Open Stack

On 09/23/2014 02:18 AM, Thierry Carrez wrote:
> The main goal of incubation, as we did it in the past cycles, is a
> learning period where the new project aligns enough with the existing
> ones so that it integrates with them (Horizon shows Sahara dashboard)
> and won't break them around release time (stability, co-gate, respect of
> release deadlines).
> 
> If we have a strict set of projects in layer #1, I don't see the point
> of keeping incubation. We wouldn't add new projects to layer #1 (only
> project splits which do not really require incubations), and additions
> to the big tent are considered on social alignment only ("are you
> vaguely about cloud and do you follow the OpenStack way"). If there is
> nothing to graduate to, there is no need for incubation.

There's no need for incubation, as such, but it's worth taking the time
to think about the technical and social functions that incubation and
integration served (sometimes ineffectively, or only as side-effects),
and what will replace them. You've identified a few there:

- learning period for new projects
- alignment with existing projects
- stability (in which gating served as a weak crutch, and the real
answer will likely lie in more extensive cross-project communication,
also carefully filtered to avoid information overload)
- respect of release deadlines (which doesn't necessarily mean releasing
all at the same time, just being cognizant of network-effects of
releases, and the cadence of other projects in an up-or-down dependency
relationship with yours)

> In Monty's proposal, ATC status would be linked to contributions to the
> big tent. Projects apply to become part of it, subject themselves to the
> oversight of the Technical Committee, and get the right to elect TC
> members in return.

And, a few more here:

- transitioning from "island" to part of the big tent
- accepting oversight of TC
- accepting responsibility to participate in TC election

Allison