Open Stack

Hi, fellow TC members,

After a noisy ML thread, opinionated blogposts, and strawmen on Gerrit,
we are at a stage where we seek potential convergence between TC members
around a common proposal (before we put it in words and RFC to the wider
community). This phase started as informal in-person discussions in
Paris, and we set up TC 5-members hangouts (2 so far) to continue that
discussion over a high-bandwidth medium. This email summarizes the
progress so far for everyone to know.

Note that there is little point in commenting on this thread, the
discussion is still very much at its early stages and we don't know yet
what the final proposal will be (nor if it will be truly consensual
amongst the TC members). At this point I prefer we continue to solidify
it in high-bandwidth discussions between TC members.


A. Apparently-consensual points (so far):

A1. A big tent for the "OpenStack Community"

So far we seem to all agree that we need to allow more projects in "the
OpenStack Community", to allow competition, diversity, and allow
projects produced by members of our community to truly exist. Projects
need to be at least vaguely aligned with the OpenStack mission, and
adhering to the OpenStack way (the 4 opens in a large sense) -- which
should be objective criteria. Even if we want to allow competition,
obvious and gratuitous duplicates/forks could be argued to not follow
the "open development / open design" values, and exceptionally be
rejected on that account. The TC would keep ultimate ownership of that
(even if we can delegate most of the checks), and be able to clean up
dead projects.

A2. Horizontal projects will reset expectations

Horizontal projects (release management, infra, QA...) naturally can't
be expected to directly handle ALL projects in the big tent. However
they need to empower all projects by providing general processes, tools
and advice to everyone. They all need to describe how their work would
evolve in a bigtent new world order. They may still choose to directly
handle some projects (think: ensure common synced release for the
release management, writing the doc itself for documentation) if that
makes sense and they can handle it.

B. Maybe consensual points:

B1. A tag-based taxonomy to navigate the big tent

It is still the duty of the TC to help downstream consumers of OpenStack
to understand what each project in the big tent means to them, and have
a rough idea of its status. The TC would define a number of tags that
can be applied to projects in the big tent, and the rules to apply them
(some tags could be set by the TC itself, some others could be delegated
to other groups). Some tags could be used to answer the Foundation Board
questions under the bylaws (like defining a set of projects that the
board can pick from to build a specific trademark license program). But
most other tags should just facilitate navigating and understanding the
status of projects in the big tent (like an "translated" tag that the
I18N team would apply to project with a satisfying degree of
translations coverage).

C. Non-consensual points:

C1. Special-case a "Compute Group" in the governance

The idea that we should codify an opinionated "Compute Group" and make
it the focus of the TC activities was consensual in the first group, but
had strong opponents in the second group. So we need to continue the
discussion on that one to see if TC consensus can be reached. I guess we
can't prevent some TC members to care more about a fundamental set of
projects, but do we need to carve it in stone in our charter ? Wouldn't
the taxonomy be enough to cover the "foundational" aspect of certain
projects (Keystone is obviously foundational, but do we need to
special-case it in our project structure itself?)


PS1: We haven't reached out to all TC members yet, and we still need to
figure out how to handle some limited resources in a bigtent world, so
this is still very much WIP. Personally I'm working on proposals on how
to cover design summit space and trademark checks in a bigtent world,
and I think we can find a solution there as well.

PS2: We used WebRTC for the second "hangout" and it worked quite well
(and my internet connection is not that fast). I think it would be worth
exploring how we could host such a video meeting open source hub under
OpenStack Infrastructure directly.

Raw notes of the hangout meetings:
https://etherpad.openstack.org/p/project-restructure-hangouts

-- 
Thierry Carrez (ttx)