Leading Linux developers are divided on whether the open source operating system should use the next version of the GNU General Public License. The first public discussion draft of GPL 3 was released last week. Over the weekend the Linux kernel mailing list started discussing whether this version could be used, as this may require a consensus among kernel developers.
Of course they should, especially if it helps deal with software patents more easily. The kernel coders that don’t want to upgrade to v3, simply drop their code from the kernel and recode, even if it takes time. I don’t have much time for Linux kernel developers that don’t want to support the FSF and GNU.
Dave
I don’t have much time for Linux kernel developers that don’t want to support the FSF and GNU.
I guess you spend your time improving the kernel, just like these developers do, right?
Not that I dislike the proposed v3 draft of the GPL, but we should avoid to rewrite code just for politics. If a licence switch can be done easily, they should go for it. Unfortunately, getting an almost-perfect consensus might be hard, let alone a perfect one.
Actually whatever Linus decides will have to either be good enough for developers or they can fork their own kernels. Somebody’s got to make the decision, and it’s going to be a decision some won’t like. Tough luck. Anybody who would actually pull their submission based on this license change doesn’t seem like somebody who’s very serious about Linux anyway. Maybe they should have been a submitter for BSD if they’re that particular about licenses.
Edited 2006-01-23 21:42
I love the strong stance against DRM and Patents, and hope it is encouraged.
If this means that Linux never will play DRM protected music, it will never be accepted by the music industry. And since more and more people are using iTunes and similar to buy music, they won’t give Linux a chance on their box. If my equation is correct it will mean the end of the dream about Linux to the masses… Please correct me if I misunderstood!
The license of Linux (which is a kernel, not an operating system) doesn’t limit you from using DRM decoders, even if it adopted the GPL v3 draft.
There are many articles posted here that will help you understand what the GPL v3 draft says about DRM.
“I don’t have much time for Linux kernel developers that don’t want to support the FSF and GNU”
Translation:
“You’re either with us, or against us.”
Typical selfrighteous comments by an advocate of “Freedom.”
Edited 2006-01-23 20:20
And sometimes perfectly valid. And your reply is just as bad, and btw. completely irrelevant to his comment.
As I wrote a few days ago I doubt that the Linux kernel will switch to GPL v3. And I still doubt it.
Anyway, it’s not unlikely that the license will change radically from the draft, so it’s far to early to come up with a decision.
I think it was short sighted of Linus to limit the kernel to GPL 2.0 only. He could have avoided any controversy like this, and there was no way to predict what the world would be like in the future. He should have realized that the current version may be outdated at a later time. Now, even if GPL 3.0 is the cat’s meow, upgrading the licence is going to be rough.
Now, even if GPL 3.0 is the cat’s meow, upgrading the licence is going to be rough.
It might even cause a split. A 2.0 kernel, and a 3.0 kernel. That would mean the end of Linux’s (slow) march into normal people’s homes.
“It might even cause a split. A 2.0 kernel, and a 3.0 kernel. That would mean the end of Linux’s (slow) march into normal people’s homes.”
People will jump on the first one to offer a stable binary driver interface.
On the one hand, we’d have Linus and his v2.0 GPL kernel (with Linus being dead set against having a stable binary driver interface), and on the other hand, we’d have a v3.0 GPL based kernel which would essentially forbid it.
I wish I could agree with you here, but this is one of those times that I can’t…
EDIT: with Linus being…
Edited 2006-01-23 21:41
You mean NetLinux, FreeLinux and OpenLinux?
Well, the march is anything but slow. OS X has been no faster mind you
I don’t licenses will cause a split. The license is going to stay v.2 until everybody agrees on using version 3.
Perhaps a few developers will quit working on the kernel, but there will be no split as such.
“It might even cause a split. A 2.0 kernel, and a 3.0 kernel. That would mean the end of Linux’s (slow) march into normal people’s homes.”
Anyone you slice it….. The gang at MS is laughing their asses off at the possibility of this happening.
My theory of the current GPL….. It ain’t broke, leave it the hell alone.
He didn’t limit the kernel to GPL2. He released his code under GPL2, without also (effectively) releasing it under the GPL3 at a time when GPL3 wasn’t written yet.
Of course releasing all kernel code under GPL3 will be hard because every contributor needs to re-release his code under the new license, but it’s possible. I can well imagine that some people prefer that, compared to effectively releasing their code under a license they can’t read yet.
Personally I trusted that my code is also well-protected under GPL3, but there certainly is something to be said for what Linus did back then.
He didn’t limit the kernel to GPL2. He released his code under GPL2, without also (effectively) releasing it under the GPL3 at a time when GPL3 wasn’t written yet.”
2 NOTE! This copyright does *not* cover user programs that use kernel
3 services by normal system calls – this is merely considered normal use
4 of the kernel, and does *not* fall under the heading of “derived work”.
5 Also note that the GPL below is copyrighted by the Free Software
6 Foundation, but the instance of code that it refers to (the Linux
7 kernel) is copyrighted by me and others who actually wrote it.
8
9 Also note that the only valid version of the GPL as far as the kernel
10 is concerned is _this_ particular version of the license (ie v2, not
11 v2.2 or v3.x or whatever
http://fxr.watson.org/fxr/source/COPYING?v=linux-2.6.9
Edited 2006-01-23 22:58
Yeah, but it’s only valid for his own contributions, and the kernel devs aren’t sure about the legal status of Linus’ note.
So Linus’ code isn’t automatically GPL3’d. That doesn’t mean he can’t release it as GPL3 after reading that license and deciding it’s actually a good choice.
I think it was short sighted of Linus to limit the kernel to GPL 2.0 only.
I rather think it was the opposite, and really shows how carefull Linus is with issues like this. When he made that decision no one knew what the GPL 3.0 would be like, and it’s still not finalized. No one knew, as Allan Cox so aptly put it, whether the GPL v3 would be a sane license. There was no way to predict how the GPL would change, so Linus played safe and removed the unknown factor. Both forward thinking and solid engineering in my book.
Edited 2006-01-23 23:19
When you look at the licences, they say you have the OPTION of using GPL 2.0 or higher. You don’t have to. Now, we don’t have a choice, whereas if Linus didn’t put that limitation on, we WOULD have a choice.
they say you have the OPTION of using GPL 2.0 or higher.
Exactly, and that was really what Linus wanted to avoid, not knowing the exact wording of the GPL 3.0. Any scenario where a distribution under version 2 is allowed, but not with version 3 would cause problems. You could even have made the distribution before version 3 was released, but the one you distribute to accept it under the new version 3. Giving him the possibility to sue you.
Given that no one really knew how the GPL 3.0 would look like at the time, you could not guard yourself against such scenarios. I find it strange that not more projects has seen this and guarded against it.
Hmmm. I disagree with your assessment, but it’s clear to me that this isn’t an easy issue. I still think Linus should have given the option to go to GPL 3 or higher, as I think the pros would outweigh the cons, no matter what GPL 3 turns out to be, since 4.0 could always be used if better (who knows). In any case, I’ll take back my comment that Linus was short sighted, as he probably just made a different call than what I would have done.
I still think Linus should have given the option to go to GPL 3 or higher, as I think the pros would outweigh the cons, no matter what GPL 3 turns out to be, since 4.0 could always be used if better (who knows).
And what would you have him do until the hypothetical GPLv4 is released? Sit there and wait? I would just worry about the case where the GPLv3 does everything Stallman wants, but doesn’t do what the Linux developers want.
It confused me too, at first, but consider this: If he didn’t specify it exactly, he’d basically be giving control of the licensing to whoever writes the next version of the GPL.
“And what would you have him do until the hypothetical GPLv4 is released? Sit there and wait?”
Yes. Just keep the GPL 2.0 like normal and continue life as we know it.
“If he didn’t specify it exactly, he’d basically be giving control of the licensing to whoever writes the next version of the GPL.”
I really don’t agree. The licence doesn’t say you HAVE to go to the next version, only that you can.
It confused me too, at first, but consider this: If he didn’t specify it exactly, he’d basically be giving control of the licensing to whoever writes the next version of the GPL.
Maybe not giving the control, since the end user would still have the choice of taking v2… However, I am pretty sure it wouldn’t have solved the issue that is arising, which is accepting “v3 or later” code, which is excluding v2.
Even if Linus’ clause is invalid, the code would still be licensed under the GPLv2 or later. If people start contributing v3 code, then it would be quite a mess for those favouring v2. And before someone saying “who would prefer v2 over v3”, some people will. After all, we just had the chance to read the draft. It could change drastically before it’s officially issued by the FSF.
How is this bad? Well, imagine if the source had to be separated in GPLv2 and GPLv3 parts. Not problematic to the end user or the average developer, but it could be a pain to maintain, especially if we get v3.1, v4… Of course, the kernel could switch to GPLv3, but it would need to ask the advice of every contributor. “Most of them” is not enough. Anyway, I don’t think subjucating to the FSF without discussion would be a good thing. The FSF keeps reminding us it’s GNU/Linux, it also demonstrates that Linux is not a GNU project, nor it should be managed like one.
All in all, I pretty much agree with you. He probably made the wisest choice, since nobody in 1991 was knowing how the GPLv3 would look like about 16 years later. I believe the statut quo is the best option. And I believe many developers think the same way since they have better things to do than meddling with politics or legal mumbo jumbo.
It might even cause a split. A 2.0 kernel, and a 3.0 kernel. That would mean the end of Linux’s (slow) march into normal people’s homes.
How so. How hard is it to go and edit text to read 3.0?
Many people have contributed code to the Linux kernel, and each individual copyright holder determines which license his code will use.
No debate of Linux kernel developers has started. Someone brought the subject up; another said it’s too early, and then a troll who is not a kernel developer started flaming the GPL.
Yup… that’s pretty much how it started in this thread.
And thanks for being a part of your summary (guy #2).
Not to mention that the person who started the thread, Jeff ‘I-want-to-buy-a-non-gpl-licensed-copy-of-the-kernel-for -$50k’ Merkey, is a known loon.
It’s impossible to change the license unless they get all the copyright holders to agree to change. So basically that would entail rewriting all sections of code where parties that don’t agree to change or those parties that are dead. It’s a non-issue and DOA.
as I understand it, the previously quoted paragraph limiting Linux to GPL 2.0 was added by Linus himself at some point after GPL 2.0 came out, without consultation with other kernel developers (certainly not ALL of them anyway).
So, if the primary devs discuss and decide they’d like to upgrade, Linus can remove this edit. They can then use the argument that all code in the kernel written before the edit was never anything but vanilla GPL2. Then, all these primary devs can officially re-release all their code, and a lot should be covered at this point.
Finally comes the only hard part: the process of tracking down those developers who contributed to the kernel some still extant code, but aren’t up to date enough to know they need to re-release. My feeling is, given the pragmatic approach of Linus and the other developers, they will simply make a best-effort attempt at this for a while, and when they’ve got most of it, they’ll declare it done and release a fully 3.0 kernel. If anyone challenges this, their work (which won’t be significant) will be quickly dropped and rewritten as soon as possible (in Merkey’s case especially, I’m sure some of them would relish the chance to remove the man’s last excuse for calling himself a kernel developer).
That is to say, the kernel developers don’t have their hands tied – they’ll consider GPL 3.0 on its own merits and see if its worth all the trouble.
That’s asking for legal trouble. Not saying there will be any, but it’s entirely possible.
Does that mean he can’t because Linus’ license restrictions supercede everybody else?. After all a driver or any other part of the kernel that talks to the core kernel code is considered derivative of Linus’ work.
Unlike the FSF, Linus never required an assignment of copyright when accepting submissions, and he’s only been tracking the provenance of patches for the last year or so. There are literally thousands of people who have submitted code over the last few years. To change the license he would need to get permission from each of these developers.
Making life worse is the fact that there’s no easy way of tracking who added which code thanks to the absence of source-control in the early days, and the various changeovers that have happened since.
More than likely, despite all the talk, the kernel developers will be forced to accept that they can’t now change the license even if they wanted.
The draft of GPL V3 contains provisions in section 7 for different licenses, which are non-GPL V3 but do give the Four Freedoms. If included, the GPL V3 is claimed to be the license of the whole collected work, but not of the individually licensed non-GPL V3 components.
This could be the escape needed by the kerneldevs to adopt the V3, without relicensing the V2 parts for which they don’t have or can’t get the permission. Assuming they would like to step up to V3 in the first place…
> The draft of GPL V3 contains provisions in section 7
> for different licenses, which are non-GPL V3 but do
> give the Four Freedoms. If included, the GPL V3 is
> claimed to be the license of the whole collected work,
> but not of the individually licensed non-GPL V3
> components.
But the license of those parts (GPL 2) says that a derivative work (the whole work, in this case) must be licensed under GPL 2. This doesn’t look like it would work.
– Morin