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."
To find out whether the gobbletygook you distribute is source code or not is simple: if you normally add features to the program by editing the gobbletygook, it's source. If you instead edit the stuff that you compiled to gobbletygook and then recompile it, then the stuff you distributed isn't source and it's a clear-cut GPL violation.
Why did it take so many posts for someone to point this out? Do people not read the GPL?
What a day to be without moderator points...
For those too lazy to read the whole thing, read section three, point #3 very carefully. Just because something compiles does NOT mean that it is source according to the GPL. That you would not do development on the obfuscated gobbledegook clearly shows that the obfuscated version is NOT the preferred form for modification. I would be highly suspicious that your lawyer is insufficiently anal when reading contracts if they missed this.
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.
(BTW, I am not a legal advisor. This is my understanding of the GPL).
If you are including other people's GPL'd source code in a program which you distribute, then you must abide by the terms of that license. Section 3 of the GPL is precise enough to disallow scrambling the source code:
There is nothing to stop you changing all the variable names, or the style of someone else's code. However, if you distribute a GPL'd binary then the source you distribute with it must be the source that you prefer to use for modifying the program yourself. You may be called upon to prove this in a dispute.
For reference, section 3:
The GPL states:
That term was written to prevent exactly the sort of obfuscation the attorney is proposing. Obfuscated code is demonstrably not the preferred version for creating modifications. So, what is being proposed is a GPL violation, and your company's attorney missed that part of the license. The talk about incidental resources isn't germane, it actually seems to be intended to confuse, because what is being proposed clearly is a derivative work, and the company attorney is acknowledging that when he suggests that the obfuscated code be GPL-ed.But there are simpler remedies than legal ones. If the free software developer community hears about a product using obfuscated code to circumvent the GPL, they will retaliate by creating a non-obfuscated version and using it to compete with your company's product. They are experienced at reverse-engineering, they have excellent tools for code reformatting and analysis, and there are a many programmers who will be angry enough to work on this.
If your employer wants to unashamedly take advantage, they are simply buying a lawsuit. The free software community does have the resources to bring one - it would probably be brought by law professor Eben Moglen of Columbia University. He wants more legal tests of the GPL, and would love to make an example of your employer. Don't go there.
Bruce
Bruce Perens.
1. Is it in violation of the GPL? This question is not a simple one, but such actions may very well be violation of the GPL. If this matter reached court, the question would center on whether the process applied to the GPL'd code constituted part of the process to create the derivative work, as derivitive work is defined in the GPL. For example, an expert might argue that code obfuscation can be part of the compilation process. It is oversimplified to say that laws are reinterpreted on the fly to capture the intent of the law. What is true is that these sorts of questions - for example, what constitutes compilation - are likely to be viewed in a manner which assists the obvious intent of the applicable contract/law.
2. If it is a violation, can it be proved? Probably. Our company works for lawyers on code plagiarism cases all the time. There are many algorithms you can apply to show statistically significant relationships between a body of code and its obfuscated counterpart. The same should be possible with bytecode. Once a reasonable basis for suspicion is established, plaintiffs could get discovery of the company's code repositories and depose employees under oath.
Christian Hicks
Elysium Digital, L.L.C.
http://www.elys.com
The GPL does address the issue of what constitutes "source code" at some length. From section 3:
The source code for a work means the preferred form of the work for making modifications to it.
I'd hardly think that obfuscated source would qualify as "the preferred form of the work for making modifications to it."
You could've hired me.
Now, the fact that this intermediate bytecode is legal C or whatever other language the original was written in doesn't make it source. It just means that that is the internal syntax of this intermediate stage. This is because the defining characteristic of source code is that it is human readable. It is what the developer wrote and would use to modify it himself. WHen you preprocess this in such a way that it is no longer suitable for human reading and maintaining, it ceases to be source...and ceases to meet the GPL source requirement.
I highly recomend that anyone who is going to talk about this actually READ the GPL
From the GPL:
-Steve
"I opened my eyes, and everything went dark again"
Yes, that would be one test for "preferred form", but there are others and the other side of any dispute will present them. The point is that the standard that you propose does not necessarily follow from the language of the GPL. In other words, your standard is more suitable than the GPL language. Of course, at trial, the credibility of your engineers and/or anyone testifying about their procedures will be at issue.