Hi Segher,
On Sun, Jun 15, 2025 at 11:07:15AM -0500, Segher Boessenkool wrote:
> On Sun, Jun 15, 2025 at 05:52:50PM +0200, Mark Wielaard wrote:
> > On Wed, Jun 11, 2025 at 10:40:02AM -0500, Segher Boessenkool wrote:
> > > On Wed, Jun 04, 2025 at 05:53:18PM -0600, Jeff Law via Gcc wrote:
> > > > If we make things simpler for those cases where it is actually simple
> > > > and increase visibility so that folks know state I think it would be a
> > > > welcome improvement to the process.
> > >
> > > Yup. And it might not be a bad plan to get some people who still are
> > > active onto the GCC steering committee, as well!
> > >
> > > It is as important now as ever that there *is* such a thing as the SC:
> > > all decisions need to be taken responsibiliity for by some final
> > > authority, which we (as a project) have all decided to have that
> > > authority.
> >
> > Agreed, having active developers taking responsibility is
> > important. But it doesn't really have to be through a steering
> > committee imho.
>
> I specifically said that there *is* a need for something like the SC.
> There needs to be a final authority. Without that there will be
> eternal (and eternally lasting) fights.
I agree with you we need some developers that are actually still
active in the project to help make decisions and take responsibility.
I just don't think that all needs to be done through an SC. I don't
know why that would have to lead to fights. We have become pretty good
at getting to consensus on things.
> > We (as a project) can decide that the authority lies
> > with the active maintainers as a whole getting to concensus. Just like
> > any maintainer can appoint new write after approval accounts. Or, as
>
> WaA is decided by the sourceware maintainers. The request form says
> "email address of person who approved request", but that is not who has
> the final call :-) Which of course makes sense, the sourceware
> maintainers primarily need to keep their system safe and working!
That is not how it works. Technically of course sourceware overseers
could just randomly add or remove accounts. But I don't remember any
instance of that ever happening. They just follow the process outlined
at https://gcc.gnu.org/gitwrite.html#authenticated checking the
details and verifying the approver is allowed to approve. But they
aren't just going to second guess the approver when everything is in
order.
> > in this case, the active maintainers of a subsystem decide on who else
> > becomes a reviewer or maintainer for that subsystem. The release
> > managers could decide if and when to accept a new front or
> > backend. etc.
>
> That is not how things work. The SC decides who does and does not
> become maintainer (reviewer is just a hobbled kind of maintainer, there
> is no real difference). Maintainers for frontends, backends, subsystems
> can recommend things, sure. But they have no separate authority, there
> can not be fiefdoms. This is Good(tm).
This case is Richard's proposal and I think it will lead to having
more active maintainers and reviewers precisely because currently the
SC is a bit stale and mostly not very active. IMHO the active
maintainers know best here and we don't need the SC for these kind of
decisions.
Cheers,
Mark
