Bzr development stopped

[Stephen J. Turnbull]

Brian de Alwis writes:
 > On 12-Sep-2012, at 1:44 PM, Stephen J. Turnbull wrote:

 > > So what?  That's not the Canonical contributor agreement, which
 > > involves no transfer of copyright.
 > 
 > It is a consequence of the GPL which considers linking (such as
 > with plugins) to invoke a derivative work.  By having ownership
 > over bzr, Canonical can grant an alternative license to itself
 > (and others) to allow linking to proprietary extensions without
 > invoking the GPL.

I don't understand your point in the context of this subthread.  We
are discussing Canonical's contributor agreement as a (non-) inhibitor
of contributions.  Matthieu proposed the attempted bzr/hg merger as an
example where we know this had a major effect.  I'm explaining why
"Canonical owns the copyright" means this is *not* a relevant example.

(Speaking of relevant examples: Curiously enough, neither Ben nor
Matthieu has stepped forward to say "I have patches I haven't
contributed, and the contributor agreement is the main reason why."
If you've just assumed that it goes without saying, it doesn't; please
make it explicit.)

More precisely, the contributor agreement does the same job by having
the copyright owner grant a sufficiently powerful license to
Canonical, *but* it allows the contributor to take their ball and go
home if they get upset with Canonical's behavior in the future.

The most important way to "take their ball" that is permitted to
copyright owners is that they are allowed to accept, even seek,
funding for a fork (including competing for Canonical customers who
use proprietary components contributed by non-Canonical employees) by
giving a similar permissive license to another deep-pockets angel.
This is not possible when you assign the copyright (modulo an
FSF-style automatic grantback clause in the assignment agreement, of
course).

Note that if (as RMS usually does for GNU Projects he controls),
Canonical removes authorship information from the relevant files
(which as copyright owner they are allowed to do), finding the holders
of grantback clauses becomes problematic.  Note that neither Bazaar
nor Mercurial provides git-style separation of author from committer,
so the VCS trail won't necessarily prove anything.

Copyright assignment really is different from the Canonical
contributor agreement.  Not only is it objectively more powerful, but
an unyielding demand for assignment tells me that monopolization is
the intent, and not merely an incidental side-effect.  That doesn't
just enrage free software advocates, that offends economists as well.

Steve