On Sun, Jul 25, 2004 at 09:15:44AM -0700, Josh Triplett wrote: > > More clearly (according to my understanding), the resulting binary > > is--it pulls in pieces of readline--but the source is not. (I'm not sure > > if this impacts your point, but it's an important distinction.) > > That's debatable. If your program is written against a library, and > there is only one implementation of that library, I would argue that the > source is a derivative of the library as well. Things get more complex > if there are multiple implementations, of course.
I think I disagree. I think a binary is arguably a derivative work because 1: it pulls in real code during execution, and 2: if I'm not aware of an alternative, I clearly intend for that to happen. On the other hand, source generally doesn't contain any actual content from the library. (I suppose identifier names could be argued, but that seems weak.) It isn't a GPL violation, I believe, for me to have a program which links against a GPL library and a GPL-incompatible library, as long as I don't distribute binaries which do both at once. This seems in line with the above: the derivative work is created at compile time. -- Glenn Maynard
