dir of the kernel source, I think it is there somewhere). X11 is not GPL'd, so there is no problem there.

Hell, isn't it similar if I write a GPL application which uses the Windows API?

No, this is totally different, the GPL explicitly says that mere aggregation does not create a derivative work, and says that linking with standard libraries etc that come with the OS is allowed as an "special exception".

I personally feel as though this is an overreaction, the ichessu site does not hide the fact its based on JIN and offers sources, or am I wrong and this is infact a genuine GPL violation?

This is the classic example of things the GPL was designed to avoid: I take some free code, enhance it, but the enhanced version is not free. The fact that they acknowledge the use of GPL code does not give them the right to violate it, just like taking responsability for a murder won't let you walk free.

Re:Right.... bit of clarification

[waynemcdougall]

Not that the EULA applies:

From http://www.ichessu.com/Download.aspx?subPage=Downl oadClientApplication we read:

Download client for lectures and games
Return to Downloads

I accept all of the foregoing terms either by clicking here or by any further use of the Software.

I do not accept all of the foregoing terms.

There are precisely NULL foregoing terms.

Yeah, sure there is an EULA that FOLLOWS, but I only accepted the foregoing terms so I didn't bother reading that part.

Re:Right.... bit of clarification

[TERdON]

1) I'm an OSX-er
2) Java IS included in OS X
3) Profit! ;-)

It is, however, not included in Windows, but who really cares about outdated operating systems? ;-)

Re:Right.... bit of clarification

[yppiz]

"In this particular case I believe the term is pwned."

I think you mean pawned.

--Pat

No it's not

[brunes69]

Isn't this like me releasing a GPL program which is linked to the nvidia or ATI blobs?It isn't, because ATI and NVidia do not link to the kernel. The portions of the NVidia and ATi driver that *do* link directly to the kernel (also known as the "kernel stub"), are indeed GPL. What happns, is the closed source X driver communicates to and from the stub indirectly, not via linking.

It's actually just a different DRM/DRI implementation, which nearly all X drivers use nowadays.

Note in this case DRM does not mean "Digital Rights management", it means "Direct Rendering Manager"

Re:No it's not

[Anonymous Coward]

the closed source X driver communicates to and from the stub indirectly, not via linking

Then what does it use? Pipes? Temp files? Squirrels? Of course the closed source driver communicates directly with the stup and the stub communicates directly with the kernel. The Stub is a technical trick which allows NVidia and ATI to deliver one precompiled module even though the kernel constantly changes in ways which would require a recompile from source. The stub has absolutely no legal relevancy.

Re:No it's not

[jaseuk]

The blob is firmware for the graphics card, not code for the CPU. So the link is through hardware not software.

Re:No it's not

[TheRaven64]

Sorry, no. The blob for many WiFi cards is firmware. The ATi and nVidia blobs are the majority of the OpenGL implementation, and X11 drivers and run on the host run on the host CPU. Why do you think you can't use the nVidia blobs on PowerPC Linux?

Re:No it's not

[Anonymous Coward]

And that make what difference? You CPU does the linking so all linking is in hardware and not software.

Re:No it's not

[ATMD]

Hey, could you clarify this? I read somewhere on the nvidia site that their Linux blob is exactly the same as their Windows blob, just with a different kernel-linking bit.

I'm guessing this means the blob is executed independant of the operating system? If that's the case, will a Windows/Linux blob work on an x86 Mac?

Re:No it's not

[TheRaven64]

A windows driver takes commands saying things like 'draw a line between points a and b in 3D space.' A Linux driver also takes commands saying the same sort of thing, but with a different call syntax (GDI Vs X11, Direct3D Vs OpenGL, etc). The code actually responsible for drawing the line (or copying a texture to the graphics RAM, or whatever) is the same on both platforms, but there is some glue code required which translates between the platform's public APIs and the hardware's functions.

Re:No it's not

[kabloom]

Incorrect. No part of the Nvidia driver is licensed under the GPL (see /usr/share/doc/nvidia-*/copyright on a Debian distribution). The Linux kernel is licensed under the GPL, but Linus does not consider all binary-only kernel modules to be derived works (only some of them). See http://kerneltrap.org/node/1735.

Thus, our graphics drivers don't appear to be GPL violations.

Re:No it's not

[Anonymous Coward]

Incorrect what? The stub is irrelevant from a legal perspective. It is linked to the kernel and the driver is linked to the stub. From a legal perspective that means that the driver might as well be linked directly to the kernel. The only reason that the driver doesn't need to be licensed under the GPL is that it is not distributed with the kernel. Under these circumstances, you only need to make sure that the driver is not a derived work. Since neither the stub nor the driver itself contain code from the kernel, they are not derived works of the kernel. Since you don't distribute the kernel, you don't need to follow the slightly stricter rule in the GPL, which considers the kernel+driver combination (not the driver itself) a derived work. Linus' opinion in that matter isn't any more relevant than that of any other kernel developer. He doesn't own the Linux kernel. Since Linus isn't a lawyer, I tend to pay more attention to other more legally educated people when the discussion turns to GPL issues.

Re:No it's not

[49152]

I fail to see what the kernel have to do with this at all,

The question was if someone writing a GPL program linking with a non-GPL library/API (like the NVidia driver) would force the non-GPL library to be made GPL too. This would clearly be absurd.

Regarding the article:
I think the important point here is if this A/V library can reasonably be seen as an independent work not derived of the chess client. I.e could it be used as a generic A/V library in any other application.

If it is then the claim that it must be made GPL is ridiculous and hopefully would be laughed out of court.

Think about it for a second, this would be like me taking this JIN client, then rewriting it with support for audio/video chat between clients using Microsofts DirectShow/DirectX library for the A/V handling and then the original copyright holder of JIN requires me or Microsoft to open-source DirectX/DirectShow because it is now used in a work derived from an GPL program. This would obviously be absurd and I dont think it will hold in any court.

Altough an open-sourced DirectX would be nice :-) perhaps someone should try this so we can have more games in Linux.

Re:No it's not

[rm69990]

So? X11 isn't under the GPL... The only part that links to the kernel, which is GPL, is the kernel module, which is also GPL.

Re:No it's not

[stsp]

ATI and NVidia do not link to the kernel.

Correct.

The portions of the NVidia and ATi driver that *do* link directly to the kernel (also known as the "kernel stub"), are indeed GPL.

Apparently they aren't GPL:

/* _NVRM_COPYRIGHT_BEGIN_
*
* Copyright 2001-2002 by NVIDIA Corporation. All rights reserved. All
* information contained herein is proprietary and confidential to NVIDIA
* Corporation. Any use, reproduction, or disclosure without the written
* permission of NVIDIA Corporation is prohibited.
*
* _NVRM_COPYRIGHT_END_
*/

This is from the top of a file called nvidia_linux.c from the nvidia driver stub version 1.0-8762.

What happens, is the closed source X driver communicates to and from the stub indirectly, not via linking.

No, what happens is that the user links the module into the kernel. As long as the user does not distribute the resulting kernel, there is no violation of the GPL.

GPLed what?

[Anonymous Coward]

Wow, did I get confused!

GPL==RIAA

[Anonymous Coward]

This proves the GPL is no different than the RIAA.

MOD ME DOWN!!!!!!!!!!!!

Re:GPL==RIAA

[tehcyder]

This proves the GPL is no different than the RIAA.
MOD ME DOWN!!!!!!!!!!!!

No, this proves the RIAA is no different than the GNAA.
MOD ME UP!!!!!!!!!!!!!

Re:GPL==RIAA

[Anonymous Coward]

No this proves GNAA is no different to Slashdot moderators.

DOWN DOWN DOWN!!!!!11

Re:GPL==RIAA

[Anonymous Coward]

What are you trying to prove anyways?

In a way, this is very lucky timing for GPLv3

[H4x0r Jim Duggan]

It's great that we will get the benefit of this ruling during the year when GPLv3 is being written. This sort of thing provides great suggestions for what should be clearer or worded differently.

Re:In a way, this is very lucky timing for GPLv3

[Watson Ladd]

Why are laws written in english anyway? English is ambiguous, and that's a bad thing. Why not some formal law language with clear semantics and syntax?

Re:In a way, this is very lucky timing for GPLv3

[Anonymous Coward]

Why not some formal law language with clear semantics and syntax?

What language do you propose to define the "clear semantics and syntax" in? Or is there to be no way of translating this "formal" language into anything that anyone can read?

Re:In a way, this is very lucky timing for GPLv3

[russ1337]

We ocasionally need management to understand it.... (it should be written in colours and cartoon pictures of small animals..)

Re:In a way, this is very lucky timing for GPLv3

[RoboRay]

By law, at least in theory, laws are required to be understandable by the general public. Otherwise, ignorance of the law would be a valid defense. If the laws were written in some "law language" that only lawyers and judges understood, they could just tell you that the law said whatever they wanted it to say. How would you know they were lying to you and abusing the system?

Of course, at least in the US, the law may as well be written in different language because even the law-makers don't generally understand what they're voting on.

Re:In a way, this is very lucky timing for GPLv3

[paskie]

Clear semantics and syntax reduces flexibility. This way it can be better applied to very diverse scale of real-world situations, based on the judge's judgement of the "spirit" of the law. Of course the cynical self must add "applied in a way sought by the party with a better lawyer", but I believe that in reality it's usually not really nearly that bad.

Re:In a way, this is very lucky timing for GPLv3

[dracocat]

We could do something where the word "may" means the person is permitted to do something, and the word "shall" means that the person must do something.

Re:In a way, this is very lucky timing for GPLv3

[tehcyder]

Why not some formal law language with clear semantics and syntax?

Like Latin? You could always update the vocabulary for such recent terms as "intarweb", "free-as-in-speech-not-free-as-in-beer", "Mac fanboy" and "DRM".

Re:In a way, this is very lucky timing for GPLv3

[bky1701]

I vote C/C++.

Re:In a way, this is very lucky timing for GPLv3

[KinkyClown]

You mean math?

Re:In a way, this is very lucky timing for GPLv3

[TheRaven64]

If the laws were written in some "law language" that only lawyers and judges understood, they could just tell you that the law said whatever they wanted it to say.

I take it you've never read a statute, or any case history then? Laws are already written in legalese, and correctly interpreting them often does require legal training. Sadly, ignorance of them is still not an acceptable excuse.

Re:In a way, this is very lucky timing for GPLv3

[Anonymous Coward]

No, Fortran is more structured. Use Fortran!

Re:In a way, this is very lucky timing for GPLv3

[Konster]

Most laws are written in English, but with an unintelligible layer as posing as mitigant in between.

This mitigant layer is known as Latin, and is only spoken by academics that have never had actual sex with another human being. It is also spoken by various other institutions that seek to lend credence to their arguments based solely upon the perceived acceptance of Latin as the tongue of choice by institutions of higher learning.

Re:In a way, this is very lucky timing for GPLv3

[Haeleth]

I vote C/C++.

Oh, great, as if our current laws weren't bad enough, now we get to have them written in an unsaf
Segmentation fault (core dumped)

Re:In a way, this is very lucky timing for GPLv3

[kripkenstein]

Why are laws written in english anyway? English is ambiguous, and that's a bad thing. Why not some formal law language with clear semantics and syntax?

There are very good reasons why this is impossible. Actually, what you are proposing is a very natural notion, that sadly turns out to be wrong. I say 'natural', because during the first third or so of the 20th century, philosophy (of language, in particular), was seeking exactly what you are driving at - a 'pure' language, free from ambiguity. This would have had benefits for legal matters, as well as philosophical ones, and even metamathematical implications. But this was shown to be a futile attempt (Wittgenstein being the major figure showing this). I'll briefly summarize why this is so.

First, when you refer to human-related things - as laws are, they mention e.g. 'assault', 'homicide', and so forth - there is no way to 'clean up' the language. It cannot be made unambiguous, because the underlying concepts are ambiguous. Try to define (as the famous example goes) 'game'. For any suggested definition, there are counterexamples (e.g. not all games are about winning or losing, not all games have scores, not all games are fun, etc. etc.). This is a simple consequence of the fact that life is complex - we use the word 'game' in many contexts, in many ways. Unlike in math, where we start with definitions, in the law we start with pre-existing human concepts and try to work with them. We therefore cannot arrive at unambiguous statements.

Second, and this is a more subtle issue, language is meaningless without a context of use. By this I mean, that if you see some scribbles on a page, they are worthless without someone to read them. A sentence + a reader are what is necessary for 'meaning' to exist. Thus, even if we write what we believe to be unambiguous text, we can never remove the element of the reader: for us, the statement is unambiguous, but for another person, with a slightly different mindset, it may not be so. You may claim that your interpretation is 'correct', but that will not avail you when a matter is put before the public, i.e. open to interpretation by many people, as the law must be.

Sorry to go on at length, but this is a fascinating topic for me.

Re:In a way, this is very lucky timing for GPLv3

[g2devi]

> Why are laws written in english anyway? English is ambiguous,

Seeing that few judges, laywers, or lay people understand Lojban ( http://en.wikipedia.org/wiki/Lojban ), there isn't much choice.

The key thing is, any language that has poetry and metaphor, is ambigious by its very nature because these things tend to create multiple meanings for a single noun or verb or phases. Any language with slang like "so hot it's cool" or "so bad it's good" (i.e. most languages) is also ambigious.

Since most popular languages have flaws and no-one wants to use unpopular unambigious languanges, it makes most sense to settle on the most popular language (which tends to absorb concepts from all languages) and work with that (even if it means rewriting contracts every 100 years as different languages go into fashion).

As a side note, the GPL has several official translations.

Re:In a way, this is very lucky timing for GPLv3

[Mr. Underbridge]

Because such a thing doesn't exist outside Neal Stephenson novels?

Re:In a way, this is very lucky timing for GPLv3

[sweetnjguy29]

Of course, not all laws are written in English...Canadian laws are written in both French and English. I would go as far to say that most countries write their laws in their own official languages, keeping in mind that most countries have more than one official language...

As far as law and ambiguity goes, you would think a simple statement like, "Thou shall not murder" is clear and precise. It is not. It leaves open to question for killing in self defense, killing for necessity, killing accidently, and killing in behalf of your country or when a diety commands it.

Laws are like computer code...on paper. Execute code on a machine, and the machine reacts precisely. Proclaim a law on people...and people should follow it exactly...except that people react and behave wildly and differently and come up with a myriad of exceptions, explanations, situations and rationalizations not to follow the letter of the law. There is no such thing as "fairness" and "equity" in computer machine code. It is all 1s and 0s...true and false...no grey in-between...at the end of the day.

Re:In a way, this is very lucky timing for GPLv3

[Anonymous Coward]

You're missing out on the Powerpoint -- gotta have powerpoint with lots of bullets and sound effects.

For EULAs, make sure to get rid of all the complex restrictions languages and replace it by these two bullet item:
      * We own your business, your data, and your programs.
      * You're responsible if anything we do goes wrong even though we're the only one who can make changes.

Re:In a way, this is very lucky timing for GPLv3

[Anonymous Coward]

Math is only unambiguous if you have unambiguous axioms. How do you plan on defining those legal axioms?

Re:In a way, this is very lucky timing for GPLv3

[R2.0]

1) There is - it's called Latin.

2) "Laws" per se are written by "lawmakers" - you know, politicians. A lot of these folks are barely literate in their own language, much less requiring them to learn another.

3) Was I the only one to find the parent remark +1 Humor?

Re:In a way, this is very lucky timing for GPLv3

[udoschuermann]

Why are laws written in english anyway? English is ambiguous, and that's a bad thing. Why not some formal law language with clear semantics and syntax?

Because then you'd have to have specially trained and licensed practitioners who understand that language and are able to interpret it to the rest of us. Ummm, waitaminute...

Re:In a way, this is very lucky timing for GPLv3

[Beryllium Sphere(tm)]

Describing the real world in a formal language is a huge project (see Cyc, for example) and a workable law written in a formal language would be big and complex. Which means it will have bugs.

With laws written in English, there's the chance for an educated and (we hope) neutral judge to make bug fixes without needing the authority to rewrite the bug-free parts of the law.

Imprecision can also be good because the gaps it leaves give room for principles to work. Where libel law isn't explicit, then free speech rights will win.

The cynical view would be to point out who writes the laws and who makes money when an ambiguity lands in court.

Re:In a way, this is very lucky timing for GPLv3

[xanalogical]

> "Laws" per se are written by "lawmakers" - you know, politicians.
> A lot of these folks are barely literate in their own language,
> much less requiring them to learn another.

That's something that puzzles me -- they do seem awfully poor at crafting readable, correctly scoped, airtight laws yet so many politicians _did_ get a law degree in which they are graded on being able to generate legal contracts with those properties. Well, maybe not always 'readable' but a lawyer who writes a leaky, poorly worded contract is going to flunk out. Yes, I know colleges churn out some lawyers with poor skills but _that_ many (or why do they all congregate in politics)?

I mean, most have law degrees so they're not semi-literate non-college folk. Yet their behavior would indicate otherwise. How did they get thru law school being unable to compose a paragraph?

Re:In a way, this is very lucky timing for GPLv3

[evil_Tak]

As a side note, the GPL has several official translations.

From http://www.gnu.org/licenses/translations.html
The reason the FSF does not approve these translations as officially valid is that checking them would be difficult and expensive (needing the help of bilingual lawyers in other countries). Even worse, if an error did slip through, the results could be disastrous for the whole free software community. As long as the translations are unofficial, they can't do any harm, and we hope they help more people understand the GPL.

We give permission to publish translations of the GPL, GFDL, or LGPL into other languages, provided that you (1) label your translations as unofficial (see below for how to do this), to inform people that they do not count legally as substitutes for the authentic version, and (2) you agree to install changes at our request, if we learn from other friends of GNU that changes are necessary to make the translation clearer.

Re:In a way, this is very lucky timing for GPLv3

[asuffield]

First, when you refer to human-related things - as laws are, they mention e.g. 'assault', 'homicide', and so forth - there is no way to 'clean up' the language. It cannot be made unambiguous, because the underlying concepts are ambiguous.

While this is largely true, there are a number of ways in which the language could be significantly improved upon. Basically, all the nouns and verbs (and noun-/verb-derived adjectives/adverbs) that do not reference mathematically defined concepts are subjective in nature (we can precisely and unambiguously define 'zero', but we cannot give an objective definition for 'pornography'). So as far as those are concerned, you cannot improve on them.

However, those are only part of a language. The other part is all the structural stuff, showing how the subjective words connect together. Human languages are flexible here because they have evolved based on what "sounds good" or "looks good", rather than what is most precise. Unfortunately, they are also ambiguous. There is no good reason why they need to be ambiguous though. It seems to me that you could construct a useful language by using English concept words as predicates, and a syntax based on formal logic. It would probably be easier and more precise to construct laws and contracts in this fashion. It would not solve all the problems of ambiguity that occur, but it could solve some of them.

For example, if you were to write "one person injuring another with a weapon" as "injure(person A, person X where X != A, weapon)", then there is still ambiguity about what "person", "injure" and "weapon" mean, but there is no ambiguity about the fact that we are talking about two distinct "people", and that one of them is "injuring" the other with a "weapon", whatever those three words happen to mean (assuming you set up the syntax that way) - so we've nailed the ambiguity down to a known set of concepts, which a court can then consider. We are certain that this statement does not apply to a person injuring themself, for example.

Re:In a way, this is very lucky timing for GPLv3

[Anonymous Coward]

Perfect idea!

if(suspect->guilty = true)
{
        suspect->goToJail();
}

Re:In a way, this is very lucky timing for GPLv3

[RoboRay]

I take it you've never read my entire post, but simply hit "Reply" half-way through it.

Re:In a way, this is very lucky timing for GPLv3

[bky1701]

Yeah, but the DMCA gives a small error on compile for some reason....

Re:In a way, this is very lucky timing for GPLv3

[Anonymous Coward]

I would like to thank you for the most interesting post I've read today.

Thank you.

Re:In a way, this is very lucky timing for GPLv3

[schweini]

well put!
i'd just like to add that i once had a homework in a class about "formlization of knowledge" where the professor asked us to 'translate' the german laws controlling heritage into formal language. we were astounded by the gaping hole we found inside the formalization - when we brought this up, he said:
"see? those holes are were humanity lies", referring to the fact that all laws must be subject to interpretation, and must be ambigous - in part, that's what we have judges for: to judge in the not-so-obvious cases what is 'right', under the circumstances.

Re:In a way, this is very lucky timing for GPLv3

[swarsron]

don't be sorry, i think it's very fascinating too. Thank you for your explanation

Re:In a way, this is very lucky timing for GPLv3

[Ambush Commander]

Precisely. Cordon off all the ambiguous stuff, and program the rest of it.

generous offer

[backwardMechanic]

Is it just me, or does $4k sound very cheap for full rights to the source code?

Re:generous offer

[DrXym]

You'd have thought so, but apparently some people aren't even going to pay that much. $4000 would buy you a contractor for 8 days. If you're lucky you might have a semi-working GUI with big chunks of missing or bugged functionality at the end of that.

Re:generous offer

[Anonymous Coward]

From TFA, it seems IchessU use russian developers. You get more than 8 days for $4000 in Russia. Other than that, you are of course right. Cheap price.

Re:generous offer

[Anonymous Coward]

RTA carefully. The $4,000 was not for full rights, but for commercial applications rights; a commercial license. I'm guessing that the terms the author offered would have allowed IchessU to modify the GPL-ed code at need, making additions as they wished, and have permission to make the resulting application proprietary. The "original" code base would have remained GPL. They were buying the right to make a commercial product from it.

Open source does not have to mean free as in speech, nor even free as in beer. It just means that the source is open to scrutiny. Also remember that the author retains copyright and may deal with his work as he pleases, barring self-imposed constraints on that freedom when making use of some licensing scheme such as the GPL.

In this case, it looks like purchasing a license for commercial use was too expensive for them, so they went the "regular" GPL route. They then overstepped the GPL bounds by abusing the terms of the license and refusing to fufill the requirements it outlines.

Re:generous offer

[Gorath99]

As I understand it, the offer was not for full rights, but "merely" a license for distribution and further development.

This difference may not sound important, but it is. If the full rights were sold, then the original developer could no longer license others, and could no longer enfore the GPL on the code. (Only the copyright holder can do so.)

It's still cheap though.

Re:The story by the developer, Alexander Maryanovs

[byolinux]

That's the first link in TFA ;)

GPL doesn't need to be tested.

[Kidbro]

Here we go again... GPL doesn't need to be tested in court. GPL doesn't restrict you from doing anything. The only thing GPL does is to allow you to do some things with copyrighted work - such as, under certain circumstances, distribute said work even if you are not the copyright holder.
What's being "tested", if anything, is copyright laws. And I believe that we all can agree on the fact that they are already, if nothing else, fairly tested in court.

Re:GPL doesn't need to be tested.

[arr28]

Those people who use the GPL to licence their software do so on the understanding that they are granting some particular rights to 3rd parties. However, for all we know, in some jurisdictions, the GPL may grant more rights (or fewer rights) than intended.

Therefore, contrary to you claim, court cases such as these do indeed "test the GPL". They test whether or not it has the desired effects.

Re:GPL doesn't need to be tested.

[squiggleslash]

I know what you're trying to say, but that's kind of wrong because it presumes all cases of "testing" are of the form "The GPL is invalid, therefore I can do what I want." This is not true. Not all cases are of that form.

If someone interprets the GPL as meaning it's ok to release a proprietary extension to a GPL'd program using some type of unusual linkage mechanism, then yes, it may be necessary to test this in court depending on the beliefs of the person who wrote the code they're licensing. The person who's interpretting the GPL that way isn't arguing that the GPL is invalid, and that they don't have a license, they're arguing that they do have a license, and it means something other than that the licensor thinks it means.

Whether that's applicable in this case is for someone who read TFA to answer.

Re:GPL doesn't need to be tested.

[Decameron81]

"GPL doesn't restrict you from doing anything."

Actually it does. Or can you merge GPL code and non-GPL code in a single codebase? This is a restriction of how you can use the code you obtained through the GPL license.

Re:GPL doesn't need to be tested.

[FooBarWidget]

Actually it does. Or can you merge GPL code and non-GPL code in a single codebase? This is a restriction of how you can use the code you obtained through the GPL license.

That is not a restriction that GPL adds: you never were allowed to do that in the first place. Suppose the code is not under any license, then standard copyright law applies. And copyright law does not let you copy that code to your code AT ALL.

Re:GPL doesn't need to be tested.

[phlipped]

GPL doesn't restrict you from doing anything

Are you kidding? Of course it does. It is one of the most restrictive "free" licenses I've seen (compared to "You can do whatever you like with this code - even claim it as your own if you want, although that would be pretty low" as is found on some exec extension to the NSIS installer framework
From the preamble to the GPL ...

To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.

You can use the vanilla software/code personally in whatever way you wish, but if you modify and/or redistribute it, then a whole bunch of restrictions come into play to define under what conditions you may do this.

Re:GPL doesn't need to be tested.

[nomadic]

That is not a restriction that GPL adds: you never were allowed to do that in the first place. Suppose the code is not under any license, then standard copyright law applies. And copyright law does not let you copy that code to your code AT ALL.

Of course it does. If YOU hold the copyright to that code, you could merge the code if you wanted--EXCEPT for the GPL. Same thing if you got permission from the copyright holder to do it. Software licensing is very, very common.

Re:GPL doesn't need to be tested.

[SashaM]

Here we go again... GPL doesn't need to be tested in court.

I think you are right in that it doesn't need testing in that sense. But a precedent does need to be set that if you violate the GPL, the court will punish you for it. Maybe if I win this lawsuit, the next IChessU will think twice before doing something like this.

Alexander (aka Sasha) Maryanovsky

Re:GPL doesn't need to be tested.

[Anonymous Coward]

You are right, of course - if it is your code you can merge whatever you like into it.

What the grandparent poster is referring to is the fact that copyright law then prevents you from redistributing the resulting merged code to anybody else (i.e. you cannot sell it or even offer it for free download). It is the GPL that then explicitly permits the act of redistribution; copyright law is the one doing the restriction in the case you mention.

I hope that clarifies things - in a way both you and the grandparent are right!

Re:GPL doesn't need to be tested.

[Kidbro]

Agreed.
However, any punishment will come from the fact that they've broken copyright laws, rather than the GPL.

I want to make it very, very clear that my comment was not directed at you, Sasha. It was directed at the myth (lacking a better word) that the GPL needs to be tested. I'm annoyed with the slashdot blurb for "supporting" that myth, but I'm definitely not annoyed with you for taking things to court. Quite the opposite, I wish you the best of luck.

Re:GPL doesn't need to be tested.

[Greyfox]

You can merge GPL code and non-GPL code all day long. You just can't distribute the results. I am not a lawyer but I've watched every episode of "Ally McBeal." Twice. I'm pretty sure that the courts would view distributing a set of patches or a binary DLL designed specifically for one application as a derivative works issue. I'd think that'd be pretty cut and dried. It's less obvious to me what would happen if you made a general API for embedding new components into an application and you or someone else patched a piece of code to incorporate that code. I suspect that it might be impossible to distribute the interface code between a GPLed program and non-GPLed modules. I'd say we were in this ambiguous area of copyright law except for the fact that the licensed code in question seems to be only designed to work with one piece of GPLed code. In my mind that very clearly makes it a derivative work.

The most ambiguous test I could think of would be if someone published a plug-in API and several authors implemented it into their code and then someone used the API to write some non-GPL plug-ins. That would make for an interesting legal test, but I think it'd still work in terms of whether copyright law itself allowed those plug-ins to be distributed.

Re:GPL doesn't need to be tested.

[Anonymous Coward]

Well, since courts can strike down parts of a license as unenforcable, it probably does need to be tested. You know, to see what's left standing so we can figure out what, as far as court enforcement is concerned, the license says? It's really no different in that regard than any EULA; any EULA author can claim they're not restricting you from doing anything but rather permitting you to do something in a very specific way. Yet, courts can still strike down parts of the license. That's how that area of law works.

But for a moment, let's pretend your statement were correct: that there were no legal need to test the license.

Ok. Except a lot of people think it does need to be tested. Sure, they're wrong (for purposes of this argument). But that's not the point. The point is, they think they're right, and a lot of people of that mindset are in positions to influence or make decisions about what software to use at such-and-such company. So they choose to avoid the software with the untested license, because they don't know what they can or can't do with it.

Of course, if the license were tested, and if it stood up (which by your logic, it must), then that concern would evaporate. So the GPL does need to be tested. Even if there were no legal need for it, there would be a business need for it.

Re:GPL doesn't need to be tested.

[Kjella]

Of course it does. If YOU hold the copyright to that code, you could merge the code if you wanted--EXCEPT for the GPL. Same thing if you got permission from the copyright holder to do it. Software licensing is very, very common.

I read what you wrote, and it doesn't make any sense. If it's all your code, then you're not bound by the GPL - you're the copyright holder and can do whatever you want. If you get permission from the copyright holder of the GPL'd code, you can do it - the copyright holder can give several independent licenses with different conditions. Licensing code under the GPL does not in any way limit other licensing options. Any and all limitations on distribution come from copyright law..

Re:GPL doesn't need to be tested.

[mseitz]

You are correct that there is no need for a court to test the "validity" or "legality" of the GPL. But there are aspects of the GPL that would benefit from being tested in court. The GPL is written in relatively broad language. There are disputes over how to interpret that language. So what needs to be tested in court is the interpretation of the GPL, and how it applies to specific actions.

In other words, what needs to be tested in court is:

a) whether the GPL allows a specific act.

For example, there is much dispute over whether the GPL allows a closed source program to dynamically link to a GPL library. It will eventually be up to a court to decide whether the GPL allows this.

b) whether a specific act fulfills the requirements of the GPL.

In this case, the court must decide whether the GPL allows what the defendant has done.

Re:GPL doesn't need to be tested.

[civilizedINTENSITY]

Of course you, as a user, can mix and match. You just can't distribute it if the licenses aren't compatible. Copyright is about distribution, not use.

Re:GPL doesn't need to be tested.

[EzInKy]

You can use the vanilla software/code personally in whatever way you wish, but if you modify and/or redistribute it, then a whole bunch of restrictions come into play to define under what conditions you may do this.


Those restrictions make GPL'd software very profitable for anyone who uses more software than they write. It's an awesome deal if you really think about it...for the mere price of distributing a few lines of your own code you gain the rights to distribute millions of lines written by others.

Re:GPL doesn't need to be tested.

[squiggleslash]

Or can you merge GPL code and non-GPL code in a single codebase?

Can you do that if you do not agree to the GPL?

The GPL isn't what's restricting you from doing that. The GPL is making some acts legal that otherwise wouldn't have been. The fact it doesn't allow you, for example, to murder Eric Raymond doesn't mean the GPL is restricting you from doing something.

Re:GPL doesn't need to be tested.

[swillden]

if it is your code you can merge whatever you like into it.

Nope. If both codebases are yours, you can do what you like with them, but if the copyright on one of them is owned by someone else, merging them constitutes preparation of a derived work, which copyright law prohibits without permission of the copyright holder.

That particular aspect of the law is generally pretty toothless -- how is the other copyright holder ever going to know what you did? If, however, the copyright holder was able to find out that you did it, he could sue you for it.

If the other work is licensed under the GPL, then you have permission to merge it with yours, as long as you abide by the terms of the GPL. If you never distribute the result, none of the GPL's requirements apply.

Re:GPL doesn't need to be tested.

[noidentity]

GPL doesn't restrict you from doing anything. The only thing GPL does is to allow you to do some things with copyrighted work [...]

And if the GPL is found invalid in court, then these allowances become void, and this you aren't legally allowed to do anything with the code. So an infringing company attempting to avoid the GPL by challenging it in court will lose either way, either by having the GPL upheld and found to be infringing, or having the GPL found invalid, and found to be infringing as well.

Re:GPL doesn't need to be tested.

[spitzak]

Huh?

If YOU own the copyright to the code, it does not matter if you previously released it under the GPL. It is your code and you can release it again any way you want. Many people do dual licensing, or discontinue supporting their GPL version. There is nothing wrong with that.

Re:GPL doesn't need to be tested.

[Anonymous Coward]

Huh?

You wouldn't be able to do that anyway. It says what you can do with the code (albeit with some restrictions) as opposed as to what you would normally not be allowed to do.

If it said that licencees of the code were not allowed to rape the authors mouth if they used it it would not be restricing them from anything as they would not, otherwise, be at liberty to rape the authors mouth.

If it said that licencees were not allowed to wear pink hats if they used it then the licencee would be perfectly at liberty to not use the code and be able to wear a pink hat.

If you don't like the licence don't use the frigging code!

Sheesh.

Doesn't matter

[brunes69]

Pipes, temp files, sockets, none of these are covered by the GPL. The GPL covers explicitly *linking only*. If a GPL'ed piece of software could not communicate with a closed source piece of software over a socket or pipe, the Apache web server would not exist.

To be specific - I am pretty sure the drivers use either a UNIX socket or a named pipe.

Re:Doesn't matter

[oliverthered]

Which line of the GPL would that be? The GPL only talks about derived works.

Re:Doesn't matter

[Ant P.]

Apache isn't GPLed anyway.

Re:Doesn't matter

[brunes69]

Yeah, but if a GPL web browser connected to it to view a web page, that would be causing Apache to be voilaitng withe GPL since it is communicating with a GPL app over a socket!

Re:Doesn't matter

[brunes69]

This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Lesser General Public License instead of this License.

http://www.gnu.org/licenses/gpl.txt

Re:Doesn't matter

[oliverthered]

That comes after the 'end of terms and conditions' bit.

Re:Doesn't matter

[mrchaotica]

To be specific - I am pretty sure the drivers use either a UNIX socket or a named pipe.

That's a pretty extraordinary claim, given that the graphics card drivers should be some of the highest-performance ones in the system. I would have thought that they, more than anything else, would require direct linking. Anyway, do you have a citation that supports your statement?

Re:Doesn't matter

[SashaM]

Pipes, temp files, sockets, none of these are covered by the GPL. The GPL covers explicitly *linking only*. If a GPL'ed piece of software could not communicate with a closed source piece of software over a socket or pipe, the Apache web server would not exist.

Umm, could you clarify what you mean? Apache is Apache-licensed, not GPL, is it?

I also don't see where the GPL refers specifically to linking. According to my reading of the GPL FAQ entry on mere aggregation, if the two pieces communicate data which is internal and specific to the GPLd piece, the other piece is also covered by the GPL. This means, for example, that you can't wrap a GPLed program in a network API (RMI and such) and extend it while keeping your extensions proprietary.

Alexander (aka Sasha) Maryanovsky - the author of Jin.

Re:Doesn't matter

[morgan_greywolf]

Yeah, but if a GPL web browser connected to it to view a web page, that would be causing Apache to be voilaitng withe GPL since it is communicating with a GPL app over a socket!

Of course, none of the mainstream Web browsers (Opera, Mozilla, Firefox, IE, Konqueror, Safari) are licensed exclusively under the GPL. And even Epiphany, the GNOME Web browser, which is GPL, uses Gecko, which cross-licensed. I think there is some GNU FAQ somewhere that says that communication over pipes or sockets is okay.

Re:Doesn't matter

[brunes69]

That phrase is clarifying items already specified in the GPL. The GPL excludes linking. If you don't believe me email RMS, or do a Google search. I don't have any more time to debate with someone unwilling to do simple research.

Re:Doesn't matter

[oliverthered]

The GPL prevents static linking because you are copying the library into you application/library and the GPL is a copyright license and prevents this because of the clauses about derived works.

However when you dynamically link you do not take a copy of the code and since the GPL is a copyright license there's nothing it can do to stop you unless you distribute both the GPL library and your work together.

I won't bother asking RMS or the FSF I'll just look at the GPL myself and take a broad opinion of the interpretation of the GPL and not rely on those who have afterthoughts about what the GPL should be but isn't.

As for Google try searching for gpl prevents dynamic linking yourself.
You could also try googling for abstraction filtration comparison to find out what exactly is a derived work.

It appears that you have failed to do the simple research.

Re:Doesn't matter

[Anonymous Brave Guy]

According to my reading of the GPL FAQ entry on mere aggregation, if the two pieces communicate data which is internal and specific to the GPLd piece, the other piece is also covered by the GPL.

Perhaps that was just an unfortunate choice of words, or a small language barrier if you're not a native English speaker. However, unless your law works very differently to most, you can't force anyone else to licence their code in any specific way, including under the GPL. To say "the other piece is also covered by the GPL" would therefore be wrong.

Of course, it may be a copyright infringement for them to use your (GPL'd) code in their product if their code is not released with a compatible licence, but that has very different implications. If they're infringing your copyright, a court might be able to issue an injunction banning the use of your code and/or award you compensation, for example. However, the code behind "the other piece" hasn't magically become licensed under the GPL, and no-one has any magical rights to see/reuse the source code.

(IANAL, don't get legal advice from Slashdot, etc. etc.)

Re:Doesn't matter

[SashaM]

You are, of course, correct, but in my post I was already assuming that the owner of the "other" piece attempts to use the GPLed piece under the GPL, and discussing whether to do so legally would require him placing his piece under the GPL as well. If I wasn't clear in my post then I hope I am now :-)

Btw, the only GPL-compatible license (in the sense you're referring to) is the GPL, as it requires the entire application to be licensed under the GPL. Usually, though, when people say that a license "GPL-compatible" they mean that it allows relicensing under the GPL.

Alexander (aka Sasha) Maryanovsky.

Re:Doesn't matter

[MadEE]

As for Google try searching for "gpl prevents dynamic linking" yourself.

It doesn't help you case when the first link google returns is this:
http://www.oslawblog.com/2005/01/dynamic-linking-g pl-and-lgpl.html

Re:Doesn't matter

[SashaM]

Yeah, but if a GPL web browser connected to it to view a web page, that would be causing Apache to be voilaitng withe GPL since it is communicating with a GPL app over a socket!

Umm, no. There is no distribution of software (especially not Apache) in the case you're describing, so the GPL is irrelevant. It's also irrelevant (in your case) for more reasons, but this is certainly the first one.

Alexander (aka Sasha) Maryanovsky.

Re:Doesn't matter

[oliverthered]

three points
1: it's a wikipedia artical.
2: the words we believe
3: It's Free Software Foundation retoric.

and belive is the correct word, the Law (or I know) uses Abstraction Filteration Comparison.

You should have checked out the second link, or the third link 'The GPL simply prevents you from _distributing_ the GPLed code'

or maybe a few links afterwards that state things like

'For example, technologies like dynamic linking and distributed object computing are often not interpreted as creating derived works. In contrast to these more traditional views, the GPL and its originators, the Free Software Foundation (FSF), propose viewpoints that they open admit push interpretation of copyright law to extremes.'

Which I take to mean, this is what we want the GPL to mean not what it actually means.

Re:Doesn't matter

[Anonymous Brave Guy]

Fair enough, just wanted to be clear.

FWIW, I was thinking of things like dual-licensing when I mentioned being GPL-compatible. I guess technically that would be permitting two separate licenses, one of which happens to be the GPL.

Re:Doesn't matter

[Anonymous Coward]

Graphics card drivers communicating through sockets... Whatever you're smoking -- pass it on!

Re:Doesn't matter

[MadEE]

1: it's a wikipedia artical.

The quote is a definition unless you disagree that that is what dynamic linking is, and even if you did as it's intended to clarifiy the subject rather then prove it, being from wikipedia means nothing. What matters is FSF's take on the issue which is the second quote.

2: the words we believe

We should probably believe the FSF since it's their license.

3: It's Free Software Foundation retoric.

It's their license that they wrote. The license itself is their "rhetoric" to argue that a document showing the intent of the license is any less valuable then the license itself is foolish. Both would be taken into consideration on the case. Until a court rules upon this everything, by your standards (including your forum quote below), is rhetoric.

and belive is the correct word, the Law (or I know) uses Abstraction Filteration Comparison.

That would be "Abstraction Filtration Comparison" test is a determination if infringement took place when there is a dispute to the origin of code within a software program. There is no dispute here just an issue of contract; the author of the infringing program acknowledges using the code from the Jin Chess client. This test is unnecessary.

'For example, technologies like dynamic linking and distributed object computing are often not interpreted as creating derived works. In contrast to these more traditional views, the GPL and its originators, the Free Software Foundation (FSF), propose viewpoints that they open admit push interpretation of copyright law to extremes.'

Right so you knock wikipedia then all of a sudden some guys post on a forum becomes gospel? Regardless he doesn't say the GPL's viral aspects end with dynamic linking he explicitly says it does argues; but without a case to back up there is no guarantee that that part of the license will hold up in a court of law. That of course is one mans opinion.

Re:Doesn't matter

[MoogMan]

How about runtime linking with ldopen/LoadLibraryEx ?

Re:Doesn't matter

[Panaflex]

It really appears here that this dispute is not about your client - but more a dispute of cash. Why not offer a per-seat licensing scheme to them - for say a small fee per initial user and/or 5% of monthly per-user revenue or something? That would be far more acceptible IMHO. If it's cheaper for them just to come clean and GPL their DLL, I'm sure they will get by with that and you'll have nothing to force them to pay up. Though you will have shown the enforcement of the GPL is viable in Israel.

Reguardless - I think the GPL is pretty explicit - the question here is whether they distribute the dll with your code.

Distributed, but merely considered a bundle:
If the extension runs in a separate process (through fork/exec, for instance) then they are considered two separate programs and there is no GPL violation. There is one possible violation here - that the main functionality of the program is merely exported between processes. But since the main functionality of the program is chess and the dll is only adding minor non-chess features I see this as a far fetched violation and very difficult to prove.

Distributed together, accessed through a plugin:
It really depends if the plugin has a special non-GPL clause. I would assume (from the lawsuite) that there is no plugin API which could be used legally, however if you have allowed a proprietary plugin exception - then this is explicitly allowed.

Distributed separatly, accessed through a plugin:
This merely puts the violation into the users hands - as the plugin API as merely a code-fork allowed by the GPL. The DLL is then "copied" by the user - this is a violation by the user.

I'm not a lawyer - and this is not legal advice. I'm simply relating my own experience.

Re:Doesn't matter

[Schraegstrichpunkt]

Of course, none of the mainstream Web browsers (Opera, Mozilla, Firefox, IE, Konqueror, Safari) are licensed exclusively under the GPL.

Lynx is, and it's been in reasonably widespread use for longer than a lot of the browsers you've listed.

Re:Doesn't matter

[rm69990]

Well, considering the Apache webserver never has been released under the GPL, that is an extremely bad example. The Apache Web Server is released under the....Apache open source license.

Re:Doesn't matter

[Brandybuck]

The GPL covers explicitly *linking only*.

The GPL doesn't cover even that. It only covers derivative works. Linking is not sufficient to create a derivative work. http://www.linuxjournal.com/article/6366

Re:Doesn't matter

[Brandybuck]

RMS' opinions on the matter are irrelevant, because the only thing that matters is copyright law. And copyright law does not say linking is derivation.

Re:Doesn't matter

[oliverthered]

We should probably believe the FSF since it's their license.

That's the one reason you shouldn't belive the FSF, they have an agenda and there not going to say you can do something they don't like even if you can.

Right so you knock wikipedia then all of a sudden some guys post on a forum becomes gospel?

No, I'm saying that people that look at the problem all seem to find a loop hole in the GPL when it comes to dynamic linking, people who don't look at the problem and belive the fsf stick the the fsf line.

The reason I mention "Abstraction Filtration Comparison" is because it's the test used for a derived work, and the GPL only talks about derived works because that's all you can cover using copyright law. (The only regerence to linking is after the end of the terms of the GPL)

In this case if they statically linked against a GPL app or library then there's a problem, and if they distributed a closed source library or application with a GPL library then there's a problem. If they dynamically linked then the GPL will probably fall down.

Re:Doesn't matter

[oliverthered]

Think about it for a minute.

The GPL is a Copyright license and not an end user license; it relies solely on copyright law.

Interfaces and data structures aren't covered by copyright because they are classed as facts. Ref Abstraction Filteration Comparison.

When you dynamically link the only thing you copy is the interfaces and data structures.

So, if I distribute a dynamically linked application without the required libraries the application that I am distributing contains no copyrightable data from the GPL library.

In that case how can a copyright license cover something which hasn't violated copyright?

Re:Doesn't matter

[MadEE]

We should probably believe the FSF since it's their license. That's the one reason you shouldn't belive the FSF, they have an agenda and there not going to say you can do something they don't like even if you can.

Wouldn't you think that GPL would be designed to fulfill their agenda if it is indeed written by them? It would be counterproductive if it doesn't fit their agenda. Secondly they provide the only information available written by legal experts (i.e. can legally practice law) readily available that I can find at least, forgive me if I hold the opinion of professionals in higher regard then random forum posts. Though if you can opinions by other lawyers I would love to see them.

Right so you knock wikipedia then all of a sudden some guys post on a forum becomes gospel? No, I'm saying that people that look at the problem all seem to find a loop hole in the GPL when it comes to dynamic linking, people who don't look at the problem and belive the fsf stick the the fsf line.

First of all a single person's post is not a consensus, it's clearly not that if this issue is in debate. Regardless, the guy didn't even say that he simply claimed its legality is in question and that until a case comes along to prove it either way you should act with caution whatever you believe. Of course if you posted the whole paragraph instead of the snippet it would have been far more obvious.

The reason I mention "Abstraction Filtration Comparison" is because it's the test used for a derived work, and the GPL only talks about derived works because that's all you can cover using copyright law. (The only regerence to linking is after the end of the terms of the GPL)

The reason I smacked down your mention of abstraction filtration comparison is because it is only used when the origin is in dispute. The abstraction filtration comparison test does not mitigate damages. Again the origin is not in dispute here he has the library on his web page in which he re-releases under GPL.

In this case if they statically linked against a GPL app or library then there's a problem, and if they distributed a closed source library or application with a GPL library then there's a problem. If they dynamically linked then the GPL will probably fall down.

The problem I see with that is functionally static and dynamic linking are the same when the application is placed in user memory since it is clear that the author intended that to happen and furthermore needed that to happen for it's operation the court will likely look at that as the intended out come and probably rule it no different then a statically linked app. If your software depends on a library such that the proper operation of the program will cease without it the app is derived from the library plain and simple. If he did not distribute the library a case could be made he didn't agree with GPL.

Re:Doesn't matter

[MadEE]

The GPL is a Copyright license and not an end user license; it relies solely on copyright law.

That is correct, and the individual is distributing the GPLed software so he has clearly accepted the agreement.

Interfaces and data structures aren't covered by copyright because they are classed as facts. Ref Abstraction Filteration Comparison.

Nice try:
"The Fifth Circuit first held that only literal copying of elements of a computer program could be considered an infringement, [1] then it simply recognized that non-literal elements of computer programs "such as structure sequence and organization" may be copyrightable, [2] and then applied the abstraction-filtration-comparison test to conclude that copyright protection could in principle exist in input and output formats and used interfaces as long as these formats were not dictated by external requirements. [3] The court found that the expressive nature of such interfaces outweighed their utilitarian function." [1] Plains Cotton Co-Op v. Goodpasture Computer Service, 1 USPQ2d 1635 (1987). In fact, the case can also be looked at as a case where the Court simply refused to give protection to features that it regarded as having been dictated by externalities of the situation. [2] Kepner-Tregoe Inc. v. Leadership Software Inc., 29 USPQ2d 1747 (1994). [3] Engineering Dynamics Inc. v. Structural Software Inc., 31 USPQ2d 1641 (1994). Input formats were also found Data structures are the 3rd level of abstraction, only after direct copying of the source code so you are going to have a hard time proving the court thinks of them as facts.

When you dynamically link the only thing you copy is the interfaces and data structures.

That may be the case in the compiled output but at runtime which is the intended outcome that is not the case, statically linked and dynamically linked apps are functionally the same.

So, if I distribute a dynamically linked application without the required libraries the application that I am distributing contains no copyrightable data from the GPL library.

If you ignore the 5th circuit's ruling perhaps but I am sure that will bite you in the ass. However he is not doing that he is distributing the library with his application which effectively makes it part of the same application to the user and likely the court.

In that case how can a copyright license cover something which hasn't violated copyright?

It does violate copyright.

Re:Doesn't matter

[oliverthered]

After reading your abstract from the Fifth Circuit case it looks like a hell of a lot of OSS projects are in deep shit (e.g. wine/ ndiswrapper)

Re:Doesn't matter

[MadEE]

After reading your abstract from the Fifth Circuit case it looks like a hell of a lot of OSS projects are in deep shit (e.g. wine/ ndiswrapper)

That would be the case if the license of the windows stuff or the network card drivers forbade this. I don't think they do.

you fool

[weierstrass]

if a 'server' that is entirely GPL'd with source code available communicates with a closed-source 'client' over a Unix pipe or similar, then the entirety of the code which defines the communication protocol is available to someone wishing to make GPL'd modifications, including to make a new, free 'client'. as your apache example illustrates, the restrictions you suggest would essentially stop GPL'd software from interacting at all with software under an incompatible license. so eg you wouldn't be able to pipe the output of grep to awk, if one came with your nonfree OS, and the other was a GNU utility.

Re:Great.

[bestinshow]

Does the GPL really have to be an infectious virus that transmits it's license to anything that comes near it?

Yes.

The whole point of someone choosing to release their code under the GPL is that they don't want the code to become less free in the future, nor non-free software to utilise the code, or part of the code. That is the right of the developer of the software, and if they want that, they can choose to do that.

If they wish to allow less free code to utilise their code, they can release it under the LGPL.

The issue is the GPL loophole of 'serverizing' GPL code, because then you can claim that you aren't directly linking to the GPL code, you're only communicating with it. Quite clearly when the 'serverizing' is being used in a manner to replace linking it should be treated as linking.

If the 'serverized' software is actually running as a server however, on a remote system, then it becomes more murky. Firstly the server is in-house, not distributed. Secondly this would logically be a valid 'serverizing'.

Re:Great.

[Jessta]

Yes, it does. If the developer wanted the source code to be able to be used in commerical non-free applications then he would have released it under the BSD licence.

Further information

[kripkenstein]

I happen to know Hebrew. Some more information from the Hebrew documents is the matter of money; the Jin programmer is requesting 20,000 NIS (about $4500) as 'damages' (for the violations made so far). This is perhaps an initial offer for settlement. Note how the $4500 is just higher than the $4000 he initially wanted from them (which seemed more than fair to me, personally).

In the lawsuit, it is mentioned that in Israeli law (which I cannot confirm or deny, I have no idea) the minimal fine possible for this type of offense is 10,000 NIS (about $2250).

I can translate other parts of the Hebrew documents if anyone is interested.

Re:Further information

[SashaM]

Umm, actually, that is false. We are asking for a total of 110,000NIS (about $25,000).

Re:Further information

[kripkenstein]

Well, you would know :) But where am I reading it wrong? In the attorney's letter, point #6 says 20,000 NIS; I can't find the 110,000 NIS figure you mention there. IANAL, so please explain - is the 20,000 (for 'statutory damages') just part of the damages you are seeking?

Re:Further information

[SashaM]

That's the initial letter Jonathan sent to them, where he informed them of the violation and offered to quickly settle for a small amount of money. The lawsuit is a different document (No. 4 on the IChessU vs. Jin page).

Re:Further information

[kripkenstein]

Ok, thanks for the clarification. Please clarify something else, as well: it says in the lawsuit that you want 10,000 NIS per violation of copyright; you claim 11 violations, for a total of 110,000 NIS. The 11 counts include - if I read correctly - 8 bitmap images of Jin's interface. Am I right in understanding that you want 11,000 NIS (about $2250) for each screenshot of your software that appears on their site?

Re:Further information

[SashaM]

Am I right in understanding that you want 11,000 NIS (about $2250) for each screenshot of your software that appears on their site?

10,000 NIS per screenshot, not 11,000. I realize it sounds funny asking damages for screenshots, but the Israeli law only allows between 10k and 20k of damages per violation. We are also suing via a legal "fast track" (otherwise it'd take years to get a decision) where we must ask for minimum damages. So, without the screenshot damages, I could be looking at spending the same amount of money on the lawsuit as the potential damages. Plus, after a couple of weeks of futile attempts to explain to Mr. Rabinovich that he is wrong, and him basically telling me to fuck off, you could imagine I'm a bit pissed and want to nail him for everything he has.

Alexander (aka Sasha) Maryanovsky.

Re:Further information

[kripkenstein]

(Oops, I meant to write 10,000, came out 11,000. Ah well.)

Ok, thanks for the clarifications. And yes, I can easily understand how you would be justifiably pissed off here. I wish you luck in getting them to stop their infringing use of your GPLed code.

Re:Further information

[Wumpus]

Sasha,

You might want to talk to your attorney before you post public comments about your case. This is always a good idea when you have ongoing litigation.

Re:Further information

[SashaM]

As I replied to someone else, (in Israel) lawsuits are public information, and I'm not divulging anything here that isn't in the text of the lawsuit.

Alexander (aka Sasha) Maryanovsky.

Re:Further information

[joe_bruin]

Yes, they're public information. But lawyers have an amazing talent for taking words you've said and twisting them against you. The more you leave them to work with, particularly with comments regarding the case, the more you arm them against you.

Better Jin link (I'm the author)

[SashaM]

Although the content is currently the same, the real URL of my Jin website is http://www.jinchess.com (could an editor please fix it - I think it can handle the residual slashdotting). I'll now get back to reading everyone's comments and reply where I can :-)

Re:Better Jin link (I'm the author)

[REggert]

IANAL, but do you really think it's wise to be publicly commenting on ongoing litigation that you are involved in? That seems like it might sabotage your case.

It all depends on the manner of linkage

[Morgaine]

I believe that the FSF have stated on various occasions that while they would *like* GPL virality to extend to all methods of linking, "unfortunately" they can't enforce that for dynamic linkages, notwithstanding what it says in the license.

The reason for this "problem" or "benefit" (depending on your philosophy) is very simple. Dynamic linking is done against an interface, and interfaces cannot be copyrighted because they are the key mechanism for interoperability. There are decades of case law enshrining that principle. (Don't forget that we benefit hugely from interoperability too.)

The GPL is a copyright license (not a EULA), and it cannot claim copyright on the interface to which separate applications are coded. As a result, the GPL has no copyright mechanism for "bridging the gap" between a dynamic library file and an entirely separate application which uses its interfaces. (Otherwise SCO too would have a case that all Linux apps are derived works of Unix, just because they use the same interfaces --- fortunately it's not so.)

And it gets worse (or better). The fact that such an application can be made to work against *ANY* other dynamic library that satisfies the same interfaces reinforces the exclusions for interoperability. Add to that the fact that no COPYrighted material ever actually gets copied, and you see why GPL enforcement is limited to two clear cases:

• code derived by ancestry -- the new source code physically contains elements of GPL'd source code within it, with or without modification -- so that the laws covering derived works can be applied (in this case the GPL license transfers to the new code by inheritance); or

• code inclusion by static linking -- the new source code does not contain any GPL'd source code within it, but the distributed binaries physically contain GPL'd code brought in by the static linkage -- so that the copyright laws which cover copying can be applied (in this case the GPL license transfers by the GPL'd code being physically bound to the new code and distributed along with it).

Both of the above cases are quite clearcut to apply, and hence supposedly safe.

The other major case (dynamic linking, in which no copyrighted source code has been used and no copyrighted binaries are included) is shrouded deep within the darkest clouds of uncertainty, since there is no legal mechanism by which the copyright can transfer to mere client code which is coded to a non-copyrightable interface.

Re:It all depends on the manner of linkage

[Chandon Seldon]

Remember that, in C, you generally include header files in order to use a library (reguardless of dynamic vs. static linking). The header file itself is copyright, so unless you pull some more complex trick you're generally succeptible to the GPL.

Re:It all depends on the manner of linkage

[grylnsmn]

By that argument, you should be paying SCO $699 per CPU to license Linux. Why? Many of their alleged copyright violations involve header files.

Remember, the header file defines the interface, and the interface itself isn't protected by copyright.

Re:It all depends on the manner of linkage

[kabloom]

Notwithstanding what the FSF says about dynamic linking, I'm sure that as cases come up in court, courts will work out a fair solution for dynamic linking that may take in other factors such as how many different libraries implement the same interface (i.e. whether one could plausibly say that their code is not linked specifically against the GPL'ed code.)

Re:It all depends on the manner of linkage

[Anonymous Coward]

Remember that, in C, you generally include header files in order to use a library (reguardless of dynamic vs. static linking). The header file itself is copyright,

Indeed, but compilation doesn't copy the include file, there is no trace of it in the resulting binary. It merely obtains the definition of the interface given in the file (and that interface cannot be copyrighted), so there is no mechanism here by which the license can transfer across.

Notice that this can't even be overcome by adding data definitions into the header alongside the interface declarations, because it would be entirely fair use for anyone to extract the interface declarations out of the header in order to achieve interoperability.

The protections for interoperability are very strong, and that's just as well too, since FOSS benefits from interoperability in countless ways.

Re:It all depends on the manner of linkage

[0xdeadbeef]

Didn't you try to spread this nonsense previously?

There is no meaningful distinction between static and dynamic linking. They are both mechanical processes by which code from one author can execute the code from another author. It is the nature of software that "interfaces" exist at all levels and that the formats by which those "interfaces" are exposed can be transformed by a mechanical process.

And with source code available, it is trivially easy for anyone to turn what is not a dynamic library into a dynamic library.

The only relevence to this discussion is that a dynamic library might be construed to indicate the intentions of its author, but with a license that prohibits unauthorised linking, there is no ambiguity. Whatever loopholes you perceive in the GPL are artifacts of the GPL, and could be remedied by additional clauses.

And even presuming a use license that would allow it, as a developer, your interoperability defense is not a valid defense if you looked at that source code to compose the "interface" which you are calling. Given the extremes taken by companies to conduct lawful reverse engineering, I'd imagine you'd have a tough day in court trying to convince a judge that your work is not derivative.

Re:It all depends on the manner of linkage

[Anonymous Coward]

You're still ignoring the crucial argument: GPL code is being distributed. For that, they need a license, which obviously is supposed to be the GPL. But the GPL places conditions on the redistribution. The interesting bit isn't your definition of derivation, the interesting bit is whether the terms of the GPL allow redistribution of GPL code under the circumstances of the current case.

Re:It all depends on the manner of linkage

[Anonymous Coward]

They are both mechanical processes by which code from one author can execute the code from another author.

Indeed. But code execution has nothing to do with copyright. Only copying the code engages copyright.

It's like driving a car. You can drive it to your heart's content without any implications on copyright law. You can even build a company based on it, because the company is merely acquiring and using the car. (Aggregation and execution.)

But if you tried to COPY the car in any shape or form (even if you merely built your company HQ to look like one) then the copyright lawyers would be down on you like a ton of bricks.

The GPL is a copyright license. It only works through copyright. And for that to happen, something has to be copied. In the case of dynamic linking, nothing is copied, so any appeal to copyright fails at the first hurdle.

Usage or execution is quite irrelevant to copyright.

Re:It all depends on the manner of linkage

[Anonymous Coward]

It is the nature of software that "interfaces" exist at all levels and that the formats by which those "interfaces" are exposed can be transformed by a mechanical process.

You seem to think that the existence of a mechanical transformation process makes everything equal to everything else, and specifically to static linking. Well, the FSF doesn't agree:

By contrast, pipes, sockets and command-line arguments are communication mechanisms normally used between two separate programs. So when they are used for communication, the modules normally are separate programs.

(My colleague here actually suggested that you're a BSD fanatic, and that you're making ridiculous claims about the GPL far beyond what the FSF is claiming purely to discredit it. That would be pretty sad. :P)

I suggest that you read the FSF's documents before trying to create your own interpretation --- especially the words of Eben Moglen, who understands the limits of copyright even if you don't.

Re:It all depends on the manner of linkage

[0xdeadbeef]

And of course I respond to your pedantic absurdity that by invoking dynamic linking, a copy is made by loading the software from disk.

But that is all beside the point. In the absence of some equivalent library with an alternate license, the infringing developer had to abide by the GPL in order to develop his software. The redistribution rights afforded to users do not apply to him and his derivative work.

Re:It all depends on the manner of linkage

[0xdeadbeef]

(My colleague here actually suggested that you're a BSD fanatic, and that you're making ridiculous claims about the GPL far beyond what the FSF is claiming purely to discredit it. That would be pretty sad. :P)

The only thing I despise more than the absolutist free software hippies are the people who wish too exploit their goodwill and intellectual property rights under some pretense of being "more free than thou".

And if they wish to make a distinction between linking and a serial communication channel in their license, that is their perrogative, but they do not wish to make that distinction between static and dynamic linking, and given how trivially that transformation can be accomplished with the source code available, the difference is moot. (And expanding the claim does not invalidate my original argument. Static and dynamic linking are different in the context of distribution, but not in the context of the developer abiding by the license when making use of the software.)

The logical conclusion of all these anti-GPL arguments is that mere knowledge of an interface allows one to call that interface, regardless of the ownership of the code being called, the license it was sold/distributed under, or how knowledge of that interface was derived. This would require that every component vendor enforce a rule in their license that their development users enforce a rule in their license that end users may not use software that executes the third party libraries that haven't officially licensed the use of those libraries. Maybe some of them do, but it is a ridiculous burden and legalistic absurdity.

Re:It all depends on the manner of linkage

[Chandon Seldon]

That's sketchy. I guess you're probably right - assuming that there is nothing in the header file other than minimal function prototypes. If it defines macros then you're probably a derivitive work.

Re:It all depends on the manner of linkage

[Anonymous Coward]

The logical conclusion of all these anti-GPL arguments is that mere knowledge of an interface allows one to call that interface

Correct.

And it's not an anti-GPL argument either. It's the very foundation on which Free Software manages to interoperate with the rest of the world. You're in danger of cutting off your nose to spite your face.

Re:It all depends on the manner of linkage

[0xdeadbeef]

And it's not an anti-GPL argument either. It's the very foundation on which Free Software manages to interoperate with the rest of the world. You're in danger of cutting off your nose to spite your face.

Moi? I'm a component vendor, not an author of free software.

And to that I say, God bless DRM and God bless the DMCA.

Re:It all depends on the manner of linkage

[Anonymous Coward]

You're still ignoring the crucial argument: GPL code is being distributed. For that, they need a license, which obviously is supposed to be the GPL.

Of course, but that simple case is not under discussion. If GPL code is being distributed then clearly its license comes into play. There is no uncertainty there at all.

The issue of dynamic linking is different though, since GPL code is either NOT being distributed at all, or else it's only being distributed as an aggregate, which the GPL expressly disclaims as a taint vector for GPL copyright.

Personally, I think aggregation is not really a safe defence for ISVs and VARs who want to go that route, as it's far too easy for different judges to make different interpretations.

But total physical separation from and non-distribution of GPL code *must* be safe, because the ramifications of a counter judgement on that would pretty much freeze the entire computer industry into immobility, since all interfaces everywhere would become vectors for copyright infection. Indeed, pretty much *all* licenses would become viral. SCO would love it though. :-(

Re:It all depends on the manner of linkage

[Anonymous Coward]

But total physical separation from and non-distribution of GPL code *must* be safe

I agree, but the typical case is that someone tries to play the wrapper game (look ma, no derivative work) and then still distributes all parts together. A program where half the required code isn't in the package isn't much of a product after all. The explanation in the license FAQ (linking is deriving, everything else is aggregation) is simplistic and I doubt that it will be the stance that courts will take. Instead, if a program is clearly designed to be used with GPL code and is distributed together with that code, there's a good chance that it will be seen as a derived work.

But not a GPL violation

[gr8_phk]

Only part of the client source code is downloadable, not the whole; this is a violation of the GPL. Also, the iChessU has an EULA which violates the GPL by placing new restrictions on how the code may be used.

Lets just remember that you are supposed to claim copyright violation in court, not GPL violation. When the defendant says they have permission to redistribute your code, you ask where they got permission and let them point to the GPL. Then you point out how they have not lived up to their end of the deal and point to the termination clause. IANAL, but if you claim GPL violation, wouldn't the court just tell them to stop? If you claim copyright infringement, clearly there are laws and penalties (money) involved right? Of course this is not in the US so DMCA type (outrageous) penalties don't apply.

Interoperability is protected

[Anonymous Coward]

their intention is plainly to try to circumvent copyright law through technicalities, which the judge is unlikely to approve of.

You can perfectly legally "circumvent copyright law" by not copying the copyrighted item in the first place. And that's exactly what they did, they created an interface mechanism which avoids COPYing the COPYrighted work. (Please note that copyright is about copying, not using.)

Interfaces and specifications and standards are a heavily protected area of exclusion for copyrights, because they allow interoperability. If it were not so, then you wouldn't be able to talk to your Windows printer from Linux, because accessing the printer interface would force Microsoft's licensing conditions upon you. Fortunately it's not so.

And that's the case here too. Those guys are merely interfacing with GPL'd code, and this can't bring the wrath of the copyright gods down upon them because there is no copying involved. Usage is not copying, entirely different things.

Re:Great.

[Anonymous Coward]

Yes, it does. If the developer wanted the source code to be able to be used in commerical non-free applications then he would have released it under the BSD licence.

Or in the GNU Lesser GPL (LGPL)

Not pipes or sockets ... merely source

[Gopal.V]

I've been watching the kororaa project for a while, ever since we did one tandem session with XGL and OS X on two machines, watching XGL rule - especially in the video across cube faces demo. But a few weeks later, the developer announced that he's stopping Kororaa because of GPL issues with properietary drivers. And here's a reply by the FSF.

Now, the point to note here is that GPL is redistribution license. The way the nVidia folks handle it is to give the user some code, a binary blob and effectively tell them to build it themselves. The code they distribute does not link to the Linux kernel *yet*, while the binary blob is the closed source bit. Now, what the user does is to link them all up and there you go - which is not the distributor's fault. And this works because they are not redistributing any code that is copyrighted by a Linux kernel author (for example laf0rge).

The whole model makes the user violate GPL in principle, while the distributor (i.e nVidia) is in the gray area of legality. This is of course, my understanding from following up all this (and then had an argument with a paralegal @ work about GPL).

But I could be wrong you know ...

Re:Not pipes or sockets ... merely source

[Hellkitten]

The whole model makes the user violate GPL in principle

Actually, since the end user doesn't distribute anything there is no violation there. So the end user is still in the white. Depending on the interpretation of the GPL Nvidia may be in a gray/black area.

Re:Not pipes or sockets ... merely source

[bug1]

Actually, since the end user doesn't distribute anything there is no violation there. So the end user is still in the white.

You should have said there is no violation so far, there is a reasonable chance that a user will distribute the software when they are disposess themselves of the machine (not everyone wipes their drives clean as they should).

The GPL will probably still be violated, just that the user is put in a position where they are unaware that they are doing it by unsrupulous hardware companies.

In Other News

[Anonymous Coward]

An elite haxxor in Mexico has completely cracked Apple's latest DRM system, and has written a .MP4 to .WAV converter. Slashdot community rejoices at this victory over burdensome copyright restrictions!

Latin == Pons asinorum

[megaditto]

Omnia dickta fortiora si dickta latina, motherfucker

Let IChessU know how you feel

[Anonymous Coward]

They have a feedback page.

http://www.ichessu.com/Feedback.aspx?type=help

I put my comments in there.

Case rules can handle much of law

[Morgaine]

>> Why not some formal law language with clear semantics and syntax?

Several people have written objections to youe suggestion here, but computer-based solutions are easy to find if you use a reductionist approach and don't try to do *everything* mechanically.

The simplest and most powerful approach by far is to apply formal semantics by case, not by generic mathemetics: For each area of legal engagement, a large number of highly specific case rules are defined and given a priority to define the order in which they are applied, with each one capable of matching only one very narrow situation. Separately from this you define a set of legal outcomes (independent of the case rules!!!), and causal links are only then drawn from the case rules to the case outcomes.

This can be made as deterministic a system as you like, merely by narrowing the scope of each case rule. Needless to say, if no case rule matches then the case pops out of the bottom to be handled by a human.

This addresses the ambiguity problem by each case rule being extremely specific, and allows lawyers to enter the loop for unmatched cases given by hard or undefined problems. It addresses comprehensibility by the public by allowing a worm's eye of the problem, one case rule at a time, and by allowing your own computer to "scan the law", which is otherwise impossible unless you practice law yourself.

Of course, anyone who has worked in AI or expert systems will recognize that the above is just a logic system, only expressed in a way that doesn't blow the mind. Well, that's good. :P

There's always a way, if you want to find one. The trouble is, such a mechanism would require endorsement from the legal profession ... and that of course would not be in their interests.

I hope that's just a figure of speech

[roystgnr]

the author of the GPL licensed Jin Chess Client is taking IchessU to court for violations of the GPL license.

It's not illegal to violate the terms of a contract you never agreed to. If you take someone to court for violating your software license, then (unless you're unlucky enough to live in a jurisdiction that takes EULAs seriously) you'd better have their signature, a record of their online agreement, or something to prove that they agreed to your license.

If you don't have any of those things and you feel like your software's GPL has been violated, then you take the violators to court for copyright infringement, not license violations. Let them bring up "but I had a license whose terms I didn't accept" as a defense if they dare.

Re:I hope that's just a figure of speech

[Wyzard]

The GPL is a copyright license, not a contract.

Section 5 of the GPL says:

You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it.

If IChessU claims they didn't agree to the GPL, they're still in violation of the Jin copyright.

Contracts are only needed when you want to impose additional restrictions beyond those imposed by copyright law. That's why you have to click "I Accept" to EULAs; those are contracts. The GPL is not.

Re:I hope that's just a figure of speech

[Anonymous Coward]

The GPL is not a contract. It is a licence. This has been debated ad nauseam on Groklaw.

Cheers,

AC

You forget the FA

[Anonymous Coward]

The article is about someone taking someone else's GPL code and using it as a distributor, not an end user.

common when a trail of argument/counter reaches a few replies deep.

Re:You forget the FA

[Tweekster]

and the poster before was correct, under standard copyright law this person would STILL be wrong because under that law he has no right to take the code and merge it either.

The GPL gets struck down, this guy is still breaking copyright law then.

I don't think that's right...

[TrekkieGod]

That's not right...intentions have nothing to do with it.

You can license your code any way you want. If you want to license part of it as gpl, write a dll that is meant to work with your program only, but license it in a proprietary form, than release that stuff together with only the source code for the part that talks to the dll, you're fine. RMS would probably show up on your door holding a bat, but his lawyer won't be with him.

The license limitations don't limit what your gpl code can do, they limit what people can do with gpl code. So, if someone grabs the source code to a dll, and dynamically links to that with their non-gpl-compatible-licensed program, they are in violation of the gpl license for the dll, which doesn't allow that. However, if you're writing gpl code, you can link it to whatever you want, as long as the other license allows that. If the other license says you can't link it to gpl programs, you're screwed, but API's usually allow you to link to them, because they'd be useless if they didn't.

nVidia clarification

[ajs318]

First point, the GPL does not forbid anything. It is a licence -- it only gives you permission to do something you would not ordinarily be allowed to do.

The Law of the Land gives you only very limited permission to copy a program (see "fair use" or "fair dealing"), and would ordinarily take a dim view of you distributing copies. The GPL gives you permission above and beyond your fair dealing rights (which are determined by the courts) to do certain acts dependent on certain conditions. One of the permissions granted by the GPL is to make and distribute copies, gratis or for money, as long as you ensure that everyone who receives a copy of a program from you receives the same permissions as you (hence the requirement at least to mention the GPL), and can exercise them meaningfully {hence the requirement to make source code available).

The nVidia drivers are supplied as a closed-source binary blob, which is distributed under a restrictive licence; and a GPL wrapper which implements an interface between the kernel and the blob. The addresses of things in the Linux kernel are subject to change from one version to the next; but you can find out what is where from files that are generated during the compilation of the kernel. The addresses of things in the nVidia binary blob are also subject to change from one version to the next; but only nVidia know what is where. So nVidia provide the source code for a wrapper that interfaces between the kernel {where addresses of functions can be determined from system files} and the binary driver {where addresses of functions are hard-coded into the wrapper}. nVidia can legally do this because they own the copyright on both bits: the Law of the Land does not prevent a copyright holder from making derivative works. The GPL grants you permission to enjoy, study, share and adapt the wrapper. In order for it actually to be any use, however, you need the binary blob (which, being pure x86 machine code and talking only to graphics hardware, does not depend on a particular OS). The combination of binary blob and GPL wrapper constitute a derived work. You are right that the GPL does not grant you permission to make such a derived work. However, that's irrelevant. The Law of the Land grants you permission to make this derived work, because that is a necessary step in making use of something you legally own -- and therefore constitutes fair dealing.

Once you have combined code covered by two different licences, you are bound by both licences. The GPL demands that you at least offer to distribute source code if you distribute anything, whereas the nVidia licence does not grant you access to the source code and would not let you distrubute it even if you had it. So, unless you live in some jurisdiction where the Law of the Land grants you access to the source code, the only way to satisfy both the GPL and nVidia licences is not to distribute the code produced by combining the GPL wrapper with the closed-source blob.

For what it's worth, I think nVidia are scumsuckers and I hope their binary blobs are reverse-engineered. But the law seems to be on their side.

Will it be in Israel?

[CaptSolo]

It is not clear whether this GPL "test" will really happen in Israel.

Added at the bottom of that webpage: "Update (29.08.2006): Alexander told me yesterday that he has relocated to the US, although I don't know whether this is a consequence of me filing the lawsuit (or whether it's at all true)."

Clearly it is hard to say from that alone if this person has really relocated to the US or not. But what happens if he has? Will he be able to evade this lawsuit this way?

Re:Will it be in Israel?

[Anonymous Coward]

He could assign copyright to the FSF and let them prosecute him in the US!

Re:Heebs? Pshaw!

[Anonymous Coward]

Cheeses, trust a fucking jew to try and get more value from something that's free!

Just like a fucking redneck to accuse the jew.
why don't you just crawl back into your cabin in the woods and go back to making bombs or something, you freak. This issue has nothing to do with Jews and Judaism. It is a legal dispute, which anyone is entitled to have. Stop being a freak and join us in the 21st century will you?

100% wrong about reverse engineering i/f's

[Anonymous Coward]

your interoperability defense is not a valid defense if you looked at that source code to compose the "interface" which you are calling

Quite the opposite. Reverse engineering for the purposes of interoperability is extremely widely protected against various types of IP-related claims (including copyright), not just in the US but worldwide.

If it were not so, Linux would work with almost no hardware at all --- the proportion of manufacturers who have supplied interfacing details to Linux driver developers is quite pitiful.

Also reverse engineering of interfaces for interoperability certainly doesn't make you fall under any licensing conditions, otherwise Linux would now be under legal restraint from several thousand proprietary licenses.

Clean-room implementations are required when developing competing applications to avoid claims of copyright infringement, but that's only because the *code* of applications is covered by copyright, and that is under danger of being alleged to have been copied.

Interfaces can NEVER be copyrighted, so there can be no claims of copyright infringement, and thus clean rooms are not required for discovering them.

Re:100% wrong about reverse engineering i/f's

[0xdeadbeef]

Gah, you got me. Trade secret protection and provisions in the license against reverse engineering don't make any sense when the source is available to anyone.

But presuming the definition of derivative work can include additions to the software, and not simply textual changes to the source code, linking is derivation.

Otherwise, all GPL'ed software is reducible to an interface that can be called by non-free software. Glad I'm not one of those free software suckers!

How I would fight the GPL

[Sloppy]

If I wanted to write a program that integrated with GPLed software but did not want release my code under GPL, here is what I would do.

My strategy, instead of looking for loopholes in GPL's terms or figure out what is permitted, would be to completely avoid getting bound by the license. If I do not accept their license, then no lawyer is going to be able to make arguments about whether I complied or not. Don't fight on their terms. Don't let any of the text in their license be admissible or relevant.

The key to doing this, is to forget that the code you want to integrate with, is GPLed. Forget you have any extra rights that you might exploit. Assume that the code you're depending on has the tightest, most restrictive and hostile license that exists. Handle integration with a GPLed chess program, the exact same way that you would handle integration with Microsoft Excel. Ask for no favors other than whatever is allowed by copyright law.

To do that, the two major hurdles are:

• My code should not be a derivative work.
• My code's distribution package should not also contain the GPLed code. Don't distribute their code at all.

If you can accomplish those two things, you avoid the license, and therefore it doesn't matter what kind of license it is, and it doesn't matter whether or not the terms of that license would permit whatever you're doing.

I'm not 100% sure what the legal definition of derivative work is, but I do know that you find out by asking your own lawyer instead of FSF's lawyer. As soon as someone at FSF starts talking about what they permit, you know they're talking about their license instead of copyright law. And you need to concentrate on copyright law.

One thing you can infer from the market overall, is that calling someone else's program does not make yours a derivative work. Even if you call the other work in a way that your program 100% totally depends on it and has zero worth without the other work, that doesn't make your program a derivative work. How do we know/infer this? From the proprietary software market. In spite of various projects like WINE, it's pretty safe to say that no Windows applications have any use outside of MS Windows. They simply don't run. If WordPerfect for Windows were a derivative work of MS Windows, don't you think Microsoft would have sued WordPerfect in order to squash the competition? Of course they would have. So don't worry about calling APIs.

And it doesn't matter how public the API is. Especially in the DOS days, there were all sorts of programs that did very intimate things to MSDOS's internal structures (e.g. disk managers, many many TSRs, etc) and nowdays on MS Windows you see some of this with the AV products. But surely McAfee's AV program is not a derived work of MS Windows. So don't worry about calling or the intimacy of calling. To avoid being a derivative work, just make sure you don't use any of the other guy's code.

Packaging. This is the tricky one. It's hard to sell a program that doesn't work on its own. The amusing thing here, is that with Linux, there are all sorts of packaging systems provided by the distributions, such as apt-get or emerge, that will solve this for you. If I wanted to write a proprietary program for Gentoo that depended on some GPLed stuff, it would be really easy to to just ship an ebuild that tells portage about the dependency, and it would get installed automatically when the user wanted to install my program. Then I wouldn't have to ship any of their stuff.

MS Windows (and MacOS AFAIK) doesn't have any sort of automatic generic get-and-install-all-the-dependencies program. (That makes it actually harder to defeat the GPL on Windows than on Linux.) Of course, you can probably just ship your Windows app with an installer than downloads and installs the dependencies, but support and maintenance would be tricky. And it would be

Re:How I would fight the GPL

[nietsch]

The test for a derivative work is very simple. Take out the GPL-ed program. Does the non-gpl program still function in a meaningfull way? Could you use another program instead of the GPL one? If so, then you have a non derivative work. If not, it clearly relies on the GPL program and cannot be distributed without violating copyright/GPL. It is not the method it uses, it is the reliance that matters.
Offcoourse, take this non-legal advise with a spoonfull of salt IANAL!

Re:How I would fight the GPL

[YoJ]

The situation is not as dire as that: under the GPL you are allowed to distribute the GPL program along with its source code. The only reason this could conceivably be a problem is that it biases an outside observer into thinking that your product is a derivative work. But objectively, whether you distribute GPL code along with its source doesn't change whether your other product is a derivative work or not.

Re:How I would fight the GPL

[spitzak]

The problem with your argument is that your counter-example of the derived works requirement is invalid.

Microsoft explicitly gives permission for you to install and run other software on the Windows system. This exception is actually worded explicitly into the copyright given to OEM's who copy the system. In fact part of the anti-trust argument is that Microsoft was failing to give sufficient freedom in what extra software was added to Windows by OEM's.

True there probably is no such wording in anything given to home users, but they own the system and can do anything they want to it, whether Microsoft wants them to or not. And they cannot distribute the result because they would be violating the copyright on Windows itself, so it is quite irrelevant if their redistribution includes some added code of their own.

Re:How I would fight the GPL

[Anonymous Coward]

By this logic, reiserfs tools for windows is a windows derivative work since it will not function in a meaningful way without it. Where is my mistake?

Re:How I would fight the GPL

[Ibn al-Hazardous]

Utter bullcrap!

Have you seen any game mods or Excell macro compilations sold independently?

No? That's because they are derivative works. They can be shared for free over the net, but not sold commercially (or at least - noone has ever dared).

Counterstrike is the perfect example here. A lot of people drew maps that were included in this mod also when it was sold commercially, but noone except the main author got any money. It was a derivative work that sold Half-Life (and the maps were derivative works of a derivative), but could never have sold independently. The lawyers would have had so much fun in court if the fans had tried!

So, why should the GPL be treated different from any other license?

Re:How I would fight the GPL

[Anonymous Coward]

Your examples are very weak...

What about all plugins that are sold for programs like photoshop and illustrator? Aren't they dependent on the main application and yet they still could cost quite a lot.

Re:How I would fight the GPL

[Sloppy]

The test for a derivative work is very simple. Take out the GPL-ed program. Does the non-gpl program still function in a meaningfull way? Could you use another program instead of the GPL one? If so, then you have a non derivative work. If not, it clearly relies on the GPL program and cannot be distributed without violating copyright/GPL.

So let me get this straight. I have Mozilla compiled for Microsoft Windows. I take away Windows. Does Mozilla still work? No. Can I replace Windows with something else? No. Therefore, Mozilla compiled for MS Windows is a derivative work of Microsoft Windows and cannot be distributed without Microsoft's permission.

Well, so much for the "browser wars."

Perhaps the test for being a derivative work is a bit more complex than "clearly relies on." If I'm wrong, then all Windows applications are Windows derived works, all MacOS apps are MacOS-derived works, etc. If that's what you really mean, ok, but I don't think anyone will accept it.

Re:How I would fight the GPL

[Sloppy]

So, why should the GPL be treated different from any other license?

It shouldn't. And my whole point was to look at the problem in terms of copyright instead of licenses. You might want to read the post that you replied to.

Have you seen any game mods or Excell macro compilations sold independently?

No (but I don't follow those scenes, so I don't know anything about them). But I have seen non-portable interpreted BASIC programs for sale, which only work with one BASIC interpreter on the market. If I write a VIC-20 program in BASIC, is my program a derivative work of Microsoft's BASIC interpreter? I don't see how Excel macros are any different that interpreted programs.

It's interesting that you think these things are derivative works of Counterstrike or Excel, but you also think "they can be shared for free over the net." I assure you that I could make a derivative work of Counterstrike or Excel, share it for free over the net, and get sued the next day. ;-)

Re:How I would fight the GPL

[Anonymous Coward]

And my point was that you claimed that it is ok to sell a piece of software, that piggybacks another piece of software - as long as it doesn't include that software. But I can claim the moon is made of cheese. Where's the proof?

Ok, so you say programs that use interpreters is the perfect example, and I say bullshit again. Interepreters are made for interpreting, and make allowances for commercial use in the licenses. Excell does not, and neither does GPL software.

Oh, and I think Counterstrike is a derivative work of HalfLife. The maps in Counterstrike are a derivative of Counterstrike, and though CS was sold commercially eventually, the copyright holders didn't get a penny.

Sure, you could make derivative works, share them for free over the net, and get sued the next day. But I didn't claim otherwise. I claimed you can't make a derivative work, sell it commercially (even without the original, but needing the original to run) without geting sued - unless the work from which you derivate expressly allows you to do so in its license.

You make claims, about my misunderstandings and about law, but as for my understanding you prove yourself wrong. Now, can you present some evidince as to the law? Without speculating? Can you even provide one example which doesn't sell entirely through the idea that others are to sell their derivative works (such as interpreters, compilers, OSes and computer firmware)?

Re:How I would fight the GPL

[nietsch]

Your mistake was reading non-gpl program as OS.

$4000, Lawyers, and Slade

[Schraegstrichpunkt]

From the article:

Eventually, Alexander told me that $4000 was too expensive for them and asked to meet me and discuss alternatives.

and then

A few days later, Alexander told me that he has consulted with his lawyers

I call bullshit. Unless lawyers in Israel are vastly cheaper than they are in North America, nobody would pay lawyers, and then take legal risks (which could involve paying the lawyers some more) in order to save only $4000.

This "Alexander" character sounds a lot like Slade from QuakeLives.

Insightful? I think not.

[49152]

>IMHO, that means Windows API = ok, Java API = ok, .NET API = ok, own API or library = not okay.

Bullshit.

It does not matter if Microsoft, Sun, you or the neighbours cat wrote the library/API you link with. Any GPL program can link with any proprietary library, it is the reverse that is a problem.

>And that makes sense. Otherwise I could just build all of my app in my "MainApp API" and GPL my "StartMainApp()" function call... :)

There is absolutely nothing in the GPL license that would prevent you from doing so. I agree it would be a bit stupid, but that does not make it an copyright infringement (which is what an GPL violation is).

What you cannot do is write a proprietary program or libary and link this with an GPL program or library. Example: This is essentially what someone once did with VirualDub, they wrote their own close-sourced video editing program but instead of doing the hard work themself they just linked with the relevant bits of VirtualDub. It was in essence no more than a new GUI for VirtualDub and a clear violation of the GPL license.

Regarding this chess client in the article I'm not so sure since it is a bit murky on the actual details. But it sounds like someone rewrote the chess client to include audio/video chatting capabillities using a third party a/v library and now they only want to give source for the modified chess client and not the a/v library. This is most likely NOT an GPL violation but as I said the parties are a bit unclear on all the details.

Re:Insightful? I think not.

[martin-boundary]

It does not matter if Microsoft, Sun, you or the neighbours cat wrote the library/API you link with. Any GPL program can link with any proprietary library, it is the reverse that is a problem.

Not actually true. You can't link with Microsoft, Sun or anyone any way you like, either. If you own a proprietary compiler, go look up the EULA. All commercial compilers have a legal section specifically allowing you to link with a specific set of libraries they supply. If you look closely, you'll find an actual list of file names that are allowed, and only those are allowed, no others.

So you can't link to Microsoft because you feel like it. You link to Microsoft because they specifically allow you to do so in some cases, and have made it easy enough that you don't notice what you're doing.

The problem here...

[raehl]

... is that you've just established that the amount of damages you're seeking is based not on actual damage done to you, or severity of the offense, but instead on how much damages you needed to claim in order to get your lawsuit fastracked.

Are you not Communist? For what you need money?

[Anonymous Coward]

Are you not Communist? For what you need money? Let swine do what he wants and you go on and make better program.

Re:Further information - Hebrew translation

[nanday]

If you are willing to do some translation, could you contact me at editors@ostg.com?

Thanks,

nanday

GPL: License to Sue

[Brandybuck]

GPL: License to sue.

I got another email today about my choice of the BSD license for software that I wrote. I was warned yet again that I am being foolish, that BSD is a "license to steal." Where the fsck do these buttinskis come from? I was told that the GPL would allow me to sue people who violated my license. Do these guys even know what freedom means?

GPL: Because freedom is about suing to get your way.

Re:GPL: License to Sue

[iggymanz]

bullshit, the bsd license also can be enforced by suing, it too is based on copyright law with the same mechanisms protecting it from violation.

Re:GPL: License to Sue

[Brandybuck]

Except that there's nothing to enforce with the BSD license! The software is essentially given away, with the license being nothing more than a spot to attach a warranty disclaimer. But the GPL has restrictions specifically designed to regulate the behavior of others. It isn't given away because it has legal strings attached.

Of all the Free and Open Source licenses out there, only the GPL has a dedicated legal team devoted to hunting down license violators. The FSF is Mini-Me to BSA's Doctor Evil.

Court is not US

[from_downunder]

My (somewhat hazy) understanding is that the 5th Circuit is a US court. This case is being held (I beleive) in Israel. While I beleive that courts will often look at other countries determinations, I dont believe that this is compulsory. Nothing in the above however, precludes the possibility of Israel following a similar legal decision.

Re:Court is not US

[MadEE]

We aren't talking about this case we are talking about GPL in general and in particular the US legal system. I am not aware how Israel handles abstraction filtration comparison, etc. Granted in a contract case because the license is an American one the Israeli court should look at American law and is likely to differ from that only in the case that Israeli law forbids it.