Open Stack

> > So while this approach would introduce new complexity to the
> > release management side, do you think it's a show-stopper level of
> > complexity, or a more a
> > give-us-pause-to-ensure-we-really-need-to-do-this level?
> 
> More of a give-us-pause-to-ensure-we-really-need-to-do-this level.

Cool, the answer I was hoping for :)

> But if we follow the oslo route (see other part of the thread) I
> think we are not adding any complexity on the organizational/release
> side of things (or at least, no complexity or process that we don't
> already have).

Interesting idea, but this doesn't really feel like oslo to me.

The reasons being that:

(a) its unlikely to ever be a truly cross-cutting concern (i.e. I
    can't see a third project needing this, other than nova & ceilo)

(b) to be fully usable across a range of hypervisors, it would drag
    in a chunk of multiple drivers, which smells very compute-specific
    and out of place compared the existing stuff in oslo.

However, I'd be open to correction if I've envisaged too limited a
role for oslo in this respect.

(And I can definitely see the value in not adding additional complexity
to the release train if we don't need to).

Cheers,
Eoghan