On Thu, 13 Jan 2005 17:02:52 -0500, Brian Thomas Sniffen <[EMAIL PROTECTED]> wrote: [snip] > Why are copies OK, and derivative works not? I see GPL 2b talking > about any work that in whole or in part contains the Program. > Eclipse+Kaffe contains Kaffe, GPL 2 then exempts mere aggregation -- > which this is not. It also exempts separate sections *when > distributed separately* -- and explicitly covers them when the Program > is distributed as a whole.
GPL section 2 is all about "works based on the Program", read in the context of copyright law. Later in that section, it says that "the intent is to exercise the right to control the distribution of derivative or collective works based on the Program." Eclipse+Kaffe isn't a "work based on the Program" because it isn't copyrightable as a derivative or collective work. In other words, you can't combine Eclipse and Kaffe, by running Eclipse on Kaffe, and then sue me for copyright infringement for doing the same. The "mere aggregation ... on a volume of storage or distribution medium" bit may clarify the contract writer's intent to some degree, but it doesn't automatically make other ways of combining two works copyrightable. To get at the cases the FSF is shooting for, they would have to use terms of art instead of "derivative or collective works", and would have to insert far more draconian provisions to create an action for breach of contract when GPL and non-GPL works are combined. > That feels like it's written to address this particular case -- or at > least a tarball containing Kaffe and Eclipse. This is similar enough > to count, it's just a funny compression scheme. It's written to address that case, all right; but I think that it's irrelevant and that the FSF is trying (unsuccessfully) to have it both ways. By inserting the "mere aggregation is OK" clause, and by publishing an aggressive stance on other forms of combination in their FAQ, the FSF tries to control interpretation and distract from the fact that courts will interpret "derivative or collective works" according to their meaning in the copyright law context that the preamble proclaims. Meanwhile, they claim the existence of a "law of license" outside contract law (with no basis that I can find), trying to avoid contract law standards such as "balance of harms" and take advantage of the relative ease of obtaining a preliminary injunction under copyright law. As I have detailed previously, the Progress Software v. MySQL case makes it pretty clear that they don't succeed in avoiding contract law standards once they actually get into court. Also, the judge in that case didn't rule on the factual issue of whether Progress Software had created a derivative work, but did make it quite clear that the FSF has no more authority on that point than any other expert witness that files an affidavit. The question "does linking create a derivative work" won't really be settled in the US unless and until a new case arises, is adjudicated in a court of fact, is appealed, and is properly analyzed with reference to precedent by a circuit court. I think (IANAL) that the outcome is quite predictable in light of precedents I have cited from several circuits. Ultimately, linking against the public interface to a published library doesn't detract from the rights in authorship that copyright law was created to protect, and US appellate courts are surprisingly good at sticking to the purpose of copyright in the face of legal and technical contortions that attempt to leverage a copyright monopoly beyond its legitimate bounds. Cheers, - Michael
