On Wed, 21 May 2014 08:16:15 +1000
Dan Callaghan <dcallagh(a)redhat.com> wrote:
Just some food for thought... You might consider adopting a
"Developer Certificate of Origin" (with agreement indicated by a
Signed-off-by footer in git commit messages) like the Linux kernel,
rather than a CLA. A good summary here:
http://lwn.net/Articles/592503/
I hadn't heard of that before, thanks for the link.
The wording of the Linux DCO doesn't give permission to
re-license
but if new code is contributed under GPLv2+ it seems like that should
give you enough leeway to deal with the GPLv3-only stuff.
From what I've been reading, DCO may not be so legally enforceable for
similar reasons why CLAs can be a mess. They do have a good trail of
who did what when to remove problematic code if something does come up,
though.
It also has the advantage of avoiding the dread "CLA"
phrase which in
many people's minds conjures up unpleasant images of copyright
assignment and proprietary re-licensing.
Yeah, my statement about not trying to do anything sneaky was an
attempt to acknowledge that. After reading a bunch of the stuff linked
in the comments on that lwn article, I'm convinced that CLAs pretty
much always have the potential for sneaky-ness and even if they don't,
it's overhead and un-levels the playing field between the
organization/group requiring CLAs and the contributors who sign them.
Tim