every other major copyleft *forces* you to allow later license versions

Posted Oct 4, 2006 8:40 UTC (Wed) by himi (guest, #340) [Link]

That is, except for "or later" loophole. That one allows parasitic forks that can pull from original codebase without permitting to merge back. _And_ it allows for GPLv2-only forks, to serious displeasure of FSF.

BTW, if you have "v2 or later" project, no extra restrictions of GPLv3 have desired effect until you get GPLv3 (or "v3 or later") fork and manage to kill development of the original. As long as the project remains under "v2 or later", anti-DRM, etc. clauses are trivial to bypass by saying "I'm distributing under GPLv2, as allowed by project license". It's not about conspiracy theories, etc. - that's simply how "or later" part of license is written.

I'm not sure I follow this argument - unless you're somehow postulating that the people doing your v2 only fork can relicense the code, I'd get it under the "v2 or later" clause, too, which means that they can apply their DRM, but when they distribute it to me I can request the source and appropriate keys under the anti-DRM clauses of the GPLv3.

So there are two possibilities: pressure authors of the original to relicense under v3/v3-or-later *OR* fork, relicense the fork that way yourself and try to make the original branch wither.

Now, put yourself in position of people who do not want to drop v2. They are going to realize that v3 fork is coming. Moreover, that fork will be able to pick their improvements, but not the other way round. About the only response other than "give up" is "curse, accept that v2-or-later is not sustainable anymore and start protecting new code with v2-only". And that's exactly what we are starting to see. And will see more as more people see the writing on the wall and decide to do something about that.

You're making an awful lot of noise about relicensing code left right and center, but you're ignoring something quite important: only the owner of the copyright on a piece of code can change the license on that code. That means that some random person coming along and creating a fork of a project under the "GPLv2 or later" license has to stick to that license on every line of code in that tree, except the stuff they write themselves. And if a person or a group is able to relicense a body of code, they own that code, and they have every right to do whatever they want with it, including taking it closed source, proprietary, and charging people millions of dollars a seat. They can put the code they write themselves under whatever license they want, whether it's a patch to some GPLv2, v3, or v20 code, but they can't simply change the terms of the original license grant from "GPLv2 or later" to "GPLv2 only" or GPLv3 only", and without that license change the only issue is the license on their modifications.

If a project starts to accept major changes under a GPLv3 only license then your horror story of a v3 only fork may be possible, but only if that fork either convinces other contributors to release their code to it under a v3 only license, or if they replace the code that's still 'tainted' with the v2 grant; likewise for a v2 only fork. But frankly, that's the project lead's own stupid fault for not taking sensible precautions about license compatibility, and it's not something to make such a big song and dance about

himi

Posted Oct 4, 2006 11:56 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

GPLv2-or-later is a no-go now for DRM proponents because of the GPLv3 forking potential, and it's a no go for DRM oponents because of the impossibility to enforce GPLv3 terms. Both camps agree on killing it.

The first fork option is offensive to people who have contributed believing in the FSF ideals, and expecting a GPLv3 someday should the GPLv2 prove insufficient to enforce them. The second one is offensive to those who have contributed based on a careful analysis of GPLv2 terms, and how they could workaround FSF ideals while complying with the GPLv2. Both forks can be taken as a betrayal.

So we have a polarization. We have hostile forks. It's easy to blame the FSF and the GPL v3 process for it.

The root cause however is the people who managed to dissocate the GPLv2 wording from the GPLv2 original intent by building "clever" business models. They knew perfectly well then they were sitting on the wishes of some GPLv2 contributors. Their actions were divisive, and their natural result (GPL update and pressure to drop the GPLv2) should have been obvious to everyone.

The free software community will have to choose sides now. Including all the people who didn't care one way or another. But make no mistake, GPLv2-only is not the neutral choice. It ceased being so once people mixed software patents and DRM in the software distribution process. That's when the community consensus was broken.