On Tuesday 04 September 2007, Rui Miguel Silva Seabra wrote: > Hi Sunnz, > > On Wed, Sep 05, 2007 at 04:32:20AM +1000, Sunnz wrote: > > > If the person chooses to use the GNU GPL they have to respect the > > > GNU GPL's conditions, not the BSD ones. > > > > GNU GPL, however, only grants the right to re-distribute (under > > certain conditions), but not re-license, right? > > No, the GNU GPL grants you the rights to > 0. run it for any purpose > 1. study & modify it > 2. reditribution of pristine copies > 3. redistribution of derivatives > > All this just like the BSD. However, unlike the BSD, it does so in a > reciprocal level: if you redistribute in the conditions of 2. or 3. > you must license it under these (the GNU GPL's) terms. > > > BTW, if satisfying requires in GPL would imply satisfaction of BSDL > > anyway, no? > > It's closer to include than imply, if you want to use these terms, > since satisfying the BSDL means allowing proprietary derivatives, > which the GPL aims to forbid. > > Rui
Hi Rui, You've been arguing in circles for days now but the real cause is there are some things about how copyright law works which you need to understand a bit better. Hopefully I can help. Only the copyright holder can modify, remove or replace the licensing of their copyrighted work. This is the law, and those silly license terms which state you cannot remove the license are nothing more than being overly redundant for the sake of idiots who do not know the law. Sadly, most licenses state the bleeding obvious. Unless the copyright holder specifically gives the recipient the right to relicense the work, the license cannot be changed. In the case of dual licensed, or better said, multi-licensed works, this law still remains in effect. Take the case of multi-licensed work of where there is no permission to create derivative works (i.e. modifications). One of the available licenses allows you to distribute verbatim copies of the work under certain terms. The rest of the available licenses do not grant you the right distribute copies. As long as you are in compliance with the terms set forth by the one particular license which allows distribution of verbatim copies, your action of distributing copies is legal, regardless if all of the other available licenses do not grant the right to distribute copies. As you can see, the right to create copies must be specifically granted by the copyright holder in order for the recipient to be legally able to exercise that right. The same is true for all other rights protected by copyright; They must be specifically granted by the copyright holder otherwise they are illegal. When given a choice between multiple licenses, the only choice you get is which license you wish to *comply* with, but you *never* receive the right to relicense the original work unless it is specifically granted. Even if you are granted the right to create derivative works, the copyright holder must specifically grant the right to modify, remove and/or replace the licensing for you to be able to do so legally. There is a vast and significant legal difference between creating a derivative work from an original copyrighted work, and relicensing the original copyrighted work. Unless the copyright holder specifically grants permission to do these things, doing them is illegal. Since the three multi-licensed atheros driver files never granted permission to modify, remove or replace the licensing, it is illegal to change the licensing in any way, and the code remains multi-licensed regardless of the misguided patches. If you create a GPL-Only licensed patch which modifies any of the original work, you now have a legal contradiction. If you apply the patch and claim the resulting derivative work is now GPL-Only, you've just broken the law because you were never granted the right to relicense the original. On the other hand, if I apply the GPL-Only patch and say the resulting derivative work is now multi-licensed like the original work, I may also be breaking the law... This is the reason why companies like TrollTech take the time to make sure all patches to the dual-licensed Qt toolkit have amicable licensing, otherwise they reimplement the changes on their own. When you think of this legal contradiction in the terms of a patch that fixes an off-by-one security vulnerability, and the obvious need to apply the identical fix to the original, you'll see the precedence will go with the licensing of the original work. Otherwise I'd be filthy rich from researching exploits on Microsoft software, copyrighting the fixes and then holding them for ransom to the highest bidder. Who would be willing to pay me more for the *RIGHT* to fix the vulnerability; Microsoft for the sake of supporting their own products, or organized crime for the sake of having an exploit that Microsoft would never be able to legally fix? Obviously, the law doesn't work that way and the licensing of the original work takes precedence over the licensing of any addition or modification. If you happen to be a GPL advocate, think about it this way: as you've probably guessed, the linux kernel is multi-licensed where ever there is a statement like "GPL Version X or any later version." Liuns Torvalds has taken a strong stance about retaining GPLv2 on the linux kernel source code in his tree. If someone took all that source code and relicensed all of it as GPLv3, as the copyright holder (on some of it) Linus would be rightfully furious, and he would have very solid cause for legal action against whoever broke the law by doing such a disrespectful thing. You not required to agree with Linus, and you are not required to respect Linus, but you are required to follow the law or you will most likely suffer the consequences. By modifying the licensing of the original work on the three dual-licensed atheros driver files, the GNU/GPL/FSF people have done something blatantly illegal, and blatantly disrespectful. Any lawyer who advocated or advised such an obviously illegal action could hear from their respective bar association about the matter and it seems some of the lawyers from the FSF/softwarefreedom, as well as Eben Moglen, signed off on this particular malpractice: http://marc.info/?l=linux-wireless&m=118857712529898&w=2 (they have all been cc'd) Everyone understands the good intent behind the entire atheros driver debacle was for the author of the derivative work to promote his license of choice, namely the GPL. Every advocate has the same intent of promotion. The serious problem here is the illegal act of modifying, removing or replacing original licensing on multi-license works. Ironically, doing this does not promote the GPL, but instead it weakens the legal standing of the GPL. Anyone can now take the supposed GPL derivative work, use it in a proprietary closed source product, never comply with the GPL terms because they chose to comply with the now missing alternative licenses, and then easily beat the GPL into the ground in a court of law, and hence, set legal precedence for defeating the GPL on a wider scale. Definitely not good. Your personal promotion of your ideals as defined by your license of choice is great as long as you think it through, but when you fail to think it through, the most likely result is your are only doing damage to your ideals by weakening the legal protection you wanted to achieve. I hope Luis Rodriguez takes the time to rethink his changes with greater insight and foresight before things get any worse than they already are. The "legal hell" he mentioned is far from over, and unfortunately, it's probably just beginning. Though good folks at the FSF/softwarefreedom have made a mistake in their practice of law with their advice on the atheros issue, it's really just a little "bug" in need of fixing, and I would much prefer to see them correct their mistake on their own, so we can end all of this nonsense amicably. Kind Regards, Jonathan C. Roberts
