the author claims that releasing derivative works of Shen under the GPL is copyright infringement
Do you have a source for this? The closest I could find was this[1]:
Appropriating other people's work by copying the code and changing the license is IMO immoral.
If this is indeed the source material, then I think your characterization isn't _necessarily_ correct. Saying it's immoral isn't the same as making a claim about law.
The link provided[2] does indicate that de Raadt thinks it is a violation of copyright law however. Finally, I question whether there is actually a consensus on this matter at all. If you relicense BSD code as GPL, then consumers needn't abide by the GPL; they can simply opt to use the original license. Does that really mean a relicensing has occurred? Of course, in non-trivial cases, there's likely additional code that has been intermixed with the original code, which would make it impossible to separate. But the release of the GPL code would still technically need to comply with the BSD license itself. Even at that point, it doesn't feel like "relicense" is actually correct.
(Note that I picked up on this because I firmly believe that such relicensing is indeed unethical, while simultaneously not knowing whether such relicensing is definitely a violation of copyright law or not, while also believing that copyright law shouldn't exist at all.)
If this is indeed the source material, then I think your characterization isn't _necessarily_ correct. Saying it's immoral isn't the same as making a claim about law.
At the bottom of that page he says "my library work is BSD... I do not give permission to relicense my work under GPL", which I think carries my point since most people would understand the former to mean the latter. I think I remember a mailing list post talking more explicitly about derivative works, but I can't find it now and could easily be misremembering.
If you relicense BSD code as GPL, then consumers needn't abide by the GPL; they can simply opt to use the original license. Does that really mean a relicensing has occurred?
People who wish to distribute derivative works based on copyrighted code need a license to do so. Some jurisdictions require licenses to be conveyed in writing, others allow oral or implicit transfers. I'm not sure any jurisdiction would be willing to acknowledge the conveyance of a license when there had been zero interaction between the parties. So it seems like if A were to license some code under the BSD license to B and B were to license it under some more restrictive license to C (which the BSD license certainly grants B permission to do), then in theory C doesn't necessarily have a BSD license to the code. Of course in practice it's moot since if B hasn't made any copyrightable changes then the only person who could ever sue C for copyright infringement would be A.
the release of the GPL code would still technically need to comply with the BSD license itself. Even at that point, it doesn't feel like "relicense" is actually correct.
Technically the combined work (which multiple contributors hold copyrights on) can only be distributed in compliance with the requirements of both the GPL and the BSD license, sure. But since the requirements of the GPL are a superset of those of the BSD license, for all practical purpose the whole is GPLed.
I'd say there's a strong consensus that you can form a work that combines (3-clause) BSD-licensed and GPL-licensed content and distribute the result under the GPL - the FSF has been publishing its list of "GPL-compatible" and "GPL-incompatible" licenses for decades now, and copying BSD-licensed libraries into source trees (GPL and otherwise) without changing the license of the whole for a similar length of time. I'd agree there probably isn't any kind of clear consensus about what happens in the case where there is no substantive GPLed contribution in the first place, because that doesn't seem like a case that would come up very often. (To me it seems obvious, but I guess to other people some other interpretation might seem equally obvious)
Comments
Do you have a source for this? The closest I could find was this[1]:
If this is indeed the source material, then I think your characterization isn't _necessarily_ correct. Saying it's immoral isn't the same as making a claim about law.
The link provided[2] does indicate that de Raadt thinks it is a violation of copyright law however. Finally, I question whether there is actually a consensus on this matter at all. If you relicense BSD code as GPL, then consumers needn't abide by the GPL; they can simply opt to use the original license. Does that really mean a relicensing has occurred? Of course, in non-trivial cases, there's likely additional code that has been intermixed with the original code, which would make it impossible to separate. But the release of the GPL code would still technically need to comply with the BSD license itself. Even at that point, it doesn't feel like "relicense" is actually correct.
(Note that I picked up on this because I firmly believe that such relicensing is indeed unethical, while simultaneously not knowing whether such relicensing is definitely a violation of copyright law or not, while also believing that copyright law shouldn't exist at all.)
[1] - http://www.marktarver.com/fsf.html
[2] - http://undeadly.org/cgi?action=article&sid=20070913014315
At the bottom of that page he says "my library work is BSD... I do not give permission to relicense my work under GPL", which I think carries my point since most people would understand the former to mean the latter. I think I remember a mailing list post talking more explicitly about derivative works, but I can't find it now and could easily be misremembering.
People who wish to distribute derivative works based on copyrighted code need a license to do so. Some jurisdictions require licenses to be conveyed in writing, others allow oral or implicit transfers. I'm not sure any jurisdiction would be willing to acknowledge the conveyance of a license when there had been zero interaction between the parties. So it seems like if A were to license some code under the BSD license to B and B were to license it under some more restrictive license to C (which the BSD license certainly grants B permission to do), then in theory C doesn't necessarily have a BSD license to the code. Of course in practice it's moot since if B hasn't made any copyrightable changes then the only person who could ever sue C for copyright infringement would be A.
Technically the combined work (which multiple contributors hold copyrights on) can only be distributed in compliance with the requirements of both the GPL and the BSD license, sure. But since the requirements of the GPL are a superset of those of the BSD license, for all practical purpose the whole is GPLed.
Sure, I guess I can't really dispute any of that, but I'm not sure I buy that what you've written has any kind of strong consensus backing it.
Anyway, I do agree with your larger point. I'd stay away from this language for a variety of reasons.
I'd say there's a strong consensus that you can form a work that combines (3-clause) BSD-licensed and GPL-licensed content and distribute the result under the GPL - the FSF has been publishing its list of "GPL-compatible" and "GPL-incompatible" licenses for decades now, and copying BSD-licensed libraries into source trees (GPL and otherwise) without changing the license of the whole for a similar length of time. I'd agree there probably isn't any kind of clear consensus about what happens in the case where there is no substantive GPLed contribution in the first place, because that doesn't seem like a case that would come up very often. (To me it seems obvious, but I guess to other people some other interpretation might seem equally obvious)