GPLv3 *compatible* terms [LWN.net]

GPLv3 compatible terms

Posted Jun 29, 2012 15:27 UTC (Fri) by landley (guest, #6789)
In reply to: GPLv3 *compatible* terms by Wol
Parent article: Linksvayer: 5 years of GPLv3

I switched my toybox project from GPLv2 to BSD last year because I think splitting the copyleft pool into multiple incompatible factions has rendered it useless. Everybody mocked Sun's CDDL but when the FSF shipped a new copyleft license the Linux kernel would never use, that was somehow different?

And yes, real projects are hurt by this. Look at qemu: it wants to copy driver code from the linux kernel and platform definition code from binutils/gdb, and it _can't_ because they're under incompatible licenses. Their newer stuff (like the tcg subsystem) is BSD licensed.

This guy is claiming that driving me _off_ the GPL entirely is somehow strengthening the GPL.

What is it with the FSF and Star Wars quotes? GPLv3 was introduced with "I am altering the bargain, pray I don't alter it any further", and the current status is "If you strike me down, I shall become more powerful than you can possibly imagine".

to post comments

GPLv3 compatible terms

Posted Jun 29, 2012 16:26 UTC (Fri) by nix (subscriber, #2304) [Link] (2 responses)

Everybody mocked Sun's CDDL but when the FSF shipped a new copyleft license the Linux kernel would never use, that was somehow different?

Well, yes. Kernels don't share much code with non-kernelspace projects whether or not the license is compatible. Code sharing in either direction between the Linux kernel and other GPLv2-only projects (nearly all of which are in userspace) is minimal.

There are significant projects out there which are not the Linux kernel. Some have shifted license. Some have not. The world does not appear to have ended, nor even slowed in its rotation.

GPLv3 compatible terms

Posted Jun 29, 2012 17:13 UTC (Fri) by dark (guest, #8483) [Link] (1 responses)

Sadly, the world did slow in its rotation. It now completes a revolution 85 μs slower than in the summer of 2007. Whether this was entirely due to the GPLv3 is of course unclear.

GPLv3 compatible terms

Posted Jun 29, 2012 19:11 UTC (Fri) by bryan (guest, #64696) [Link]

And yet they have to insert another leap second on June 30 2012.

You can ignore the difference, but it's kind of obvious

Posted Jun 29, 2012 18:33 UTC (Fri) by khim (subscriber, #9252) [Link] (16 responses)

Everybody mocked Sun's CDDL but when the FSF shipped a new copyleft license the Linux kernel would never use, that was somehow different?

Well, sure. Kernel developers had no way to anticipate CDDL and thus no way to make their stuff compatible with it. They had a way to anticipate GPLv3: the GPLv2 itself talks at length about "revised and/or new versions of the General Public License" and explains how to support newer versions. It also explains that stuff released under "GPLv2 or later" will always be available under GPLv2 no matter what - but that new FSF's stuff can only be released under GPLv3. And it was not even a theory back then: FSF already did one such switch (from GPLv1 to GPLv2). IOW: the kernel developers made conscious decision to make kernel incompatible with binutils/gdb/etc - it's not as if they were not warned.

GPLv3 was introduced with "I am altering the bargain, pray I don't alter it any further"

Well, you may not like it, but why then you've adopted the GPLv2 in first place? GPLv2 was introduced in the exact same manner, after all.

You can ignore the difference, but it's kind of obvious

Posted Jun 30, 2012 11:42 UTC (Sat) by Wol (subscriber, #4433) [Link] (15 responses)

Well, you may not like it, but why then you've adopted the GPLv2 in first place? GPLv2 was introduced in the exact same manner, after all.

The difference is that the change from v1 to v2 (I wasn't around at the time) was widely perceived as fixing bugs in v1. I'll freely admit v2 is buggy, and needed fixing, but v3 is perceived as introducing NEW things, not fixing old bugs.

One quick example of a bug in v2 - if you put source and binary next to each other on your website as SEPARATE downloads, you trigger the "make source available for three years" clause. This *feels* wrong, and the FSF say it wasn't the intention, but that's what the black letter of the licence says.

Cheers,
Wol

You can ignore the difference, but it's kind of obvious

Posted Jun 30, 2012 22:50 UTC (Sat) by pboddie (guest, #50784) [Link] (14 responses)

v3 is perceived as introducing NEW things, not fixing old bugs

Let us ignore the perceptions and focus on what it actually did, then, because it certainly did tidy up a lot of the ambiguity in GPLv2.

Going beyond the fixes, the "controversy" appears to originate from the fact that version 3 of the GPL was written in a way that prohibits vendors from simultaneously shipping GPL-licensed software to end-users while denying them the right to modify that code and run the result on the same hardware, and from holding patent threats over the heads of end-users in order to deny them the right to redistribute the code in question. That these technical and legal licensing evasion tactics were addressed in a revision of the licence shouldn't really be much of a surprise to most observers familiar with what the FSF is about.

I suppose some people get upset about such matters because they believe that getting their code into blockbuster products and not upsetting "risk-averse" corporate stakeholders is more important than what the end-user gets to do with their code, but one can't honestly expect the FSF to uphold those priorities.

You can ignore the difference, but it's kind of obvious

Posted Jun 30, 2012 23:43 UTC (Sat) by raven667 (subscriber, #5198) [Link] (13 responses)

Hey, nice to talk to you again.

I agree with your assessment. I think though that in their PR the FSF has promulgated some myths about the GPLv3 that they think are important for keeping hardware open. I don't think anyone wants to end up in a dystopia where all hardware is boot locked to only Apple, MS or Google. The confusion in many recent articles about the compatbility between GPLv3 and boot time signature checking and some of the bad FAQs on this issue show the results of their PR, intentional or not.

You can ignore the difference, but it's kind of obvious

Posted Jul 1, 2012 0:39 UTC (Sun) by cmccabe (guest, #60281) [Link] (12 responses)

The thing is, when you're dealing with legal matters, the "perception" among judges and legal scholars IS the "reality."

For example, most parts of the "Patient Protection and Affordable Care Act" (also known as "Obamacare") were recently perceived as constitutional by the Supreme Court. So the law is now on a much firmer footing than it was a month ago, even though the law itself didn't change. The perception did.

In the same way, all it takes is a few judges who DO believe that the GPLv3 prohibits boot time signature checking to make that the reality. I am not a lawyer, but the license as written does seem to me like it could be interpreted that way.

I really believe that people who use complex and poorly understood licenses are doing themselves a disservice. Political activism through licenses just doesn't make sense.

The worst part of the whole situation is that most of the licenses, software and otherwise, that we are forced to agree to on a daily basis are not written by open source hackers, but rather by companies like Apple, Microsoft, and so forth. Apple has already set terms for their app store which forbid GPL'ed programs. It wouldn't be a big surprise if any Microsoft app store had the same kind of terms. We should *all* be hoping that complex and EULA-like licenses are ruled unenforceable, not contributing to the problem.

You can ignore the difference, but it's kind of obvious

Posted Jul 1, 2012 16:09 UTC (Sun) by forthy (guest, #1525) [Link] (11 responses)

"Political activism through licensing doesn't make sense".

Well, the GPL always was political activism through licensing. Always, from GPLv1 on. And the GPLv3 fixes the bugs in GPLv2, which were so many that it required a complete rewrite. That some dumbasses like Linus Torvalds are thick and slow, and take ages to "get it" doesn't mean they are right. The GPL is political, it always was, it always will be. Get over it, if you don't want to be political, release under BSDL.

You can ignore the difference, but it's kind of obvious

Posted Jul 2, 2012 5:46 UTC (Mon) by cmccabe (guest, #60281) [Link] (10 responses)

There's no need for name-calling. We're all reasonable people here.

I think you know full well that using a license doesn't require that you agree with the political beliefs of the person who drafted the license. If it did, we would all have to write our own licenses, because consensus is hard to come by in politics.

As for the rest of the issues, Linus and some other people have discussed them all here: http://lwn.net/Articles/200422/

Politics is consensus

Posted Jul 4, 2012 13:21 UTC (Wed) by man_ls (guest, #15091) [Link] (9 responses)

If consensus was so hard to reach in politics, we would all be creating our own political parties. Even worse, we would all be living as hermits. In fact, politics is all about reaching consensus.

Licenses can only require that people agree to the trade-offs expressed in them. For example, when Apple released WebKit under the GPL they agreed (grudgingly) to the political beliefs behind it, or at least convened that the benefits outweighed the cons. But WebKit was based on KHTML which left little margin to Apple.

In the case of releasing new software, choosing a license requires at least that you agree with its explicit goals. You may think that the GPLv2 preamble is not political, but I would disagree: it is highly political, and also it is easy to agree with it. Of course nobody has to agree with Stallman on everything and like the same things just because they use the GPL. But it would be foolish to release new software under the GPL unless you agree that "people should be able to share, study and use software freely".

Politics is consensus

Posted Jul 4, 2012 13:34 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

WebKit is LGPL.

Politics is consensus

Posted Jul 8, 2012 11:00 UTC (Sun) by man_ls (guest, #15091) [Link]

Sorry for that. Anyway, the LGPL has the same political background as the GPL (in fact LGPLv3 is GPLv3 with a few additions); I think the argument with respect to Apple stands.

Politics is consensus

Posted Jul 5, 2012 13:49 UTC (Thu) by mpr22 (subscriber, #60784) [Link]

If software's direct contribution to one's business is as a cost centre (e.g. because what you sell is widgets, not packaged software), then releasing one's software under the GPL may make sense for entirely pragmatic/selfish reasons whether one thinks Richard Stallman's politics pertaining to software to be despicable and wrongheaded, or noble and correct.

Politics is consensus

Posted Jul 7, 2012 23:21 UTC (Sat) by cmccabe (guest, #60281) [Link] (5 responses)

I'm pretty sure that Apple doesn't agree with the political beliefs of Richard Stallman, even grudgingly. You can't even distribute a GPL'ed application on the Apple App Store.

WebKit was not created by Apple, and is LGPL, not GPL. Apple came up with its own open source license, the APSL, which it uses for certain things. It is not GPL compatible.

Politics is consensus

Posted Jul 8, 2012 11:06 UTC (Sun) by man_ls (guest, #15091) [Link] (4 responses)

In fact WebKit was created by Apple; based on KHTML, but it is a different project. The LGPL (sorry about that) embeds the same political beliefs as the GPL; it was also created by Stallman, so the argument doesn't change.

Apple agreed with Stallman's beliefs to the extent that they thought that using WebKit (and releasing source code under the LGPL, therefore helping their direct competitors) was better than developing their own rendering engine from scratch, which they surely could have done. Politics makes strange bedfellows, as they say.

Apple doesn't believe that the trade-off holds for everything, and Apple takes advantage of everything that they can; we might think that they are cynical, but I don't see how we might be surprised.

Politics is consensus

Posted Jul 8, 2012 12:01 UTC (Sun) by hummassa (guest, #307) [Link]

> Apple agreed with Stallman's beliefs to the extent that they thought that using WebKit ([...]) was better than developing their own rendering engine from scratch, which they surely could have done.

The problem is that Apple did not (at the time) have the resources to deal with Yet Another HTML Rendering Engine. Trident, Gecko, Presto and KHTML/WebKit were the most used ones at the time for testing by webdevs, Trident and Presto are closed-sourced and Gecko was at the time MPLd -- so LGPLd KHTML was the natural choice for them (IIRC they tried to buy Opera at the time in order to acquire Presto, but my memory can be playing tricks with me, and a quick google only brings "facebook wants to buy opera" stories)...

Politics is consensus

Posted Jul 10, 2012 6:18 UTC (Tue) by cmccabe (guest, #60281) [Link] (2 responses)

Just repeating over and over that "Apple agreed with Stallman's beliefs" doesn't make it true. Did you read Stallman's obituary for Steve Jobs?

Politics is consensus

Posted Jul 11, 2012 13:06 UTC (Wed) by marcH (subscriber, #57642) [Link] (1 responses)

It looks like you've misquoted "...agreed to the extent that..." into just "agreed" when the whole point of man_ls was about the "extent".

I am afraid this is damaging the credibility of everything else you wrote (which looked interesting)

> Did you read Stallman's obituary for Steve Jobs?

This is getting worse... "X agrees with Y" does not imply "Y agrees with X".

Politics is consensus

Posted Jul 11, 2012 17:41 UTC (Wed) by cmccabe (guest, #60281) [Link]

I would characterize Apple as an occasional supporter of Open Source and an opponent of Free Software.

I feel that the way that man_ls phrased things was confusing (but of course, you may disagree.)