> It
> seems like the pattern in the positive examples is a deep and positive
> contribution to some aspect of the project.

Yes. For PPMC membership, a continued contribution to governance
issues like release voting or design questions.

> I think in some of those examples, where the committer hadn't touched much
> code or done any code reviews, we'd expect the committers to exercise their
> discretion and abstain from +2ing code reviews (instead deferring to other
> committers), right?

Yes. This is what other Apache projects do, according to Todd, one of
our incubation mentors.

> In
> that case would the argument for committership be that we want their input
> on other things to have some official weight?

This story is a little fuzzy, but I think committership is not only +2
powers on CR, but also:

1. A first step towards PPMC membership
2. A recognition of work already done
3. A method to encourage continued and further contributions

IIUC, these are some of the ways other projects use the commit bit.

> An alternative view I've heard is that committers (at least ones whose
> committership is based on their code contributions) are people that other
> committers trust to maintain the codebase and that we should determine this
> based on their history of code contributions and code reviews.

Perhaps we could lower that to a committer we trust to continue doing
the work they are doing without active tutoring and mentorship from
other committers. This implies that committers still review each
other's code, that different committers have different specialties,
and that maintenance of any sizeable codebase over the long run is a
herculean task.

> I think regardless we'd expect
> committers to use their discretion in deciding whether to +2 something - in
> many cases there's one or a handful of people who know parts of the
> codebase best and should have input on non-trivial changes.

Agreed.

Reply via email to