OSNews

What the FSF says is not really relevant. What matters is what the law says. As far as I can tell, this ruling points out that API's (i.e. things like names and signatures) are not copyrightable, only the actual implementation of that API is. Therefore it would be odd that you can end up with a derived work if you only use the API and do not include an actual implementation of that API (which is what dynamic linking does).

So from an engineering point of view, it would seem that when linking dynamically, you would be creating a derived work from something that is not copyrightable. I'm not a lawyer, but my guess is that this no longer makes it a derived work, in which case GPL would be equal to LGPL for dynamic linking (and interpreted languages), at least in the EU. It remains to be seen if that is the case in the USA as well.

Wow, ndrw really unleashed quite the debate here, it turns out I was right this would turn into quite the drama. Unfortunately I think the court left us with more questions than answers. Much of the problem is that the structures and function prototypes present in header files represent the idea of the API and are an expression of the API at the same time. There isn't really any difference between the idea of a function prototype in our minds, and on screen. So which interpretation would the court have us use? That would have been helpful.


Assuming for a moment that structures & function prototypes are not copyrightable (because they're directly representative of the ideas of them), then we might consider them to be a type of "copyright barrier", where two different programs/DLLs can explicitly use the same structures & function prototypes without being legal derivatives. I think this would have huge implications for both proprietary and OSS licenses.

But the court was just too unclear, I think the issue needs to be kicked back in until they take a more assertive stance.