Torvalds: No GPL 3 for Linux (eWeek) [LWN.net]

Linus is not planning on using the GPL3 license for the Linux kernel, according to this article on eWeek. "In a message to the Linux Kernel Mailing List on Jan. 25, Linus Torvalds made it plain that the Linux operating system is going to stay under General Public License 2 and not migrate to GPL 3. Torvalds announced this in response to a discussion on the list of Linux developers, which had been started by Jeff Merkey. The former Novell and Canopy Group developer, Merkey is best known in Linux circles for his attempt to buy a non-GPLed version of the Linux code."

to post comments

not much choice, really

Posted Jan 26, 2006 23:32 UTC (Thu) by stevenj (guest, #421) [Link] (13 responses)

As others have pointed out, there isn't really much to discuss — the license for the Linux kernel (which never contained the "version 2 or later, at your option" language) seems almost impossible to change because it would require obtaining the permission of untold numbers of contributors. (Something Merkey never seemed to grok.)

Linus' original post is here. See also the requisite Slashdot flamewar. (LT also seems to misunderstand the GPL v3's draft provision about providing keys required to run software, and makes an unfortunate incendiary comment in his typical style.)

not much choice, really

Posted Jan 27, 2006 0:08 UTC (Fri) by bk (guest, #25617) [Link] (5 responses)

I'm not a lawyer, but would it be possible to put out a message saying something like: "As of January 2007 [or whatever], we plan to license the kernel under the GPL with a minimum version of 3. If you do not wish to have your code included, you must reply to nogpl3@kernel.org with your full name, address and the precise code you own the copyright to and we will remove your code from the kernel prior to relicensing. If no response is received by 23:59 December 31, 2005 UST it will be interpreted as permission to license all covered code as described."

Wouldn't that work?

Don't think so!

Posted Jan 27, 2006 0:11 UTC (Fri) by felixfix (subscriber, #242) [Link]

Imagine trying to do the same to any other copyrighted work, such as, oh, I don't know, say a Walt Disney animated movie.

I don't think either would get very far.

um, no

Posted Jan 27, 2006 0:14 UTC (Fri) by stevenj (guest, #421) [Link] (3 responses)

Imagine that I sent you a letter in the mail saying, "We plan to take possession of your house, your dog, and your Aunt Petunia on February 23, 2006. If you do not object in writing before that date, it will be interpreted as permission."

um, no

Posted Jan 27, 2006 1:24 UTC (Fri) by bk (guest, #25617) [Link] (2 responses)

The difference being that permission to distribute copyrighted work is always revokable provided the copyright is still valid. So, for instance, if Linux contains code copyrighted by you today, you could always revoke permission tomorrow and have the code removed. In the hypothetical situation above, if an individual was in a coma and awoke a year after the GPL 3 switch they could have their code removed immediately at will.

um, no, part II

Posted Jan 27, 2006 2:21 UTC (Fri) by Ross (guest, #4065) [Link]

I don't believe that is true. You can prevent new people from copying your work, but if you have promised other people they can copy it as many times as they want, indefinitely, you can't go back on that. Now the GPL (and most other licenses) have provisions for revoking permission, but that's different than taking it away without warning.

revokability of GPL licenced code ?

Posted Jan 27, 2006 10:04 UTC (Fri) by copsewood (subscriber, #199) [Link]

I don't see that code once licensed under the GPL, of which a licensed copy has been obtained by anyone else willing to continue distributing it, can be revoked by the copyright owner. The GPL explicitly makes permission to continue distributing and modifying affected code transferrable between recipients who uphold the GPL's terms. This doesn't, however, allow a recipient to redistribute it under terms different (e.g. a different GPL version) from those under which rights to redistribute were irrevokably granted.

The copyright owner may opt to discontinue distribution themselves, or put the same code out under a different license, but can't revoke permission to continue distributing GPL licensed versions already out there.

not much choice, really

Posted Jan 27, 2006 15:07 UTC (Fri) by forthy (guest, #1525) [Link] (5 responses)

A whooping 2558 .c files in Linux (from SuSE 10.0) contain the standard FSF copyright disclaimer (with "any later version" at your option), so the point is moot. Linus isn't going to change those either, and the whole thing therefore is a mixture of GPLv2 code and GPLv2+ code (which is possible). After the GPLv3 is out, you simply can't continue to release it as if nothing happend. Linus should have denied the inclusion of all those 2558 files since they state the "any later version" he dislikes so much.

His writing on top of COPYING is not legally binding. It's not part of the license text as such, it's just his interpretation.

not much choice, really

Posted Jan 27, 2006 15:39 UTC (Fri) by bronson (subscriber, #4806) [Link] (4 responses)

Your post shows some misunderstandings of the GPL.

1. The release of the GPL V3 doesn't change the provisions of V2 at all. It's true that some of the code in the kernel allows GPL V2+, but that only applies to those files, not to te rest of the kernel as a whole!

2. Why should Linus deny the inclusion of those files? They are perfectly compatible with the GPL V2.

3. Finally, Linus didn't write on top of COPYING. Read the GPL: http://www.gnu.org/copyleft/gpl.html Notice how "at your option" comes after "END OF TERMS AND CONDITIONS". Therefore, "at your option" is not a part of the GPL and Linus didn't write on top of anything!

In short, Linus's decision to release Linux only under the GPL V2 is definitely legally binding. I have no idea how you managed to convince yourself otherwise.

not much choice, really

Posted Jan 27, 2006 18:31 UTC (Fri) by paulj (subscriber, #341) [Link] (3 responses)

... Therefore, "at your option" is not a part of the GPL and Linus didn't write on top of anything!

In short, Linus's decision to release Linux only under the GPL V2 is definitely legally binding. I have no idea how you managed to convince yourself otherwise.

I suspect the poster concerned had, unlike either Linus or yourself, read section 9 of the GPL.

not much choice, really

Posted Jan 28, 2006 4:20 UTC (Sat) by kmccarty (subscriber, #12085) [Link] (2 responses)

I suspect the poster concerned had, unlike either Linus or yourself, read section 9 of the GPL.

You mean the part that reads as follows? (emphasis mine)

If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.

I think the following quote in the top-level COPYING file in the kernel source keeps Linux from falling under the purview of either of the above clauses of Section 9:

Also note that the only valid version of the GPL as far as the kernel is concerned is _this_ particular version of the license (ie v2, not v2.2 or v3.x or whatever), unless explicitly otherwise stated. Linus Torvalds

not much choice, really

Posted Jan 28, 2006 6:17 UTC (Sat) by paulj (subscriber, #341) [Link]

Ug, bad quoting on my part, I had only intended to quibble with the Therefore, "at your option" is not a part of the GPL part. The GPL per default does allow any licence version to be chosen. The Linux kernel though obviously does have a preamble stating "v2 only" - I hadn't intended to dispute that, apologies.

not much choice, really

Posted Jan 30, 2006 17:03 UTC (Mon) by forthy (guest, #1525) [Link]

I don't think this preface from Linus concerns all the 2500+ files, which explicitely state GPLv2 or any later version. That's the point of "unless explicitly otherwise stated". 2500+ files have explicit other statements. And whether a file or a bunch of files constitutes a program, the answer is easy to tell: Is the NTFS driver a program (just one example, there are at least four file systems which completely run under the GPLv2 or later copyright; or, more important for DRM: ALSA)? It can come as a module, load into the kernel, and provide functions through a rather well-defined kernel API and ABI. It's a program. It's under GPLv2 or later. Linus can use it under GPLv2, but when he distributes it, it's again the next user's choice under which GPL he wants to use the NTFS part of the Linux kernel. The GPL clearly states that the user always gets the license straight from the horse's mouth, i.e. the original copyright holder:

6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License.

The rights "granted herein" include section 9., the right to choose another version of the GPL when the program doesn't specify a particular GPL version.

BTW: This logic guarantees that the GPL can become only more permissive. If you are fighting a patent war with Novell, while you still use Linux, you can always choose GPLv2, which does not allow a termination of the license for this reason (GPLv3 does).

not much choice, really

Posted Jan 27, 2006 16:20 UTC (Fri) by chel (guest, #11544) [Link]

IANAL, but just some thoughts. Linux is around for a while, with a general notice: it is GPL V2, and with some files with a notice: "V2 of the License, or (at your option) any later version". I have doubt about the legal status of "or (at your option) any later version". It is clear that with this clause, Linus is free to use, modify and redistribute the code under GPL V2. I have some doubts about the legal posibilities to change the license for existing code from V2 to V3, both from a user and a writer view. From a legal point of view it is strange to use an "Open License", a license that can change in time.

The current situation: a general notice that files in the tree are GPL V2, and the current headers, implies the distribution of the tree is completely GPL V2. Discussion if you want to change it or not, is without value, as no way to do the change is known.

When there is a GPL V3, there is room for discussion about new files: Can new files be GPL V3 only, and how should that affect the tree copyright? Should there be V2 and V3 makefiles to enable a V2 binary and would that help?

Hey, I understand the "Linux is GPL V2 for the time being" position. I think this licensing issue should not hinder development.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 26, 2006 23:41 UTC (Thu) by atai (subscriber, #10977) [Link] (5 responses)

The COPYING file in Linux should specify explicitly that it only applies to the code of Linus Torvalds, not anyone else. Every file should contain the complete notices about its own copying conditions.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 4:53 UTC (Fri) by xoddam (subscriber, #2322) [Link] (4 responses)

Every file in the core of the kernel contains code from numerous
contributors. The COPYING file pertains to the entire kernel and
contributions were always made under that assumption. Indeed
the contributions could ONLY be made under the provisions of the
copyright licence granted to the contributors in the first place
which was very explicitly GPLv2, and that licence always required
contributions to be distributed under its own provisions. While
code might have been separately licenced under alternative conditions
in addition to the GPL, it is most definitely licenced to all users
under the GPLv2.

Moreover Linus Torvalds owns the 'compilation' copyright to the
kernel, despite other parties' legitimate claims to their
particular contributions. The work of integrating the contributions
has explicitly been his.

The ONLY way to change the licence of the kernel as a whole is for
every contributor -- most importantly Linus Torvalds himself -- to
relicence every component under a new licence. While some major
contributors might agree to do so, Linus is one of the few with an
absolute veto (as the work of smaller contributors might be weeded
out), and he is exercising that veto.

So no GPL 3 for Linux.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 8:18 UTC (Fri) by Felix.Braun (guest, #3032) [Link] (1 responses)

As Alan Cox has pointed out here, I don't think it's reasonable to expect a final decision on whether there will be a relicensing before the final version of GPLv3 is released. Remember, the text that is currently being circulated is only a first draft. Linus is known to be flexible enough to admit he was wrong or to adapt his opinion to changing realities.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 9:13 UTC (Fri) by atai (subscriber, #10977) [Link]

But Linus is clearly responding in a hot head way. He used the phrase "for example" so the key stuff may not be the only thing that he does not like about GPL v3 (and we don't know what else he objects to). And since the FSF has allocated a year for public commenting on the GPL, now it is time for Linus to make constructive suggestions if he wants to. But he does not try to explain his objections in detail but instead calls a draft "insane" and tries to reject it as if he does not want to see the GPL v3 at all. Should the FSF miss something technically with the attempt to fight DRM in the GPL, Linus can point it out rationally rather than using words like "insane." The FSF is behaving reasonably and Linus is not.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 9:49 UTC (Fri) by atai (subscriber, #10977) [Link] (1 responses)

What you said is not necessarily true. The final outcome may require a court to decide... Alan Cox said:

It isn't clear that this will be a problem. Very few people specifically put their code v2 only, and Linus edit of the top copying file was not done with permission of other copyright holders anyway so really only affects his code if it is valid at all.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Feb 9, 2006 14:18 UTC (Thu) by forthy (guest, #1525) [Link]

Linus also altered the GPL, which is a copyright violation. Read the GPL from top on, and you find right below the copyright notice

Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.

That makes Linus a "pirate", doesn't it ;-)?

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 3:30 UTC (Fri) by huffd (guest, #10382) [Link] (8 responses)

I thought there is a provision that subsequent versions of the GPL must be adopted if the kernel already is at GPL-2? At any rate if this is allowed it will definately hurt the adaptation of GPL-3.

Linus has changed so much since the days the uploaded his toy to a bbs for the whole world to work with, without any license.

I hope his decision this time creates less of a furor than his decision to go with BK, but I doubt it.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 4:39 UTC (Fri) by nicku (subscriber, #777) [Link] (1 responses)

...uploaded his toy to a bbs for the whole world to work with, without any license.

I thought that Linus first applied a more restrictive license to his work than the GPL, yet you say there was none. Can you provide a reference to that?

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 5:58 UTC (Fri) by jmtapio (guest, #23124) [Link]

Quoting the release-notes from 0.01 (ftp://ftp.funet.fi/pub/linux/kernel/Historic/old-versions/RELNOTES-0.01):

Full source must be available (and free), if not with the distribution then at least on asking for it.
Copyright notices must be intact. (In fact, if you distribute only parts of it you may have to add copyrights, as there aren't (C)'s in all files.) Small partial excerpts may be copied without bothering with copyrights.
You may not distibute this for a fee, not even "handling" costs.

So the early license was actually more restrictive than GPL with respect to fees.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 7:46 UTC (Fri) by dlang (guest, #313) [Link] (5 responses)

quote:
I thought there is a provision that subsequent versions of the GPL must be adopted if the kernel already is at GPL-2? At any rate if this is allowed it will definately hurt the adaptation of GPL-3.

think for a little bit about what this would mean if there was such a provision in the GPL. This would mean that it would be possible for some company to take over control of the GNU project and publish any license they wanted and have it apply to your code if they called it GPLv50.

you can argue that it's unlikly for someone to be able to take control of the EFF, but it's not impossible.

the common (and suggested by the GNU project) wording of 'or any later version' implies that the author has personal faith that all future actions of the GNU project for all time will be things that they agree with

personally I think it's silly to agree to licensing terms that will be defined in the future.

however the cost of not doing so is exactly what we are seeing now, to get people to use the new license the programmers must agree that the new license is better then the old one, it doesn't happen automaticaly.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 10:04 UTC (Fri) by gallir (guest, #5735) [Link] (4 responses)

> This would mean that it would be possible for some company to take over
control of the GNU project and publish any license they wanted and have
it apply to your code if they called it GPLv50.

Even so the user/developer can choose which version of the GPL he will
use.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 29, 2006 16:11 UTC (Sun) by kmccarty (subscriber, #12085) [Link] (3 responses)

dlang: This would mean that it would be possible for some company to take over control of the GNU project and publish any license they wanted and have it apply to your code if they called it GPLv50.

gallir: Even so the user/developer can choose which version of the GPL he will use.

Yes, but the real danger (for works licensed under the GPL with the "any later version" option) is not that a future version of the license becomes too restrictive, but that it becomes too permissive. If it becomes too restrictive, downstream developers (Debian, RedHat, forkers) can always continue using the earlier GPL version under which the author originally licensed the work, and the author can explicitly specify use of an earlier GPL version in the work's next release with no harm done. [1] But if GPLv50 is very permissive, say like the BSD license, the author's code could be taken proprietary by some large corporation. Of course the author still owns his/her code, but the corporation would have no obligation to make its changes to the source public.

I don't think this turn of events is too likely, so I've included the "any later version" option in my own code, but in principle the danger is there.

[1] Of course the author could explicitly force the use of the more restrictive GPL version, but then the author always has the right to take his/her own code completely proprietary anyway.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 8:45 UTC (Mon) by Wol (subscriber, #4433) [Link] (1 responses)

Actually, if it's so permissive that it's not copyleft, I think it would be easy(ish) to get gpl-v50 overturned in court as self-contradictory ...

Any new GPL must be "the same in spirit" as the earlier one(s). It's taken as given that the GPL is copyleft. Should someone manage to take over the FSF and remove the copyleft from the GPL it shouldn't be too hard to get a judgment that the new version violates all the assumptions used by authors who released their work under earlier GPLs, and as such the new version would not be a valid "or later" GPL.

Cheers,
Wol

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 13:33 UTC (Mon) by kmccarty (subscriber, #12085) [Link]

Actually, if it's so permissive that it's not copyleft, I think it would be easy(ish) to get gpl-v50 overturned in court as self-contradictory ...

This is a relief to hear; I hope you are correct, or better, that things never come to that state :-)

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 17:10 UTC (Mon) by forthy (guest, #1525) [Link]

Later versions of the GPL can not take away rights granted by earlier versions (if the program in question allows to choose the version). So if GPLv5 doesn't require to ship source, GPLv2 still does, and you are entitled to choose GPLv2, if you like to (and then you are entitled to see the source). That's why a GPLv2+ program can't exercise the patent fight-back clause in GPLv3. Permissive is always permissive to the end user, who gets a license from the original licensor, not from the man in the middle.

I think Linus is right

Posted Jan 27, 2006 7:57 UTC (Fri) by dlang (guest, #313) [Link] (3 responses)

I origionally thought that Linus was mistaken in his statement that GPLv3 would require him to hand out his private keys, but after thinking about it a while I think he's right (at least as far as the current draft goes)

the purpose of the provision is to block things like the Tivo (which will only boot signed kernels, at least until you modify the BIOS)

but what is the real difference between distributing a kernel signed by Tivo and distributing a kernel signed by Linus?

if you say that it's becouse the Tivo kernel has hardware that requires that signature then (takeing things to an illogical, but legal extreme) I could build hardware that required the kernel be signed by Linus and then by the same argument he would have to reveal his secret key (note that to build such hardware I only need his public key)

as for the suggestion that this requirement of the GPLv3 be modified to allow for cases where the key can be changed as an alturnative to revealing the secret key. remember I control the (theoretical) hardware so I can eliminate that option.

now I realize that Linus doesn't commonly distribute kernel binaries, only it's source, but what if gentoo was to get built to only install and compile the source if it is signed by Linus, put this on a otherwise closed box and you are back to the same problem.

or, apply the same logic to rpm's from distros, it's possible to build equipment that will only accept signed rpm's, now by the same argument the distro would be required to had out their secret key so that I would be able to sign rpm's so that they could be installed on this hardware.

if you can close this sort of loophole without opening up one that would allow a company that wanted to avoid allowing people to put their own stuff on the hardware please speak up.

(and think of the case where company A builds the hardware to require signed stuff from company B)

Private key distribution

Posted Jan 27, 2006 9:46 UTC (Fri) by tgb (guest, #745) [Link] (2 responses)

if you say that it's becouse the Tivo kernel has hardware that requires that signature then (takeing things to an illogical, but legal extreme) I could build hardware that required the kernel be signed by Linus and then by the same argument he would have to reveal his secret key (note that to build such hardware I only need his public key)

No, if you wanted to distribute the kernel on that platform, you would be required to distribute Linus' private key. The responsibility would lie with you, not with Linus in this example, as I understand it.

Private key distribution

Posted Jan 27, 2006 18:53 UTC (Fri) by dlang (guest, #313) [Link] (1 responses)

there is a cablemodem out there that downloads the kernel when it boots.

the manufacturer of such a device is not distributing the kernel so how in the world could they be violateing the license.

if such a device were to download the kernel rpm from the RedHat website and refuse to boot unless it's signed by RedHat then the current wording would appear to require that RedHat (who is distributing the kernel) reveal their signing key.

note that a device configured this way would not be a practical thing, but it would be pretty cheap to produce such a device, especially for a company that wants to make trouble for Linux and Open source software.

Private key distribution

Posted Jan 27, 2006 21:26 UTC (Fri) by tgb (guest, #745) [Link]

From the draft, section 0:

To "propagate" a work means doing anything with it that requires permission under applicable copyright law, other than executing it on a computer or making private modifications. This includes copying, distribution (with or without modification), sublicensing, and in some countries other activities as well.

So by causing a kernel to download onto the cable modem, the manufacturer is propagating the binary, which means that they are bound by the GPLv3. By the manufacturer not being able to distribute the signing key, they cannot propagate the binary. The manufacturer is breaking the terms of the GPLv3 in this example.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 9:41 UTC (Fri) by russell (guest, #10458) [Link] (4 responses)

That's fine for Linus but others don't have to follow his lead. For example, the GPLv3 crowd could contribute or relicense their code under GPLv3. This code would have to be stripped out for anyone using the kernel under GPLv2 only. Eventually there would be too much code to strip out, i.e. crippling the kernel or requiring a GPLv2 only fork.

It would only take a minority to impact the kernel this way.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 12:55 UTC (Fri) by copsewood (subscriber, #199) [Link]

Until the lead maintainers are able to accept contributions with a particular license, such contributions are by definition ineffective. Anyone can offer a patch with an incompatible license, but this code is only of use to the private individual offering this patch. This is shooting yourself in the foot, because if your patches can't be accepted you have to reapply your patch every time new features or fixes appear in the mainline kernel which you wish to use together with your patch; essentially you have all the effort of maintaining your private fork of the kernel.

A minority could fork a kernel e.g. to GPL3 only, but for any kernel publicly redistributed, would have to exclude code which was GPL2 only. The minority is inherently at a disadvantage, unless and until it can demonstrate better working code than the majority.

So instead of playing silly games, it would be better to try to identify the substance of Linus' opposition to GPL3, to see whether his critique is valid and if it is, to identify whether this is a bug in the draft that needs to be fixed, or whether it concerns an important feature genuinely reflecting the intentions of the GPL3 draft authors that needs to be left as is.

Nope

Posted Jan 27, 2006 18:12 UTC (Fri) by stevenj (guest, #421) [Link] (2 responses)

For example, the GPLv3 crowd could contribute or relicense their code under GPLv3.

No, they couldn't, because the GPLv3 is not compatible with GPLv2 (in the absence of "or any later version" language). Thus, it would be illegal to distribute a kernel with a mix of GPLv2 and GPLv3 components.

The kernel is effectively impossible to relicense, period.

Nope

Posted Jan 30, 2006 8:52 UTC (Mon) by Wol (subscriber, #4433) [Link] (1 responses)

Actually, I think v3 has been written with the explicit intention of being compatible with v2.

So if the aim was to move forward to v3, all Linus would have to do is change the wording of the COPYING file and say "as of now, v3 is the default licence", and that would be that. The kernel as a whole would be v3, and the v2 stuff would fade. Including v2 stuff in a v3 work is not a problem.

The problem is if linux remains a v2 work. Because of the extra restrictions (DRM) and permissions (patent retaliation) it's difficult, if not impossible, to include v3 stuff in a v2 work.

Cheers,
Wol

Nope

Posted Jan 31, 2006 5:58 UTC (Tue) by jamesh (guest, #1159) [Link]

The only way GPLv3 code can be compatible with GPLv2 code is if the GPLv2 code includes the "or a later version, at your option" clause.

The GPLv2 on its own forbids you from imposing additional restrictions on the work when you distribute it. The GPLv3 includes restrictions not found in the GPLv2 (e.g. the patent retaliation clause). Therefore you need additional permission from the licensor to combine GPLv2 and GPLv3 code into a single work (which is what the "later version" clause many people use is).

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 13:20 UTC (Fri) by sylware (guest, #35259) [Link]

When I look at the GPLv3 I can see that "software patents" (yes those ugly things) are melted into the core flow of the text. That assumes the existence of such ugly things. It is not the case. Parts of the GPLv3 regarding "software patents" should be kind of retired of the main flow in order to make it exceptional and to limit it's validity to geographical regions where they are still enforceable (hope that will change in the futur).
I understand Linus, because if I had such influence, since I don't like the GPLv3 text right now (has some definive nonos), I would do the same.
But we have one year to shape the GPLv3...

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 16:09 UTC (Fri) by X-Nc (guest, #1661) [Link] (22 responses)

GPL 2
GPL 2+
GPL 3
Blah, blah, blah.

This is what happens when you use non-free, restrictive licenses like the GPL. Mind you, the GPL is just as good a license to use any any other, and I like it myself. But it's just to non-free and restrictive to make transitions like this easy. It will always cause a big muddy mess that will take years to sort out.

Using a truly free license like the BSD or MIT license would make this whole issue moot.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 17:01 UTC (Fri) by allesfresser (subscriber, #216) [Link] (2 responses)

Momentary total individual autonomy is not what we're after. Guarding the freedom of future developers and users has the price of vigilance to maintain it, and the GPL reflects that.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 18:31 UTC (Fri) by X-Nc (guest, #1661) [Link]

You misinterpreted my comment. What the GPL stands for is a good and very positive thing. It keeps the code "open" at the price of restricting it's freedom. Nothing wrong with that. My statement was pertaining to the issues brought up my Linus' stance and the posts by others here in this thread. There is and will be some pain involved with the movement to GPL 3 whenever it comes out. There will be licensing issues to deal with and lots of code to filter through. And you know that no matter how good GLP 3 is there will be a good portion of the world that finds something wrong with it. This is the same thing that any of the other restrictive licenses would go through with such a major change in their licenses. If the license was MIT'ish this wouldn't be an issue. There would be other issues to deal with, of course. But not this one. The GPL is very important and much needed. After all, we wouldn't be here in the Linux world without it. It's what made the F(L)OSS world what it is today and assures it will remain so in the future. But the GPL is not free by any means.

Free \Free\ (fr[=e]), a.: Thrown open, or made accessible, to all; to be enjoyed without limitations; unrestricted; not obstructed, engrossed, or appropriated; open

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 6:02 UTC (Mon) by kirkengaard (guest, #15022) [Link]

I have never been so concise saying the same thing. Nice. :)

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 17:30 UTC (Fri) by ballombe (subscriber, #9523) [Link] (1 responses)

Speaking of the MIT/X license, you missed the whole xfree86 debacle
apparently.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 18:35 UTC (Fri) by X-Nc (guest, #1661) [Link]

Heh, no. I didn't miss it. But that is a very good example of what free source code licenses are capable of. The license for xfree86 was changed so the community took the code and X.org was born. Those who wanted to do something less open with the code are free to do it and those who want to do something more open are also free to do so.

There was little or no problems with the split that were caused by the license.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 18:07 UTC (Fri) by Ross (guest, #4065) [Link]

There is only one way to distribute something with absolutely no restrictions (your "truly free"), which is to deliberately and specifically put it in the public domain. Do you remember the transition from the old BSD license to the new one? Was that not messy? Has all of the code been converted, even today?

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 20:26 UTC (Fri) by GreyWizard (guest, #1026) [Link] (14 responses)

Apart from the apt comments above by allesfresser and ballombe, you seem to have overlooked the fact that exactly this problem exists for the BSD style licences. There's the original BSD license with and without the advertising clause, MIT, X and a thousand other little variants. Moreover, each of these imposes some restrictions such as requiring attribution and not using the original author's name for promotional purposes. So in fact none of these licenses are truly free (in the crabbed sense which allows you to call the GNU GPL non-free) either. There are no truly free licenses by this measure. Only works which have expired copyright or which have been released into the public domain can qualify.

Of course, all of this presumes that the true measure of freedom is how much a single person in possession of the code is permitted to do. A better goal is preserving as much freedom as possible for as many people as possible. By that measure the GNU GPL is much more free than any BSD variant because it forbids restrictive relicensing.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 22:13 UTC (Fri) by X-Nc (guest, #1661) [Link] (13 responses)

<sigh>

I know better than to get into these kind of discussions but for some reason I just can't seem to keep out. Ok, last time...

I like the GPL. I use the GPL. I believe in what the FSF is doing. I promote FLOSS with the GPL as the backbone of the entire Linux community. GPL - Good!

You are right that the only truly unrestrictive, unencumbered "license" is Public Domain. There are issues that would, and have, cropped up when BSD/MIT/X licenses are involved. I addressed that in the reply about X.org I posted earlier so there's no reason to go over it again. There is only one point I was making and that is this:

There will be problems that will come up which will be a result of the GPL 2(+) and GPL 3 licenses differences. These will mostly be minor but there will be some that are major (i.e. the whole reason for this thread of comments). These kind of problems would not be encountered if the code were licensed under a non-restrictive, or at least a lesser restrictive, license. If the code were under a less restrictive license there surely would be difficulties, but they wouldn't be of the kind and magnitude as those under the GPL's.

As for this:
> these licenses are truly free (in the crabbed sense which allows you to call the GNU GPL non-free)

The GPL is not free. It is not about freedom of the code. It's about the continued access to that code for future developers and users. The GPL is a highly restrictive license. It is that restrictive access that makes it the good and proper thing that it is now and should (will) continue to be for decades to come. Said developers and users are not free to use the code as they might need or wish to. But the GPL is not about usage of the code. It's more about access to the code. See the definition of the word "free" that I posted earlier.

And on that note this part of the discussion should be closed. There is no rebuttal that anyone can make to what I've said without tearing up the guts of the GPL. If anyone dislikes and does not want to see the GPL proliferate then continue with the definitions and bastardization of the word "Freedom." Just as GreyWizard and others in here have.

I'm off to Haru in March so I need a lot of keiko as I believe I got a promotion to Makushita on the next Bench Sumo Banzuke.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 22:32 UTC (Fri) by GreyWizard (guest, #1026) [Link] (12 responses)

Allow me to observe that this is a relatively open forum. The privilege of closing the discussion does not become yours simply because you declare that any rebuttal is, according to some absurd excuse for reasoning, an attempt to "tear up the guts" of the GPL (whatever that means).

Furthermore, your understanding of the word freedom is simplistic at best. I have not bastardized its meaning by pointing out that you are concerned only with software developers and not to the world at large when you declare the restrictions imposed by the GPL to be non-free.

Finally, you are mistaken when you claim that the kinds of problems referenced on this thread are less severe or missing for BSD style licenses. As I and others have pointed out, examples of such problems and worse are available for anyone to see.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 29, 2006 1:32 UTC (Sun) by mepr (guest, #4819) [Link] (11 responses)

> when you declare the restrictions imposed by the GPL to be non-free.

I just wanted to point out that the above is a non sequitor
:)

a thing which has restrictions is by definition non-free. It is the choice of the expositor to choose whether to emphasize the freedom of potential users of the source code or of the distributors of the software.

The GPL ensures the rights of the large community to the source code at the expense of the the rights of anybody who chooses to distribute binaries based on source code they have modified.

This is therefore a restriction of a person's ability to pour creative energy into a piece of software based on someone else's work and then choose the terms on which she will make that innovation available to others. To wit, it guarantees the freedoms of one group precisely through the means of restricting the freedoms of another.

This is largely an analog to the situation where a person creates remixes based on copyrighted recordings. If the music industry is been an abusive parasite towards the artists, there are still artists that want control over the future use of their work and will never agree to such a definition of freedom.

It is OK by me to refer to the GPL as a Libre license, but there are differend kinds of freedom, and I defend the likewise valid explanation of why the GPL is also a non-free license.

I for one came to linux and gpl software for the free beer and have stayed for the freedom. But some may think it is more free to use software that is free as in public domain or bsd or the Artistic License, and I'm not going to argue with them.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 16:17 UTC (Mon) by GreyWizard (guest, #1026) [Link] (10 responses)

What a pile of pedantic nonsense! The notion that "a thing which has restrictions is by definition non-free" is absurd and untenable. Were it otherwise, the restriction against shouting fire in a crowded theater would mean there is no free speech. The restrictions imposed by BSD style licenses would mean that they too must be declared non-free. And so on. Restrictions are often imposed for good reasons that are not incompatible with freedom.

Securing the freedom of people who might use software by refusing to grant permission to impose further restrictions is indeed what the GNU GPL does, but this is not a net loss for freedom.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 20:06 UTC (Mon) by X-Nc (guest, #1661) [Link] (9 responses)

$DEITY save us!
How can you be so blind.

I have, in nearly every post I've made in this thread, made it clear that I AGREE WITH, PROMOTE & EVANGELIZE the GPL. It is the cornerstone of the F(L)OSS movement and the reason that the last decade has seen such great achievements in application and systems development. Your statement, "Restrictions are often imposed for good reasons that are not incompatible with freedom," is only an echo of what I have been saying this whole time. The only point I made was that the GPL is a restrictive license, which it is and that's a GOOD THING©!

Now please go away and leave rational debate to those who are rational.

That said, I'm going, too. My state of rationality is, and has always been, a point of question for everyone who knows me. Just ask erat.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 30, 2006 20:33 UTC (Mon) by GreyWizard (guest, #1026) [Link] (8 responses)

My statement about restrictions and freedom was made in response to a post by mepr, not you. Nowhere have I made any comment on whether you agree with or promote the GNU GPL. I couldn't care less. I have been responding to your claim that this license is not truly free. That claim is false. Even writing it in bold, italics and all caps won't make it true.

And again, I remind you that this is a relatively open forum. Go ahead and imply that I'm not rational if you find it easier to attack my character than my arguments, but you have no power to exclude me from this debate.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 31, 2006 3:19 UTC (Tue) by X-Nc (guest, #1661) [Link]

My apologies. I got lost in the thread and confused who was talking to whom. There is no excuse for my behavior. My mistake was I thought. "I thought you were directing your comments toward me." My father, who was a great teacher and football coach, would always point out that most mistakes were because of those first two words. "I thought..." His reply was "Don't think!" By this he meant to do your job and rely on your training. Don't go off and ad-lib your position. Well, I am guilty of doing just that. So, officially and formally -

I apologize to GreyWizard for my unfounded and inappropriate flamage and hereby ask forgiveness.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 31, 2006 15:46 UTC (Tue) by sepreece (guest, #19270) [Link] (6 responses)

There are many ways in which we give up some freedoms in return for other benefits. That in no way changes the fact that we are giving up freedoms. The GPL is clearly NOT free, whether you say so or not. In fact, the license says, quite clearly, that it makes certain *restrictions* in order to guarantee other freedoms.

It's important to remember what a license is - it's a grant of certain rights to distribute something you would otherwise have no right to distribute. It's silly to claim that the "most free" license (evaluated solely in terms of the license as a grant of distribution rights) wouldn't be one that gave you the absolute right to redistribute.

However, the goal of this particular license is to make sure that certain other freedoms are preserved (the freedom of the recipients of the software) are preserved. There is a net increase in the freedom reserved to those recipients at the expense of freedom in the license itself.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 31, 2006 16:53 UTC (Tue) by GreyWizard (guest, #1026) [Link] (5 responses)

Wrong. As a person who acquires work created by someone else, you have only the rights that copyright law grants. Accepting a free license like the GNU GPL gives you new rights and requires you to give up nothing. All licenses, including BSD variants, refuse to grant permission for some things so if such restrictions make a license "NOT free" then no license can be free and your entire argument is tautology.

Perhaps when you said "NOT free" you meant to say "less free" in which case you should try saying what you mean. Obviously the GPL grants fewer options in some circumstances than some alternatives as a matter of strategy. That you suppose I don't understand this is evidence of poor reading comprehension on your part. What I don't accept is the notion that "most free" should be evaluated solely in terms of distribution rights.

Common sense recognizes that freedom is not reducible to the right to add restrictions to work other people create and that the increase in freedom copyleft provides can be more important than the lack of permission with which it is purchased. Declaring the GPL less free on the basis of such a myopic perspective is simplistic and silly.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 31, 2006 19:36 UTC (Tue) by sepreece (guest, #19270) [Link] (4 responses)

Actually, "free," to me, means "public domain". No restrictions == free.

Since it's a license that serves no legal purpose but to convey distribution rights (that's all a copyright license can possibly do), then restrictions on distribution are the only metric for "freedom". The particular restrictions that the GPL imposes clearly do enhance the freedom of the recipient, they just don't enhance the freedom of the licensee.

But, if ad hominem arguments are where you're at, I'll leave you to vituperate in peace.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Feb 1, 2006 15:48 UTC (Wed) by GreyWizard (guest, #1026) [Link] (3 responses)

Pointing out flaws in your reasoning hardly qualifies as an ad hominem argument. On the other hand, if that's just an attempt on your part to use fancy words to object to the tone I've taken then consider your own comments, which are every bit as vituperative. Either demonstrate the same level of courtesy you would prefer to receive or develop a thicker skin.

Meanwhile, your emphasis on the freedom of the license itself over the freedom of the recipient is pedantic and pointless. And you're mistaken when you claim that a copyright license can only convey distribution rights. Courts have ruled many aspects of proprietary end user agreements to be binding, for example. Take a moment to read the GNU GPL and you'll discover that it deals with quite a bit more than distribution.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Feb 1, 2006 20:17 UTC (Wed) by sepreece (guest, #19270) [Link] (2 responses)

Sorry, the original point was pedantic and so was all the discussion about it - since "more free" has no useful meaning, we've just been talking about words. I thought that was clear from the beginning.

Attacks on logic aren't ad hominem, "evidence of your poor reading comprehension" is.

I admit to insufficient pedantism in saying "distribution rights" when I meant "rights under copyright." I'm not sure there's a practical difference in terms of the specific discussion, but I was inaccurate.

Note that I DO agree with you that the original post in the string was wrong in characterizing BSD and MIT as "truly free" licenses, though I would say they were "more free" than GPL, judged solely as licenses, rather than on the basis of total net freedom of society...

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Feb 1, 2006 22:04 UTC (Wed) by X-Nc (guest, #1661) [Link]

As the instigator of this mess... I mean "discussion", I want it to be known that my only point is that the GPL is a restrictive license and the whole "Linus says no GPL 3 for the kernel" would be less of an issue if the license was less restrictive than the GPL. I used the BSD and MIT licenses only as an example. They are not "truly free" and if I said that it was a mistake in wording. They would also not be without problems and issues, just that they would be different than the ones dealing with GPL2 <> GPL 3.

I like and support the GPL and F(L)OSS. The real problem is that RMS picked "Free" and "Freedom" as the key words for his movement/cult. No, I can't think of a better word to describe the fundamental essence of the GPL and FSF. But the word free has to many other connotations that cause a LOT of confusion.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Feb 2, 2006 2:51 UTC (Thu) by GreyWizard (guest, #1026) [Link]

"More free" does indeed have a useful meaning for people who apply it sensibly. Only when one focuses on irrelevant and uninteresting questions like how close a license is to the public domain status when isolated from the effects it has on society does it seem meaningless. As a simple example, all else being equal a state in which the government does not censor the press is more free than one without. While obvious, there is nothing meaningless or pedantic about that.

Meanwhile, you are still confused about the meaning of ad hominem, despite the link to a clear explanation I included in my last comment. An ad hominem argument is not merely one that hurts your widdew feewings. For this fallacy to apply, I must first attack your character and then use that attack to dismiss one or more of your claims without further argument. I'm not convinced that pointing out your failure to comprehend the clear meaning of the post you replied to counts as an attack on your character, but suppose for the sake of argument that it does. Which claim of yours did I dismiss on this basis? None that I can think of.

But this is exactly the problem. Rather than depending on the accepted meaning of terms like "freedom" and "ad hominem" you declare, like Humpty Dumpty, that words mean just what you want them to (or sometimes nothing at all). That's every bit as inane on this side of the rabbit hole.

Re: Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 27, 2006 20:36 UTC (Fri) by patipati (guest, #28059) [Link]

it's just to non-free and restrictive to make transitions like this easy.

Transitioning from a MIT or BSD license is trivial because anybody can use that code and redistribute it under any license they want, free or otherwise. There are reasons why this isn't desirable, so these licenses aren't comparable to the GPL (apples vs. oranges). If you know a license which makes free-to-free transitions easy and free-to-nonfree transitions impossible, then I'm interested.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 28, 2006 2:50 UTC (Sat) by sbergman27 (guest, #10767) [Link] (1 responses)

With all this hoopla about how we might change the Linux kernel license, I find myself taking solice in the idea that it is immutable. Going into the future, GPLv2 may not be *optimal*, but it is a close enough approximation of optimal. Do we really *want* the Linux license to be mutable? Or do we want it set in stone such that no matter how many people any hypothetical threatened parties (with lots of cash or other temptations) might try to buy, the Linux kernel license remains the same?

I want it set in stone.

Torvalds: No GPL 3 for Linux (eWeek)

Posted Jan 28, 2006 10:20 UTC (Sat) by copsewood (subscriber, #199) [Link]

In one scenario, we could end up with a Linux which is only modifiable by end users on old hardware, or which will only run on newer all DRM hardware. The license may be the only leverage available which prevents the loss of freedom a general move by hardware suppliers acting in concert with proprietary software suppliers would result in. This is more likely to occur in the case of consumer devices than servers, because the purchasers of the latter kind of hardware are more likely to discriminate against DRM hardware. Given the fact that the commercial advantages of copyleft software is now becoming evident in consumer embedded devices, it seems to make sense to use the possible leverage appropriate anti-DRM provisions in copyleft licenses might provide.