On Mon, Feb 14, 2022 at 2:01 PM Brennan Ashton <bash...@brennanashton.com> wrote: > > Background: > I am generally opposed to both of these. It is quite rare that we need a > crazy fast merge turn around on a PR. And if something is approved and > straight up broken in master that needs to get in then I think forgiveness > can be used to self merge. > > > I also generally do not have a big issue about people from the same company > reviewing and merging. I could see the arguments for shared code but then I > think we are nitpicking. I prefer the velocity with a few oops that can > be reverted along the way if needed. There is also parts of the code base > where the best people to review are on the same company. > > > I think most of the concerns here are best addressed not by process but > increasing the number of contributors who can participate. (more committers > and PPMC)
Feel free to correct me if I'm mistaken, but I think David is bringing this up because of time zones. Indeed, most of the PR merging activity seems to occur during what I would call nighttime or early morning, and I think that might be more pronounced in David's time zone. Still, I think things have been working well, more or less, and I don't think we need to make up any new rules right now. Instead, I would only urge committers to give complex PRs 12-24 hours to percolate, even if there's an approving review, so other time zones have a chance to look at them. Obviously that doesn't apply if it's urgent. For example, if the build is broken and people can't get work done, or a serious error was merged and needs to be reverted ASAP, don't wait, do it! Also, it's not necessary to delay for trivial PRs. What are the definitions of "complex," "trivial," "urgent," etc? I say, committers should just use their best judgment and try to find a good balance. Don't rush too much, don't delay too much. :-) David brings up a good point about time zones and we do have to remember that NuttX is a global project, and I think that's the main point to keep in mind. To Brennan's last point: as we grow the committer base, we are likely to have more people in more time zones and more PR reviewers, so this should become less of a concern. Nathan
