On 5/19/05, Raul Miller <[EMAIL PROTECTED]> wrote: > But the ambiguities have to be valid ambiguities. > > That's where we seem to differ on this issue.
I think there is little question that the "work based on the Program" definition + erroneous paraphrase in Section 0 is either: 1) a "valid ambiguity" (to be construed against the offeror on the licensee's request), or 2) unambiguously readable only as "derivative work under copyright law", because the paraphrase is so weakly attached as to be an implausible candidate for a definition even if the licensee wanted it that way. Perhaps you would now agree to this "either/or", without any implications for whether my reading of the phrase "derivative work under copyright law" is correct? > Eh... ok. I was remembering this: > > "Existing case law with respect to copyright _licenses_ is > always, always, always based on contract law (in the US, > anyway)." > > And was thinking you meant that there could never be any > other kind of case. That's understandable. I have at times acted more like I was disputing with the author of the FSF's FAQ (and "Philosophy of Free Software" article) than with you. There's no question that other parts of the law come up in real-world cases where a copyrighted work is in dispute. Let's go with, "whatever the cause of action under discussion, a claim of (non-exclusive) license to a copyright is always viewed through the lens of contract law, and construed accordingly." OK? > > No, but the district court's subsequent ruling did say this; and even > > if it hadn't, the legal principle that analysis of scope of license > > must precede any ruling on copyright infringement claims would still > > hold. > > But this seems rather irrelevant in the case of the GPL. > > Either there's a breach -- in which case there is nothing within > the scope of the license (which is the default state under > copyright law) -- or there hasn't been, in which case the > GPL has granted you some rights, and there's little or > no problem. Can we focus on this question of what "scope of license" means for a bit? I am _not_ talking about what happens once a court rules that a license has been terminated for breach of contract. I am talking about a situation where the contract wouldn't authorize the defendant's conduct irrespective of whether it had been breached. In this situation -- and only in this situation -- can the court skip the rest of the contract analysis. That's the question on which the Sun v. Microsoft appellate decision hinged -- the district court couldn't rule correctly on a copyright infringement claim until it had evaluated the scope of the license. And on remand, it correctly (IMHO, IANAL) concluded that the scope of the TLDA included the conduct that Sun had alleged constituted infringement, and denied Sun's copyright infringement claims. > > > I think we can agree that Microsoft was violating the license. > > > > I have seen no ruling to that effect that was not vacated by a higher > > court. Microsoft was enjoined under a state law regarding unfair > > competition -- a tort -- for which the license terms were part of the > > evidence. Do you have any indication that Sun succeeded in obtaining > > any judgment on a breach of contract theory? > > I'm not sure I understand you here. As I read it, the injunction was > based in part on the belief that Microsoft would continue breaching > the contract. Are you saying that this could be the case when > Microsoft had never breached the contract? It still appears to me that no remedy, injunctive or otherwise, for breach of contract was ever ordered in that sequence of cases (although I still have more to read). But on review, there's an element of breach of contract theory in the "likelihood of success on the merits" part of the unfair competition judgment. The finding that Microsoft probably violated the compatibility provisions of the TLDA was a significant part of the support for the conclusion that their competitive practices were unfair. It appears to me (IANAL) to be possible that the unfair competition ruling would not stand without adequate evidence that Microsoft's conduct violated a binding agreement with Sun -- i. e., breached the contract. Score one for you. The "recurrence" part also had to do with the facts surrounding the unfair competition claim -- but you're basically right. The court judged that, in the absence of a court order, Microsoft was likely to go on playing shell games with the "real" specification-compliant MS Java run-time and SDKJ. It accordingly enjoined Microsoft to "put up or shut up" (not a literal quote) -- either give the "real" Java equal standing with its "embraced and extended" edition, or stop distributing any Java Technology at all. > Or are you referring to this being a preliminary injunction? > That nothing was actually proven -- that it was only shown that > the case was likely to succeed? This seems standard for > preliminary injunctions. > > Anyways, I'll quote from what appears to be the current > injunction -- as far as I know, this was not vacated? > > "Preliminarily, Sun has established a reasonable likelihood > of success in demonstrating that Microsoft's distribution of > non-compliant Java technology violates the compatibility > provisions of the TLDA. See Sun, 21 F. Supp. 2d at 1119-22, > 1122-25; Sun, 188 F.3d at 1120 (holding that substantial > evidence demonstrates that Microsoft violated the TLDA)." > > http://java.sun.com/lawsuit/012400motioncal.html You're reading that correctly; I stand corrected. Just note that no remedy for breach of contract as such was ordered, least of all termination of the TLDA. [snip general agreement] > > > In the context of the GPL, the remedy contained in the > > > agreement is that the license terminates. > > > > Funny how that's not how it worked in Progress Software v. MySQL, isn't it? > > I think that's because MySQL didn't properly charge > that Progress Software had violated the GPL. > > I certainly didn't see any discussion by the judge of > the issues which I would have thought were pertinent. > This implies to me that those issues weren't raised. > Or, if they were raised, that they weren't backed up > with facts. A claim of "GPL violation" was raised, all right. But as Judge Saris explained, the facts of the case didn't begin to match _any_ of the criteria for a preliminary injunction, let alone all of them. And (just to harp on this a little more) she very consciously used contract law standards, not the "automatic presumption of irreparable harm" applicable to copyright infringement claims. If a copyright infringement claim had been advanced by MySQL, it might have been wise on the judge's part to go through an explicit "scope of license" construction in the text of her opinion, just to avoid having it remanded to her on that technicality. (I don't think it could possibly have affected the outcome, but judges prefer to avoid being justly criticized on appeal.) But it doesn't actually seem that MySQL claimed that Progress Software's conduct was in this sense not within the "scope of license" granted by the GPL. As far as I can tell from the opinion (I have not yet obtained a copy of the full docket), MySQL's claims with regard to the GPL were entirely stated in terms of breach of contract. (Compare Eben Moglen's affidavit, which was not given in his capacity as an attorney at law but as an "expert witness".) Evidently MySQL's lawyer cared a little bit more about his credibility in the long run than Eben Moglen seems to; though again, that judgment is outside my qualifications. > > > So, in the context of the GPL -- unless you can show that > > > some other remedy applies -- breach of contract means > > > that B is infringing copyright. > > > > No, breach of contract means breach of contract. If the remedy > > granted by the court includes recognition of the offeror's grounds for > > contract termination as of a certain date -- applying statutory notice > > requirements and all that -- then continued copying, etc. beyond that > > date is copyright infringement. The remedies for that are specified > > by statute and the GPL's text has nothing to do with them. > > As I understand it, the only remedy which appears in the > GPL is termination. Do you have some reason to believe > different? > > If not, what does your "nothing to do with them" comment > mean? The text of the GPL has no effect on what remedies may be granted for copyright infringement, which is a statutory tort. It does have some effect (though not complete say) on what remedies may be granted for breach of contract. > I mean... I can imagine that the judge can construe some > other remedies -- I can imagine that the judge could order > the licensee to comply with the terms of the GPL, for example, > thus removing the breach. But, I don't see this as having > nothing to do with the remedies specified in the GPL -- I > see this as being motivated by those remedies. Different causes of action, different applicable law. > > > But maybe I'm wrong -- how could that be? > > > > > > Maybe the court has tossed out that remedy -- unless > > > some other remedy is provided by law, there is > > > no contract at this point. This would mean that the > > > case must proceed using some other principles > > > than contract law. > > > > Nope. Unless and until the contract is ruled to be terminated, the > > breach-of-contract part of the case proceeds under contract law using > > the fully construed contract, including whatever remedies for material > > breach and/or standards of cure that the applicable law and/or > > principles of equity may suggest. > > Can you be more specific? Injunctive relief, monetary damages, contract termination (opening the way for copyright infringement claims), and in extreme cases compulsion to perform specific acts (such as release source code) -- although I think that's usually phrased as an exception to an injunction, as in Sun v. Microsoft. I think any or all of those remedies might be ordered for breach of contract alone. Probably some other things are possible if state unfair competition law, the Clayton and/or Robinson-Patman Acts, RICO, or something like that can be brought to bear. No prison time or corporal punishment in the US without a criminal conviction though, AFAIK. > After all, I'm listing options here. You seem to want > to respond to each option with some text indicating that > there are other options, and you're right about that, but > I'm trying to construct a complete list. I don't begin to have the legal background to construct a complete list of what remedies may be available for breach of contract in an arbitrary jurisdiction. But the above list is ample for most purposes, I should think. > > > We're also left with the possibility that by law some other > > > remedy is a salient part of the contract. I'll be interested > > > if you find something significant along this line. > > > > > > One other possibility remains: B is guilty of copyright infringement. > > > > Not unless B has acted outside the scope of the license or the > > contract is ruled to have been invalid or terminated according to > > contract law and the facts of the case. > > Isn't that what I said? > > In other words, are you objecting purely to my choice of > words? Because it looks to me like you're offering > the same basic structure with more formal phrasing. You don't seem to be acknowledging the legal significance of "scope of license" as articulated in the Ninth Circuit's Sun v. Microsoft opinion and op. cit. You also don't seem to be clear on how a claim of contract termination is handled in a courtroom. As I understand it, if there is evidence that there was at one time a valid license, you can't win a copyright infringement claim (in front of a competent judge) without jumping one of those two hurdles. (Except via a statutory bypass such as droits morals, in a jurisdiction where it exists; and that's not exactly a copyright infringement claim anyway, and according to Batist can't be applied to software.) > > > For someone in breach of contract, the scope of license seems to be: > > > no license has been granted. That seems to me to be a rather narrow > > > scope. > > > > Bzzzt. That's not what "scope of license" means. > > It is when there is no license. Bzzzt. That's the alternate hurdle -- proving that the license has been terminated for cause -- and takes you through the entire "contract construction" gauntlet. It's the route that MySQL took -- and lost, despite a factual background that could hardly have been any more prejudicial to Progress. Oh, they had released the Gemini source code by the time the preliminary injunction hearing took place; but the judge very explicitly said that MySQL wouldn't have gotten an injunction anyway. In fact, they wouldn't even have met the "likelihood of success on the merits" criterion. > > > Of course, you acknowledged that in your statement (thus my > > > question about whether you meant it as a tautology). > > > > > > But basically, you've come up with a different way to describe > > > what we've been calling "violating the GPL". > > > > A very different way to describe the procedure of analysis, yes; and > > hence one which draws a very different picture of what is and isn't a > > "GPL violation". > > If I didn't have some severe problems with some of the more > concrete statements you've made, I'd love to go into this. > > At the moment... I'd prefer to defer any discussion of this view > of things until after we've reached agreement on two more > basic issues: > > [1] The whether derivative works and collective works are disjoint > or merely different. Especially in light of the "mise en scene" discussion -- a whole 'nother way for X+Y to be a derivative work -- I am now willing to retract "disjoint". The mathematical term doesn't fit legal reasoning that well anyway. (Sour grapes.) Score another for you. How about: "X+Y+... is sometimes an uncopyrightable collection containing Y and sometimes a collective work containing Y; but neither case adds one whit of support to a claim that X, X+Y, or X+Y+..., is a derivative work of Y." As long as you are OK with that, I will happily concede that X+Y+... might also wind up being described by a court as a "derivative work" of (the selection-and-arrangement-creative-expression in) some other collective work, or might be a derivative work of Y by some other theory (such as mise en scene). I reserve the right to argue differently in a courtroom if I think better of it later. :-) > [2] Whether the GPL is a valid license for any software where > section 0 of the GPL does not apply. It's also a valid license to create at least some "collections" containing the "work based on the Program". Whether construed from the "mere aggregation" clause, from industry practice with respect to software in general, or from 17 USC 117 and the simple fact that you can't do much with it without creating some "collective work" (distro CD, hard drive contents, whatever), a GPL licensee is given the right to create various collections, copyrightable and otherwise. There's simply no basis in the GPL text for defining some magical boundary at exec()/fork() or between interpreter and bytecode. If it weren't for the "mere aggregation" clause, it might get punted out to some computer industry expert witness, and who knows what they might declare to be standard practice -- the FSF's strong-arm tactics have helped establish a status quo in which people keep GPL code at arm's length. (I suspect that of being a big part of their agenda in maintaining the fictions in the FSF FAQ.) I think (IANAL) that 17 USC 117 protects dynamic linkers at the minimum, but an expert witness could conceivably convince a judge otherwise, and there's the rest of the world to consider. But given how loudly the FSF screams "no contract! all copyright!", perhaps a judge ought to look for a definition of "aggregation" in a strictly legal context, as close to copyright as possible. Per Merriam-Webster anyway, the closest fit is from patent law: "a collection of separate parts that is unpatentable because no integrated mechanism or new and useful result is produced". Don't just jump to "integrated mechanism" and say, "But that's exactly what a linked binary is!" That's a _patent_ law standard of "integration" and "new and useful", and judges still usually treat that with the respect it deserves. What's for damn sure is that the FSF's idea of where the line ought to be drawn -- to maximize the incentive to release new works under the GPL (if you trust them) or to give the FSF a monopoly on the supply of GCC, GDB, etc. to embedded systems IDE vendors (if you don't) -- shouldn't influence the construction at all. That's the #1 respect in which the FSF's claim that the GPL is a "copyright-based license" is not merely false but IMUnqualifiedHO dishonest. > We may have some other fundamental disagreements but > these are the two that trouble me the most. It may be starting to look like thesis-antithesis-synthesis after all. Cheers, - Michael