Open Stack

On Fri, Sep 28, 2018 at 5:00 AM Jeremy Stanley <fungi at yuggoth.org> wrote:

>
> If memory serves, the biggest challenge around that solution was
> determining who approves such proposals since they still need
> per-project specs for the project-specific details anyway. Perhaps
> someone who has recently worked on a feature which required
> coordination between several teams (but not a majority of teams like
> our cycle goals process addresses) can comment on what worked for
> them and what improvements they would make on the process they
> followed.
> --

This is definitely the biggest challenge, and I think it is one of
those things that is going to be on case by case basis.

In the case of neutron smartnic support with ironic, the spec is
largely living in ironic-specs, but we are collaborating with neutron
folks. They may have other specs that tie in, but that we don't
necessarily need to be aware of. I also think the prior ironic/neutron
integration work executed that way.  My perception with nova has also
largely been similar with ironic's specs driving some changes in the
nova ironic virt driver because we were evolving ironic, as long as
there is a blueprint or something tracking that piece of work so they
have visibility.  At some point, some spec has to get a green light or
be pushed forward first. Beyond that, it is largely a tracking issue
as long as there is consensus.