FSF Releases Third Draft of GPLv3

johnsu01 writes "The Free Software Foundation has announced publication of the third discussion draft of the GNU General Public License Version 3. Because quite a few changes have been made since the previous draft and important new issues have surfaced, the drafting process has been extended and revised to encourage more feedback. The most significant changes in this draft include refinements in the "tivoization" provisions to eliminate unwanted side effects, revision of the patent provisions to prevent end-runs around the license, and further steps toward compatibility with other free software licenses. The FSF has also explicitly asked the community whether the new patent provisions should apply retroactively to the Microsoft-Novell deal."

Comments

Can they do that?

[venicebeach]

The FSF has also explicitly asked the community whether the new patent provisions should apply retroactively to the Microsoft-Novell deal

Is that really an option? Wouldn't that be changing the terms of the license (v2) after it was distributed and agreed to? I don't understand how they can affect the Novell deal without going through the trouble of upgrading Linux to GPLv3-- and even then Novell should be able to use old Linux released under GPLv2, no?

Re:Can they do that?

[Bruce Perens]

Yes. What RMS was asking for was whether GPL3 code should include pernicious terms, from the start, that apply to people who had already committed a deal like the Novell-Microsoft one on GPL2 code but not yet on GPL3 code. The other option is to wait until said scoundrels commit the same deal on GPL3 code.

Bruce

Re:Can they do that?

[Bruce Perens]

You missed the part about "upgrading Linux to GPL3". Some people out there think Linus controls the license to all of what goes into a distribution, not just the kernel.

Everybody: Linux is just the kernel. Linus does not control anything else, and has less than absolute control over that.

Bruce

Re:Can they do that?

[Bruce Perens]

Linus specificaly stated that he is against using GPL v3 in its current form.

Actually, he probably meant draft-2 form. The current form didn't exist when he said that :-) .

Besides, specific objections are more helpful. Like he is against some DRM-related terms. I have gone over some of those terms here, you might find that useful.

Bruce

Re:Can they do that?

[Bruce Perens]

I'm going to have to correct myself. The specific sentence in square brackets, which means it's proposed but not accepted, would exempt anyone who has made a Novell-Microsoft-like agrement before today from enforcement of the terms even on GPL3 software. So, FSF is really asking "should we let Novell and Microsoft off, and just apply this to future violators?" I don't think they'll do that.

Bruce

"retroactively" was just a bad choice of word

[H4x0r Jim Duggan]

GPLv3 cannot be retroactive.

The question asked is whether the provisions that prevent deals such as the MS-Novell deal should have an explicit exclusion for that deal by Novell. i.e. such deals will be blocked in future, but should people who've already made such deals be prohibited from distributing GPLv3'd software?

That's the question asked.

Re:"retroactively" was just a bad choice of word

[Bruce Perens]

The provisions in section 11 say, essentially, that you can't transfer the grant of patent license to your customers when distributing GPL v3 software.

You don't have that right. You can transfer a patent license to your users as long as you do so to everyone. The point is that you can't create privileged groups like "people who have paid lots of money for protection" who have more rights than others.

I don't see how this is anti-user. It's an attempt to assure that everyone has a right to run the program.

Bruce

Re:"retroactively" was just a bad choice of word

[ivan256]

You don't have that right. You can transfer a patent license to your users as long as you do so to everyone.

The fact of the matter is that under a deal like the Microsoft/Novell deal, the distributor of the GPLed software doesn't have the right to grant the protection to everyone. Thus the clause means they can't grant it to anybody. The only loser is the end-user of the software who has no hope of protection from patent suits if they want to use GPLv3 software.

I don't see how this is anti-user. It's an attempt to assure that everyone has a right to run the program.

It's an attempt, sure. It's a noble effort even, and I'm as opposed to software patents as the next developer who works on GPLed software for a living. But I think that it is a failed attempt.

Re:"retroactively" was just a bad choice of word

[Bruce Perens]

It's an attempt, sure. It's a noble effort even, and I'm as opposed to software patents as the next developer who works on GPLed software for a living. But I think that it is a failed attempt.

It's based on the principle: we must all hang together or we will surely hang separately.

Bruce

Re:"retroactively" was just a bad choice of word

[g2devi]

Yes. You're correct. I still have the early Computer Languages (or was it Dr. Dobbs?) magazine article where they interviewed Stallman about the newly released Free Software Manifesto. The tone of the article portrayed Stallman as being a Don Quixote-like idealists that had little chance of succeeding (after all, everything was proprietary and the moment you make something free like in the 60s someone will come along and make it proprietary again). But he was looked at favourably in the same way that kindly and generous old grandfather that's out of touch with reality is.

Things have definitely changed since that day, but the threat of finding loopholes in the GPL to lock it up again and return us to the 1980s still remains. It remains to be seen of the GPLv3 helps or hinders free software (it has to maintain a fine balance between pragmatism and idealism). But at least the "additional permissions" feature of getting the GPL to be more compatible with other licenses (and reduce license fragmentation) and the Novell-Microsoft patent feature are definite improvements.

Re:"retroactively" was just a bad choice of word

[raehl]

The only loser is the end-user of the software who has no hope of protection from patent suits if they want to use GPLv3 software.

You have this backwards. The requirement that the patent license is provided to everyone *IS* the protection against patent suits for using GPLv3 software.

If a company distributes software under GPLv3, then elects to sue someone for using that software in violation of a patent, they open themselves up to being sued for violating the copyrights in that software by other contributors. Because the contributors have said 'You may not copy my code if you don't give everyone a patent license to use it and the derivative works.'

Separately, you are also only looking at half of the equation. Some users who are willing to pay to not be sued may lose out on being able to have the software. BUT, that doesn't matter any more than it matters that I've 'lost' the ability to run Windows unless I buy it. What *IS* important is that the contributors to free software have the copyrights in THEIR work protected. They have agreed to distribute their work to the community so long as it can be freely redistributed. If you do not prohibit distribution of software with a patent license, you are allowing companies to take the free software, modify it with patent-encumbered software, then UN-freely redistribute it.

This strikes to the very core of what free software is. If you're going to use free software, then you need to provide free software. Patent encumbered software isn't free, and just like we wouldn't allow GPL software to end up in a proprietary binary sold for profit without source, EVEN THOUGH it denies some users that software, we shouldn't let it end up in patent-encumbered source either, EVEN THOUGH it also denies some users that software.

Preventing non-free software from being created with free software is the whole point, isn't it?

Re:"retroactively" was just a bad choice of word

[F452]

Would be interesting to read that. I did a quick search and turned up this byte interview from 1986 (it references publication of the GNU Manifesto in Dr. Dobbs in 1985):

http://www.gnu.org/gnu/byte-interview.html

Which now I'll have to invest some time in reading. :-)

Re:"retroactively" was just a bad choice of word

[mrchaotica]

some major projects... in no special order examples are WINE, KDE, QT, Java and Firefox, GCC, Gnome (I assume we will have a LGPLv3, too)

GCC is the biggie there; since it's an official GNU project (which means its copyright is owned by the FSF) you know it's going to become GPL v3 as soon as the license is ready.

Retroactively?

[Mr.Ned]

"The FSF has also explicitly asked the community whether the new patent provisions should apply retroactively to the Microsoft-Novell deal."

I didn't see that in any of TFAs; does anyone have a link?

There is no "retroactive" change

[H4x0r Jim Duggan]

There is no "retroactive" change. That comment refers to the last setence of the 2nd last paragraph of section 11. That sentence, which is in square brackets, would make the ban only apply to deals that are made starting from today, so that deal by Novell and MS would not trigger the ban on distributing the software.

So the public are asked: should Novell be banned from distributing GPLv3'd software?

And, imlicitly, I guess, Novell are asked: What assurances can you give us to win our trust so that giving you this exception is justified?

Re:There is no "retroactive" change

[Hijacked Public]

Maybe, to assure everyone, Novell could adopt the slogan "Do No Evil". You people fell pretty hard for that one last time.

Why GPL v3?

[MS-06FZ]

Do we REALLY need a GPL v3? Probably - v2 was written in a time in which the scenario of the GPL being tested in court was rather more remote. One of the important reasons for v3 is to further lawyer-proof the license.

The question most people seem to be wondering about with v3 is whether it's too ambitious - seeking to prevent abuses of the license in ways some disagree with. Personally, I haven't made up my mind, exactly. I think the underlying premise of the GPL is great - that it is a license that allows free usage in a way that encourages more free usage - and GPL3 is taking that further, by trying to keep people from taking advantage of free software while simultaneously using patents against it, by trying to prevent people from using free software to create devices that restrict users' freedom (the idea being, that if someone wants a big DRM box, they can write the code themselves)

The flip side, of course, is at some point free software has to be something you give. At some point you need to let go, and let people use the stuff. That's why you wrote it, right? So people would use it. This is the sticking point for me - I like what GPLv3 is trying to accomplish - I even want to support what it's trying to accomplish - but sometimes, if you want your gift of software to be really useful, you need to stop attaching quite so many strings to it.

But all that aside, the real problem with the GPL v3 is that new clause that RMS will personally strangle a kitten every time someone uses GPLv3 code in a DRM box. We've got to see about getting that clause removed.

Re:So I no longer have to give up my private keys?

[Bruce Perens]

You never had to give up your keys. That's just FUD. I've written a longer explanation here.

Bruce

Re:I thought it was out already?!

[cronius]

This is a not a program you can change if it's broken, this is a license that could possibly have far reaching effect on the nature of free software. The last license was released over 15 years ago, you want to make it right so that v3 can last another 15 years or more. The license is complicated, and quite political, there are no easy answers.

Pluss, they want to take their time so that anyone who wants can voice their oppinion and be heard. Why rush it? Let them take their time and make it right, the first time.

Reaction to GPLv3

[Experiment 626]

I'm curious how the adoption of GPLv3 will play out. The kernel is going to stay at v2 for the foreseeable future, so the new version will mostly apply to the GNU tool chain. There are enough companies out there who like the loopholes of v2 (TiVo, SuSE, etc.), will they maintain a fork of the code that stays licensed under v2, perhaps individually, perhaps as a collective effort amongst those with reason to balk at v3? Another possibility is to just keep on using versions of the code that were released under v2. Some things, like /bin/ls, really don't change enough that everyone will feel compelled to step up to the latest version. On the other hand, if the GNU software the company depends on is gcc, staying at a particular release and not having support for new processor technologies in your compiler would start to become problematic after a while.

So, how do you guys think the companies for whom adopting GPLv3 would eliminate loopholes will react to the new license? Somehow, I don't think they will just all go, "Oh, so that's how you intended Free Software to be used. We will play nicely from now on."

Re:What happened to web apps?

[cyclop]

And I've never understood why this is bad.

A web app on a website is a source code usage, not distribution. The code runs on the web server and never leaves it. So why should I bother about it? In what sense it's different from me modifying a GPL program on my machine only and having my friends using it on my machine?

Adoption?

[Bruce Perens]

I don't think they have to worry about adoption. They have all of the FSF-owned software going to GPL3, which means that you can't really make a distribution without accepting GPL3. And most likely things like Samba, MySQL, Solaris and other Sun offerings, essentially anything owned by people who don't want the trick that Novell and Microsoft pulled to apply to them. I think that will be a lot of people. In the end, it might even be the kernel team. But that will take 1 to 2 years to play out.

Bruce

Re:Sadly...

[init100]

Freedom under the GPL has always been complicated, because it means free-as-in-FSF. As I've observed before, that isn't the same as "free" by any English language definition, which would be more akin to a BSD-style licence.

Would it? The BSD is akin to "You can do anything you want", while the GPL is akin to "You can do anything you want, except killing, raping, robbing or otherwise harming other people". The GPL is free, it just tries to stop people from restricting other peoples' freedom.

Re:I thought it was out already?!

[osu-neko]

So? does a low slashdot id equate to a knowledgeable, reasoned poster? I think not.

Sounds like a good metric to me! ;)

On the original topic: There's no need to rush GPLv3 out the door. There's a perfectly good GPLv2 out there serving the community as we speak, so why rush? Might as well take the time to make sure everything is the best it can be before release. It's not a matter of being "complete" -- the first draft was a complete document. It's a matter of being as good as it can be. If there was nothing like it out there already, that'd be a good reason to release quickly, but since there is, might as well take all the time desired, heck, take all the time in the world, it's not like we need a GPLv3, the GPLv2 is perfectly serviceable.

Re:Good luck.

[Bruce Perens]

but it isn't possible to run a gpl2 userland without the linux kernel.

Say what? Where do you think it was developed before Linux came along?

On Sun. RMS used to program on a Sun. GNU LIBC existed before it was ported to Linux. GCC did. Emacs did. Most of the userland did. Linus Torvalds did the last part, not the first.

I can think of a lot of kernels besides HURD and Minix. You could start with BSD and Solaris, but that's hardly the end of the list.

Bruce