[openstack-dev] [all] The future of the integrated release

Thierry Carrez thierry at openstack.org
Wed Aug 13 09:45:51 UTC 2014


Nikola Đipanov wrote:
> While I agree with motivation for this - setting the expectations, I
> fail to see how this is different to what the Swift guys seem to be
> doing apart from more red tape.

It's not different imho. It's just that nova as significantly more
features being thrown at it, so the job of "selecting priority features"
is significantly harder, and the backlog is a lot bigger. The slot
system allows to visualize that backlog.

Currently we target all features to juno-3, everyone expects their stuff
to get review attention, nothing gets merged until the end of the
milestone period, and and in the end we merge almost nothing. The
blueprint "priorities" don't cut it, what you want is a ranked list. See
how likely you are to be considered for a release. Communicate that the
feature will actually be a Kilo feature earlier. Set downstream
expectations right. Merge earlier.

That ties into the discussions we are having for StoryBoard to support
task lists[1], which are arbitrary ranked lists of tasks. Those are much
more flexible than mono-dimensional priorities that fail to express the
complexity of priority in a complex ecosystem like OpenStack development.

[1] https://wiki.openstack.org/wiki/StoryBoard/Task_Lists

-- 
Thierry Carrez (ttx)



More information about the OpenStack-dev mailing list