Beets: MIT License

[clounie]

Problem

Curious - if beets is using Mutagen (GPLv2), then how can it validly state that it has an MIT license? Am I missing something here?

[sampsyo]

IANAL, so I don't have an authoritative answer to this question. But from my completely uninformed perspective, we're not distributing Mutagen code at all, so I don't see beets as derived work. The distribution does include instructions to obtain and install Mutagen yourself, but we don't package any dependencies or anything. This would be a problem if we, for example, bundled up a complete executable package for Windows and offered it for download—that whole would seem to be a derived work and need to be licensed under the GPL, even though the pieces are dynamically linked.

A little googling suggests that there is not a definitive answer to this question—it seems to be a gray area in the GPL. This thread, for example, summarizes some of the disagreement: https://stackoverflow.com/questions/6748333/import-a-library-that-imports-a-gpl-library

[alfiepates]

Yeah, IAANAL but at no point does Mutagen code actually mix with code distributed by beets.

It'd be different if beets offered a tarball/etc with all the deps included, but as of right now that isn't the case (and is unlikely to be the case, considering how easy installing deps is these days).

Again, IAAANAL but this is not what GNU would be able to call a "derivative", as much as they'd _like_ to make everyone use GPL.

[clounie]

Makes sense - thank you for the replies.