I have to say this, just because I don't want this ever to be misused: * As always, this is not the opinion of the ftpteam.
* Don't use this to justify a GPL violation. The other comments on this thread are the correct way to read this. Don't use this to back up silly arguments. I'm just bring it up. On Wed, Dec 11, 2013 at 07:21:18PM +0100, Andreas Metzler wrote: > Hello, > > I think this is a very simple question but I somehow failed at > search^Wfinding: > > We know this is in not distributable: > [GPLv2 program] linked with [LGPLv3+ library] > > How about this, with indirect linkage: > [GPLv2 program] li.w. [LGPLv2+ library] li.w. [LGPLv3+ library] > > One way to look at it is as a huge aggregated work including two parts > with incompatible licenses. > > The other way round is to look at the two pieces on the right and say > that it is perfectly fine for a LGPLv2+ library to be linked against a > LGPLv3+ library without being upgraded to LGPLv3 [1], and therefore > the GPLv2 program can use it. > > Is there a canonical interpretation? Look to the others on this thread for a sane interpretation, but here's another that I think has a useful way of looking at the issue. IMHO, sharing memory shouldn't always force the GPL's licensing on your program[1]. As example, if we have a well defined header file, which contains a general use library with some well defined behavior (say, for doing Crypto, or some common set of system calls), the API it's self ought not be considered a derivitive work of the GPL, since there's nothing original to that implementation. The program can be linked against any number of libraries, and the fact the library is GPL'd shouldn't force linked code to result in a violation of sorts. Put this the other way. What if we re-implemented a BSD library from the headers, and built an API (hell, even ABI) compatable library that's GPL'd. Does this mean when we switch out the system library, we have undistributable software (or force 3ed parties to hand over source?). I can see the argument that says "no" to that. For me, the question isn't "is this transitively linked" - but is it transitively a derived work? I'd say some cases, yes, some, no. Devilishly advocatingly yours, Paul [1]: The FSF disagrees with me, so listen to them. -- .''`. Paul Tagliamonte <paul...@debian.org> | Proud Debian Developer : :' : 4096R / 8F04 9AD8 2C92 066C 7352 D28A 7B58 5B30 807C 2A87 `. `'` http://people.debian.org/~paultag `- http://people.debian.org/~paultag/conduct-statement.txt
signature.asc
Description: Digital signature
