Raul Miller <[EMAIL PROTECTED]> writes: >> Raul Miller <[EMAIL PROTECTED]> writes: >> > On Sat, Jan 15, 2005 at 02:31:13PM -0500, Brian Thomas Sniffen wrote: >> >> Again, this isn't about the copyright holder's right to control >> >> production of derived works. This is about the copyright holder's >> >> right to control copying and distribution of copies. Reading GPL 2b, >> >> I cannot see permission to distribute a CD with Eclipse and Kaffe on >> >> it, such that Eclipse runs on top of Kaffe when I insert the CD. > >> > Copies of what? > > On Sat, Jan 15, 2005 at 10:40:08PM -0500, Brian Thomas Sniffen wrote: >> Copies of his work -- of Kaffe, in this case. > > Kaffe is GPLed. I can distribute Kaffe unless I violate the terms of > the GPL. > > So it's up to you to show that I would be violating the GPL when I > distribute both Eclipse 3.0 and Kaffe... > >> > The license on Kaffe applies to Kaffe and ... >> > >> > "This License applies to any program or other work which contains a >> > notice placed by the copyright holder saying it may be distributed >> > under the terms of this General Public License. The "Program", below, >> > refers to any such program or work, and a "work based on the Program" >> > means either the Program or any derivative work under copyright law: >> > that is to say, a work containing the Program or a portion of it, >> > either verbatim or with modifications and/or translated into another >> > language. ..." >> > >> > If Kaffe + Eclipse 3.0 is not a derivative work of Kaffe under copyright >> > law, then the rest of the GPL is irrelevant. >> >> It says, and you quoted: "a work containing the Program or a portion >> of it." If with modifications, that's a derivative work. If without >> modifications that's "the Program" in the phrase "the Program or any >> derivative work". > > Indeed it does. So what's your basis for saying that "Eclipse 3.0 and > Kaffe" is "a work"? Is it a "work based on the Program"? If it is, > then which of the following is it:
The Debian OS is a work containing a copy of Kaffe. > [a] the Program? (obviously not -- there are no GPL notices on Eclipse > 3.0). > > [b] any derivative work under copyright law? > > I'm guessing you're thinking that the combination constitutes a > derivative work under copyright law. But what is your basis for this? > This is not a rhetorical question, I'm asking for a specific cite of > some relevant law or legal precedent. No, I'm not talking about derivative works. I'm talking about direct copies. In order to distribute copies of GPL'd works, you must obey GPL 2b. It doesn't talk about derivatives; it talks about bundling of any kind! Then the "mere aggregation" example carves out a broader license later. But it doesn't apply to entwined programs. > But that's a pretty general request, so I'll try to point out, as clearly > as I can, what the overall issues are and what I think are the significant > points within those issues -- in other words I plan on outlining in this > message the things which I'd want to make sure are addressed before I > considered a legal cite relevant. > > In other words, I'm asking "why does eclipse 3.0 + kaffe constitute a > derived work of kaffe under copyright law"? I don't think they are a derived work of kaffe. I just think that in order to distribute Ecliipse+Kaffe, you must be distributing copies of Kaffe. >> GPL 2b says that if distributing a combined work which contains a copy >> of a GPL'd work, then the entire result must be under the terms of the >> GPL. This is that case exactly. > > Only if Eclipse 3.0 + Kaffe constitutes a derived work. But how can > that be, when the total content which distinguishes this "work" from the > "non-work" is less than 80 bytes of text? No, it doesn't say anything about derived works! It says "combined work which contains a copy of a GPL'd work". > So what part of copyright law makes "Eclipse 3.0 + Kaffe" a copyrightable > work, other than the fact that they're on the same volume? > > What copyrightable work was done to make this combination happen? Selecting the two of them and pairing them together. The Debian OS is a copyrightable collection. > Remember, we already have Eclipse 3.0 as a running program without Kaffe. > We don't have it in main, but your argument would keep the combination > out of non-free as well as main. And we're considering moving it based on a creative criterion: a desire to maintain a group of self-buildable free software. >> > Can you provide such evidence? >> >> I can provide evidence that it doesn't qualify as mere aggregation -- >> the dependency and automated invocation of Kaffe by running "eclipse" >> at a command line. > > But is that enough evidence? For example, depending on how the user has > put together their system, eclipse might very well not invoke Kaffe. > (Which, combined with basic information about java, is sufficient to > show that "eclipse works with kaffe" is not the result of any specific > work to combine the two, but instead is the result of both following > the same standards.) Eclipse *could* run on some other JVM. But in shipped Debian CDs, it will be paired only with Kaffe, and there's a specific Depends: line to show that we really did know and intend that they would work together. >> It seems plainly evident to me that the Debian OS is a work containing >> the Program (Kaffe), so all the parts of Debian that aren't merely >> aggregated with Kaffe have to be distributed under the GPL. Emacs, >> for example, is clearly merely aggregated. > > You're ignoring Kaffe's explicit permission, stating this kind of thing > is ok even if someone might think the GPL says otherwise. But ok, > let's focus on proof that the combination is not mere aggregation. What explicit permission? If there is such, throw it into the copyright file and we can be done with this. > Copyright law protects works of art, science, etc. Copyright does not > protect ideas, procedures, methods of operation or mathematical concepts > as such. > > As near as I can tell, the copyrightable work consists solely of: > > [a] Depends: kaffe | ... > > [b] Build-Depends: kaffe | ... > > [c] The work of choosing both kaffe and eclipse to go onto the machines > of Debian's users. (c) alone is enough to qualify it as a copyrightable work. > [d] The program runs. > > [e] The association of eclipse -> jvm -> kaffe which will exist on some > users' machines. > > [a] and [b] constitute significantly less than a sentence, so aren't > copyrightable -- they're too trivial to count in and of themselves. > > [c] is covered under mere aggregation. (c) alone would be mere aggregation. (c) and (a) and (b) and (e) mean that it's no longer mere aggregation. > [d] and [e] not protected by copyright law. > > In the more general case, [d] and [e] would indicate that more work > than mere aggregation had taken place to form the derived work. But we > already had Eclipse and Kaffe. We were just preventing them from being > installed in a fashion where eclipse would work with Kaffe. We didn't have Eclipse. We had it in contrib. > But that distinction -- preventing from running vs. running -- is not > in and of itself a concern for copyright law. That's true. But it is a concern for the GPL's "mere aggregation," and we have to care about what the GPL thinks because copying Kaffe is a concern for copyright law. >> It's easier for me to think about this when considering stripped down >> distributions; a distribution of Emacs+Kaffe wouldn't be anything but >> mere aggregation, but Eclipse+Kaffe would be integrated, more than >> just aggregation. > > How about eclipse + kaffe, where we've two words different such that > eclipse won't invoke kaffe? Because if it's illegal with those two > words it has to be illegal without those two words. No, I think if there were a BSD-licensed JVM in main, and Eclipse depended on that, this would all be fine. >> > If not, do you have any suggestions about good ways to find such evidence? >> > >> > If not, what grounds do you have for you claiming that those GPL >> > restrictions are relevant to this discussion? >> >> I think that's pretty clear evidence that Debian does have to worry >> about this. > > If we do, then just getting rid of the Depend: and Build-Depends: > won't be enough. Those phrases aren't sufficient content to matter for > copyright law. > > And, just to repeat myself, copyright law does not concern itself with > what the programs do. What the programs do is just evidence that might > be used in court to show what kind of work was done. But copying Kaffe is a concern for copyright law; once it's involved, our only way out is to deal with the much more finicky GPL, which does care about entwining of the programs. -Brian -- Brian Sniffen [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
