Bzr development stopped

Stephen J. Turnbull stephen at
Mon Sep 3 08:44:05 UTC 2012

Ben Finney writes:

 > One of the biggest points holding back development is the barrier of a
 > one-sided contribution agreement.

I really don't believe that.  It would be easy enough to prove though:
Fork!  If it's really a big barrier to development, you'll get a huge
round of applause and a deluge of code contributions.

I don't think you will, though.  Bazaar, while it is the preferred VCS
of some highly skilled hackers, is basically a product aimed at people
who either can't or won't put thought into their VCS.  They'd rather
switch than hack, and in order to keep satisfaction levels high,
Canonical pays salaries to a number of the primary contributors.

I wonder how many people reading this post have patches (to any
Canonical product covered by that agreement) that they have not
contributed yet but would be more likely to do so if the agreement was
changed in this way?  (Please reply to me and I'll summarize.  The
results of "bzr diff | wc" would also be appreciated, against an
unmodified bzr tree.)

 >     <URL:>
 > It's great to have Bazaar under the GPL, ensuring further
 > redistribution is always under the GPL. But Canonical is not
 > content with equal treatment and insists code contributors give
 > them special rights to relicense under non-free terms

I suspect that if you ask the legal beagles, Canonical does not
request those rights to permit release of a proprietary Canonical
products.  Rather, they want to be free to mix any software they
develop into improvements of proprietary software already in use by
their consulting clients (ie, without triggering the GPL on 3rd-party

 > – breaking the equality essential to software freedom.

"Essential"?  I don't think so.  I am free to use only GPL Bazaar, and
I do.  None of the projects that I use that use Bazaar use a non-free

And it's not just me.  The FSF standard assignment agreement is also
unequal -- authors, but not the FSF, are always allowed to relicense
under non-free terms, just like Canonical under the Canonical
agreement.[1]  Not to mention that software licensed permissively,
licensing anybody to enforce unequal terms on their version, is also
free.  This doesn't bother the FSF, so I think you should retract the
word "essential".

I do, however, think that Bazaar should lose its GNU project status as
long as the contributor agreement allows Canonical to release non-free
versions no matter what the rationale.  But that's up to RMS, not me.

 > Me too, agreed on all points. I dearly hope Canonical can be
 > convinced – ideally by its developers and employees – to release
 > its contributors from the agreement

That would be nice, and probably smart in the long run.

 > that keeps development of Bazaar (and other software) a funnel of
 > exclusive rights to Canonical.

And a funnel of non-exclusive rights to the 7 billion inhabitants of
planet Earth.  I don't think that's such a bad trade.  Of course, we
should encourage the 7 billion to use only free versions.

[1]  Of course I recognize the difference here, in that the FSF
agreement diffuses market power away from the FSF, while the Canonical
agreement concentrates it in Canonical.  Nevertheless, both reserve
the right to make non-free versions to one side only.

More information about the bazaar mailing list