Re: Eclipse 3.0 Running ILLEGALY on Kaffe
> > 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:
On Sun, Jan 16, 2005 at 11:46:36AM -0500, Brian Thomas Sniffen wrote:
> The Debian OS is a work containing a copy of Kaffe.
Sure. And, in general, there's a small part of that OS which constitutes
the program Kaffe, and the rest of that OS is what the GPL calls "mere
aggregation". We'd have all sorts of problems if this were not the case.
If that's your only basis for saying that "Eclipse 3.0 and Kaffe" are
"a work", then the work is one of "mere aggregation, and the GPL's
requirements for the work are not an issue.
Do you have any other basis for saying that "Eclipse 3.0 and Kaffe"
is a work?
> > [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!
Direct copies of Kaffe do not include Eclipse 3.0. Direct copies of
the Debian OS are merely aggregation unless you can show some other form
of derivation.
> Then the "mere aggregation" example carves out a broader license
> later. But it doesn't apply to entwined programs.
Entwined is not a term from the GPL, nor from copyright law.
I'm going to guess that you're using "entwined" to refer to the GPL phrase
"work based on the Program"?
> > 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.
Ok, so you claim that Eclipse+Kaffe is not a derived work of kaffe.
In that case, your argument fails at first sentence of section 0 of
the GPL.
If the GPL does not apply, the GPL's requirements do not apply.
> >> 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".
That's not what the GPL says.
> > 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.
Which seems to be addressed in the GPL in the mere aggregation sentence:
In addition, mere aggregation of another work not based on the Program
with the Program (or with a work based on the Program) on a volume
of a storage or distribution medium does not bring the other work
under the scope of this License.
If Eclipse+Kaffe is not "a work based on Kaffe", I don't see any GPL
basis for thinking that this sentence does not apply.
> > 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.
This is a procedural issue, and is not relevant under copyright law.
So this can only matter if the GPL states that this is an issue, but
the GPL provides an explicit exemption for works not based on Kaffe.
> > 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.
$ cd /usr/share/doc/kaffe
$ grep -c exception copyright
21
> > [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.
In my opinion, this falls under the GPL's mere aggregation clause.
If there were any evidence that Eclipse had been developed for
Kaffe, I'd have a different opinion.
As it is, I see no difference -- in the context of the GPL or in
the context of Copyright law -- between Eclipse + Kaffe and any
other combination of content with a program designed to process
that kind of content.
I see only functional differences between Eclipse+Kaffe and
less + text document
ghostscript + postscript document
electric eyes + image file
g3dav + wireframe objects
xpdf + pdf file
gzip + compressed file
...
But neither the GPL nor copyright law makes any mention of these kinds
of functional issues.
So your logic must apply equally to all these cases.
> (c) alone would be mere aggregation. (c) and (a) and (b) and (e) mean
> that it's no longer mere aggregation.
(a) and (b) were depends and build-depends, (e) was the os level
association based on file type. I see nothing in the GPL which makes
me think that these constitute anything other than aggregation.
The GPL is concerned with derivative works, and you've already asserted
that you don't believe that we're talking about a derivative work here.
So what's your basis for saying that "mere aggregation" does not apply?
> > [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.
That's a DFSG issue, not a copyright issue.
The DFSG does not concern itself about license compatability between
viewers/players and their content.
If you can show that there IS a copyright issue, then the DFSG would
matter when judging that copyright issue.
However the only other relevance that I can see the DFSG having for
this case has to do with the number of users at any point in time.
And I don't see anything in the GPL which would make that relevant.
> > 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.
Where in the GPL does the distinction between running and preventing
from running make a difference? What is your basis for saying that this
difference overrides the mere aggregation clause?
> >> 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.
Why would that be?
I don't see anything in the GPL that applies to the distinction between
having a BSD-licensed JVMs in main, or not.
> > 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.
The GPL doesn't have any explicit mention of "entwining".
The GPL concerns itself with issues of derivation, and makes it very
clear that other issues that a person might think of as "entwining"
are irrelevant (running the program, aggregating the program with other
content).
--
Raul
Reply to: