Slashdot Mirror
Abusing the GPL?
"How, you may ask?
Integrate the highly useful GPL code we're eyeing into our only slightly more complex (but much more lucrative) project, thereby saving us at least 30% of the coding involved. The company then go all the way to production with it, but instead of finally compiling the actual project for distribution, they instead compile a bunch of incomprehensible gobbledygook that just happens to compile to the same bytecode. You know the game: globally replace every function name, variable name, and so on from our code with nonsensical names (or random characters), remove all of the comments, and any other form of obfuscation they can introduce. They will then GPL the obfuscated gobbledygook, which isn't much more useful to anyone than reverse-engineered bytecode would be (it is a complex project). 'Voila!' All the benefits of a huge GPL project and countless thousands of volunteer hours and unreadable, incomprehensible source tree.
For the record: I do not think this is right yet, I have not been able to find any precedent for why the GPL should protect against this kind of abuse.
I'm not trying to snitch on my company -- or lose my job, which is why I am posting anonymously -- but hopefully some lawyers out there could point out some iron-clad legal reason preventing this sort of thing. I've read the GPL through at least a dozen times since yesterday, and so far it looks like our lawyer is right. I have not found any relevant linkage either, as I have mentioned. Links to extended legal analyses of the GPL from a technical standpoint (if any exist) would be the most helpful. All help is appreciated."
If you take some code and switch out all the variable names and change the spacing around, it's still the same code. If your lawyer is advising you differently, I'd be very suspicious of his motiviations.
"We returned the General to El Salvador, or maybe Guatemala, it's difficult to tell from 10,000 feet"
Nowhere does it say that that code has to be non-obfuscated. Nor do I think it should. Do we really want to try and formalize that gray area between "obfuscation" and just plain "sloppy code?"
Not all of the code released under the GPL is what we would consider "good code." By that, I mean people release all sorts of toy projects and junk code under the GPL, for learning purposes. They use bad variable names and inefficient algorithms, but when do we start to consider code "obfuscated?" And more importantly, do we want to leave it to a lawyer to make that decision for us?
I say if you're really concerned about it, then leave the company. Otherwise, just write it off as mean-spirited. There's no law against being mean. :(
Like woodworking? Build your own picture frames.
Pretty boring stuff, but the overall point is that once the end product is GPL'd, it won't take long for someone in the bazaar to figure out a meaning for "asdfgh", and do a s/asdfgh/meaningfulName/g through the whole thing. Or even figure a way to diff it with the original source.
As long as it's GPL'd, the source will be available, and it'll be figured. You're wasting a lot of your time (and the rest of the community's) for very little reason.
No matter how complex your obfuscation, it's likely much less complex than, say, CSS or DES was.
jer
We may be human, but we're still animals
- Steve Vai
Although logicaly it doesn't sound like a violation of the GPL because you still can see the source code. I question the motivation. How would this benifit your company? The source will still compile right? It still can be obtained free. right? This just seems silly. The problem people have making money off of GPL'ed software lies not in the open source code but in the fact that people can get for free what you are trying to sell.
I mean when was the last time you looked at the source of a project that you just wanted to use, not develope.
From my reading, that is not the problem. It appeared that the company did release the code with source as GPL along with their product. They just obfuscated it before releasing it. That is not directly a GPL violation.
There have been cases before of obfuscated GPL code (Some video drivers in the Linux Kernel I believe) but those were original source from the manufacturer.
This article is about taking someone elses GPL code, obfusacting it, then re-releasing it with GPL intact.
There is nothing so silly as other peoples traditions, and nothing so sacred as our own.
I don't see how this violates the spirit of the GPL, since there are no provisions in it for the quality or readibility of code.
.asm file that is just the disassembly of your binary isn't very useful for preserving the right to modify the program. Neither is deliberately and cleverly obfuscated source.
The "spirit" of the GPL is about being able to make modifications to the code. That is one of the rights that the GPL is trying to preserve. It isn't just about being able to get a free copy of the code you can compile (and if you're lucky for different platforms).
As at least a dozen other posts under this article have already said, there is language in the GPL providing for quality -- or at least editability. The source must be in the "preferred form" for editing. Because releasing a
The authors of the GPL understood that "openess" depended on at least the level of usability that was present when the code was written. Hopefully we've cleared this up (and this guy's company lawyer has been sacked).
The enemies of Democracy are
IANAL, etc... etc... yadda, yadda, yadda.
The company then go all the way to production with it, but instead of finally compiling the actual project for distribution, they instead compile a bunch of incomprehensible gobbledygook that just happens to compile to the same bytecode.
You know the game: globally replace every function name, variable name, and so on from our code with nonsensical names (or random characters), remove all of the comments, and any other form of obfuscation they can introduce.
They will then GPL the obfuscated gobbledygook, which isn't much more useful to anyone than reverse-engineered bytecode would be (it is a complex project). 'Voila!' All the benefits of a huge GPL project and countless thousands of volunteer hours and unreadable, incomprehensible source tree.
Here is my take:
Other things to take into account:
Conclusion?
Bad idea. VERY bad idea. Release code under GPL, play nice, and nobody gets hurt... (wink! wink!)
IMHO, any company who tries that kind of stunt is going to end up on the trash-pit of dot-coms faster than you can say "GNU General Public License".
The right to offend is far more important than the right not to be offended. (Rowan Atkinson)
Yes, it does:
'The source code for a work means the preferred form of the work for making modifications to it.'
Incomprehensible gobbledygook does not the preferred form make, any more than machine code.
What a lot of people appear to miss a lot of the time is that the GPL is _not_ one of those 'thrown together in a week' opensource licenses. It was developed over several years, and reviewed and rereviewed by the FSF legal counsel. It doesnt have holes like this.
Newbie lawyers looking at it for a few hours always misinterpret it. They dont have the technical savvy, nor the persistence to grasp the actual meaning and how thorough the GPL actually is when it comes to accomplishing its task.
The current MySQL AB/Nusphere legal issue isnt the first court case on the GPL because nobody has tried to violate the GPL before. It's because everyone else has realized they dont have a chance in court, and have given up rather than trying to persue a case which their lawyers have eventually realized they will lose.
Go to a public library. Logon to the internet terminal found in most public libraries. Create a Hotmail or Yahoo Mail account. Use that account to E-Mail the FSF with your company's name, the project's name, and as many details about the project as you can without personally identifying yourself.
Then, later, you can sleep like a baby, knowing you did the right thing.
The next Slashdot story will be ready soon, but subscribers can beat the rush and slashdot the links early!
First of all, IAAL. Second, the GPL's definitional distinction between source and object/executable form relies on two key terms that cannot be objectively measured: "preferred" and "normally". I defy you to provide me with objective metrics for measuring what is "normally distributed...with the major components...of the operating system on which the executable runs." Equally imnpossible is a definitive response to the question "what is the preferred form of the work for making modifications to it?"
In order to impart meaning to the GPL distinction between source vs. object/executable, one must go on a fact-finding parade to measure industry practice, and other wishy-washy standards. In the context of a dispute over a GPL'd bit of code, you can be damn sure that the GPL will collapse under the weight of this fact-finding process, and that the party with more patience and money will win that battle.
There are some things that lawyers understand better than geeks, believe it or not. We are (generally) excellent at spotting weakness in prospective arguments. In the case of the GPL, there are drafting holes big enough to drive a Trident submarine through. I've said it before, and I'll say it again: the GPL won't hold water in a dispute. The reason no one has given you any precedent (as per your request) is that the GPL has not been truly tested in court. Since the GPL eschews the lessons that lawyers have learned about drafting in the past (largely in order to score points with geeks by being colloquial in manner and sounding un-lawerly), it cripples itself with imprecision and ambiguities. The weakness in its core definition of source vs. object/executable is merely one of many fatal flaws in the document. To be perfectly frank, the GPL is a POS contract and I would arguably be liable for malpractice if I advised a client to use it for reason other than their unbending adherence to open source dogma.
In conclusion, you are likely to see many companies "abusing" the GPL. Rather than use the loaded term "abusing", I would prefer to characterize this behavior as "exploiting" the unsophisticated and niave drafting of the GPL's language.
Since I said "IAAL", I must also say that the above does not represent a formal legal opinion, that I do not represent you (the reader) as your lawyer, and that you should not treat this message as my legal advice to you. Laugh all you want -- I'm just sticking to my ethical directives, kids.
As for precedent, can anyone find a discussion of GPL'ed yacc/bison grammars? This would fit exactly the case above - the original source that must be distributed is the .y file, not the result of compiling the .y to a .c file. Unfortunately, I don't think that anyone has ever been tempted to rip off a GPL'ed grammar.
.y file with bison and still use it in a closed source product. This didn't used to be the case, but it is now. Before that happened, everyone who needed to do that just used yacc (or should have!)
Yacc has a BSD license, not GPL, and so this was never an issue at all for yacc. You can do whatever you want with BSD licensed code.
Bison makes a specific exception to the GPL for the code that it includes in your parser. You can compile your
I'm not an expert with legalese, but:
First arent all the copyright notices inside comments ?
Removing comments with the copyright notices would immediately violate T&C section 1. (while indicating acceptance of the whole document as per section 5), but then you aren't allowed to remove the comments. The obfusciation is seemingly permitted so long as the copyright comments still remain along with additional comments documenting the changes as required by section 2.
The obfusciation is seemingly a process of derivation, that is you start with GPL product and do some M-x replace-string's... This derivation process means that the "proprietary intellectual property" is still GPL'ed...
The GPL does NOT apply to sections not derived from GPL code, but only when they are published apart from the GPL portion. when the whole package is published it is still GPL'ed by inclusion of the GPL code (does anyone remember the Nvidia driver issues?)
Also according to section 5 the fact that you edited the GPL code at all indicates acceptance of GPL terms and conditions. Failure to accept prohibits you from making modifications (such as the string search and replace described)
The whole process seems expressly in violation of section 4, but i am no expert...
What I fail to see is how anyone can avoid GPL except by producing clean-room-code. I seem to recall Nvidia having this problem with their drivers a while back.
As an aside, isnt "chicken noodle soup" less than 30% chicken by volume? (but it is still considered a chicken product.) Your company's project might be 30% GPL code that was heavily edited (IMHO the only real weakness in the GPL is no "real" definition of "derived", however the common meanings of derive include "to trace the deveolpment of", which has been done...)
A couple of questions: Is it possible to write a perl/awk/sed script (or otherwise algorithmically describe the obfusciation? (since global replaces are used i would dare way yes...) If this is true then an argument can certainly be made that the work was "translated" from "ANSI c++" to "ANSI c++" (hasnt anyone done english-to-english translation between say a lawyer and an engineer? or perhaps heard of such things?). This translated copy would seemingly be covered by section 0 and all other sections (as incorporated into the defitition of modification)
just a few cents worth
-j.
I think that this is a distinction that is much easier to make than the previous one you mentioned. All you have to do is to go to the computers where the people are actually writing the code and see what form of the program they are modifying. If they're working on the code in a format different from what is distributed, it's an easy case that the form that's being distributed isn't the preferred form for making modifications. That's especially true if you can find:
You're correct that this is not an open and shut thing, but it's not an intractable one, either. Most people have fairly sensitive BS detectors, and they're going to be able to tell that code that's been deliberately messed with to make modification more difficult is not in the preferred format for making modifications. All you have to do is show that a deliberate attempt has been made to obfuscate with the code and you're set.
There's no point in questioning authority if you aren't going to listen to the answers.
IANAL so this is only from my observation:
Several posters have pointed out that obfuscation is a violation of GPL, or at least the spirit of the GPL.
Unfortunately, until your company actually releases a product based on obfuscated GPL code (commits a violation), you can't take legal action in the courts; you can only get a GPL-friendly lawyer to send nasty cease-and-desist letters.
In other words, you can't stop it until it's too late. And if you do sue, the copyright holder (the creator of the GPL code which was borrowed) will probably have to be named as a plaintiff, as the violation was commited against HIS copyright, or possibly the FSF as a plaintiff's representative yadda yadda yadda. YOU probably will not be able to file suit as a plaintiff directly, unless somehow you can do it as a representative of the party claiming loss.
If you do nothing else, inform the writer(s) of the original code of your company's intentions.
Give me my freedom, and I'll take care of my own security, thank you.
The GPL is a license that, when I apply it to my code, says that you may take my code, modify it (or not) and redistribute it. However it places the restriction that you MUST also distribute or offer to distribute the source code for anything you compile my code into.
This is why the GPL defines what source code is. And source code is HUMAN READABLE. Thast the point of source code. Code that has been preprocessed in some way, even if it is not a machine readable binary, is NOT source code unless it is in a human readable and inteligable language.
I do believe that this issue is specifically addressed in the GPL (along with a few other situations, which is why the GPL is so damned long).
This is definitly a violation of both the letter and the spirit of the GPL, and I urge the person who asked this question, or anyone else working for this corperation to blow the whistle on this project. It is a direct attempt to subvert the free software community.
-Steve
"I opened my eyes, and everything went dark again"
I recall a friend of mine who worked on a project that was designed obfuscated in the first place. Var names like vEh45c01, etc.
all proprietary, all aobscure, god forbid if the guy who designed it were to have an accident.
That said, would keeping the clear code represent a legal obligation? Since after all, the clear code is really the source, not the obfuscated stuff that you compile from.
"It is a greater offense to steal men's labor, than their clothes"
You may be a lawyer, but no other lawyers seem to agree with you.
:) When the company goes out of business it'll show they were lying.
I went across the hall at work yesterday and asked two lawyers who I often see over lunch about this. They said that while "preferred" and such terms are often fairly vague and cases hinge on those, in this case, where you can simply show the inability of the company to use the obfuscated code, and the obfuscating programs used, that it's dead simple.
Too bad modern judges can't hand down rulings that really cut to the heart of the problem...
Ruling that the company must delete all other source code and forever maintain the project using only this source code and other code in this form would quickly show if this was the preferred method.
(With creative and honest judges we could get by with a lot less of your type.)
While some judges might let you bring your industry usage of the term "preferred" to bear on the meaning and intent of this contract, just as many other will demand that the agreement be interpreted "within the four corners of the document". And why should one have to resort to an "industry standard" definition, when the question could easily have been addressed in the agreement itself? And how will you go about proving that your usage of preferred is widely-accepted? Will you bring in witnesses? How many? How many will your adversary bring in to prove that your usage of the term is not "industry standard"? How long will the parade of witnesses take? How much will it cost? Why wasn't this dispute clarified in the agreement before all the lost time and money? Do you have answers to these questions?
It's not a ploy, it's an assurance. Before the FSF is going to commit money and resources to defending a project, they want some proof that the author is truely commited to Free Software. Authors have refused to persue GPL violations in the past, and the FSF doesn't want to let authors pick and choose who can violate the GPL. They also do not want to defend a project that could potentially change it's liscence and become proprietary software sometime in the future. As long as the FSF holds the copyright that can't happen.
So while you are right, they could defend people without assigning copyright to them, they are trying to protect the interests of Free Software, and protect hemselves from being abused.