[anon1385]

Somehow I imagine you take a rather different stance when it comes to the many aspects of the GPL or AGPL that haven't been tested in court. On those issues the position of the FSF is very much 'illegal till proven otherwise' (that's if you actually succeed in getting Eben Moglen to give a straight answer).

[jacquesm]

That's one of the most often repeated bits of nonsense regarding the GPL, the FSF has always been more than willing to test these aspects in court it's just that nobody has felt sure enough they would prevail in such a lawsuit that they bothered to follow through on it.

Effectively this is testimony to how well the GPL has been put together from a legal point of view, it is bad contracts, agreements and licenses that are tested in court.

[anon1385]

So you think the FSF/SFLC wanted to go to court over this case, where Moglen was giving private legal advice to linux developers that was very much in conflict with copyright law[1]? I don't.

Uncertainty over what adaption means serves their purposes. Uncertainty about the legality of taking somebody else's code and just relicensing it as GPL also serves their purpose. As does uncertainly over what results in a combined or derivative work with GPL binaries. The more solidly those lines are drawn then the less risk people face when calling GPL code from GPL incompatible code. That means people are more likely to do all the kinds of things that RMS fears (e.g. proprietary IDEs that call out to gcc and all that kind of thing, which RMS was talking about recently in relation to emacs interfacing with gcc).

The problem they face is that ultimately they can't create a license that prevents interoperation with proprietary code without also violating freedom zero. The only real way they have to discourage people from doing that is legal uncertainty. That is why it's very difficult to get a straight answer from the FSF or SFLC about what constitutes a combined work (other than them just claiming everything in the world is a combined work, which is the usual nonsense reply you will get). If they answer the question then it's them flagging up the best way to get around the GPL.

For example the FSF claim that distributing FooApp that links to a user chosen library at runtime and calls frobulate() requires FooApp to be GPL licensed, even if the developer of FooApp doesn't distribute the library with the app (or at all), so long as there exists a library implementing frobulate() that is GPL licensed. If you really press them on this point then they will argue that it depends if other non-GPL libraries also implement frobulate(). Of course that creates a large loophole - just create a very basic crappy implementation of the API and make it available, with the expectation that users will actually use the superior GPL library. So it's worth noting that the FSF doesn't actually accept that argument when it comes the readline and editline. Distributing a binary containing the symbols for the readline API [2] is a GPL violation as far as they are concerned. Yet they have not taken anybody to court over this yet even though the FSF owns the copyright for readline.

You can easily find people with quite different views on what the linking clauses in the GPL mean[3], so it's clear that there is a great deal of confusion out there. Nobody really knows what it means till it gets tested in court and the FSF have no interest in getting this cleared up because there is a non-zero chance that the courts don't agree with the FSF's interpretation of title 17. Once the red line is drawn over what is and isn't a combined work then people will be free to work around the GPL with minimal legal risk.

That much legal uncertainty is not the sign of a well written license, if you think the aim of a license is to clearly enumerate what rights people have. If you think a software license is a political weapon then you probably don't care that some people are being scared away from doing things they have a legal right to do with your code (but that you don't like).

[1] http://marc.info/?l=openbsd-misc&m=118963284332223

[2] e.g http://tuomov.bitcheese.net/b/archives/2005/12/23/T22_53_01

[3] https://news.ycombinator.com/item?id=1877926

[belorn]

I would take Theo de Raadt suspisions with a large grain of salt, especially when he is arguing that BSD licensed software can only be used with other BSD licenses software. The whole point of permissive licensed software is that you can use it under a different license, and many proprietary software companies would get quite angry if they suddenly had to close down shops because Theo has a new theory. They would also likely stop funding bsd projects.