How can we ensure Bazaar (bzr) remains active?

[Daniel Vedder]

I would like to emphasize the quality of the Bazaar code base - when I
tried to get a grasp of how things worked, I was amazed by how
well-written the code was, and how much documentation (both for users
and developers) was available. I think Bazaar is a very well-engineered
product, which makes its decline even sadder.

Right now, one of the things that I think has been turning new
contributors away is the long lag between submitting a patch and getting
it accepted (currently 73 branches are waiting for a merge into lp:bzr).
I understand we had a technical problem there, not sure if it has been
fixed by now, but these seventy branches have been sitting there for
several months.

I don't think that the battle is completely lost, if the Bazaar
development team were to be reinvigorated and enlarged I think it might
gain in popularity again, playing on its strength of being easier to use
than git.

Another big problem we'd still be likely to face though is the dominance
of Github vs. Launchpad. While Launchpad is definitely functional, it is
by no means as snazzy as Github, and, what's more important, not as
well-known or popular. Everybody tells you to have a good Github profile
in your CV, I've never heard anybody say that about Launchpad. Apart
from the Ubuntu (and derivatives) world, most people seem to be putting
their open source projects on Github. And Github happens to be using
git, not bzr.

P.S. Nice to see some activity on the mailing list again :-) It had
become somewhat quiet here...

On 22.09.2015 05:09, Stephen J. Turnbull wrote:
> Kevin writes:
>
>  > What is it with bzr being so left behind?
>
> As open source projects go, it was hard to contribute to.  The code
> itself is deeply layered and not easy to understand for "drive-by
> contributors".  And the QA process was so burdensome that when
> development was active, they had a "patch pilot" who basically spent
> half-time or more guiding third-party contributions through and around
> Scylla and Charybdis.  There were several (more than two, I forget how
> many there actually were) paid developers working at least half-time;
> it was hard for "casual contributors" to keep up (and a few who did
> were hired by Canonical).  There was a real social divide between the
> core team (mostly Canonical employees) and the users.  Not in a us v.
> them sense, just there weren't very many casual contributors in the
> middle to pick up the ball when core developers became less active.
>
> Then over a period of a few months, Canonical transferred all the paid
> developers to higher-priority projects.  Oof! as the bad guys say in
> Batman comics.
>
> On the plus side, the core team was *extremely* competent (the product
> is very high quality on code quality metrics as well as featureful)
> and as responsive as any open source team I've seen to bug reports.
> You have a really good, clean code base to build on, and copious
> internals (design) documentation.
>
> Of the old core, a couple (Jelmer Vernooij and Robert Collins) still
> post to this list.  It's a good bet they'll help out with technical
> advice and occasional patches.  Others would probably reappear as
> casual contributors if the project became active.  So there are
> (likely to be) valuable human resources new developers can access, as
> well as the code base and documentation.
>
>
>