Last week, I went through and added Github owner data to every module
in Extras, and then I went through and tagged "owner_pr" to every PR
in which the requestor was also the module owner. There were a very
few (5, I think) and they've all since been merged.
Today I made another pass through all of the Extras PRs, and looked
for "submitters of new Extras modules who are already owners of at
least one other module", and tagged them with "owner_pr" as well (not
perfect, but I didn't want to make another tag.)
There are currently 91 new PR requests for new modules. 12 of them are
from extant module owners -- which, of course, means that 79 of them
are not. You can find them here:
I'd like to review these 12 quickly, in hopes getting some insight
about a key question: if someone is a "module owner", does that also
imply that they're a competent reviewer?
I'm going to assume, for now, that we will end up saying "yes" to that
question. I hope we can, because it's now clear that what we need most
to move Extras forward is to have more reviewers with actual
authority.
With all that in mind, here is my latest proposal for Extras merge
policy -- much simplified from my previous proposal.
1. For all existing modules, a +1 from the owner means "merge". I'd
previously recommended that this come from the owner's branch, but I
now believe that's an unnecessary step. Obviously, a push to an
existing module from its owner is an implicit "+1". Thus: if the
owner of a module says "merge it", we merge it.
2. All new modules need two +1s and no -1s -- and these +1s/-1s can
come from any current module owner. If the submitter is also a module
owner, that's an implicit +1, so any other +1 with no -1s would mean
"merge".
I think this policy is a good mix of delegating authority, guarding
quality, and keeping things simple.
Thoughts?
--g