Thanks Julian for starting the discussion! I'm spending my spare time to contribute to Calcite, usually at weekends, and sometimes in the break of weekdays, hence my time would be limited because the spare time may varies. Review work is not that simple for me because Calcite has many complicated components and evolves many years which means we need track a lot of background. I'm still learning some part while doing the review work.
The complexity of PRs varies a lot, simple ones would be easier to get in because it only cost me minutes to hours to review. But the complex ones, usually I need to spend more time to understand the background, new design, the effect to the whole project, and the future direction we want to take. These kinds of PRs may be preempted by small ones, and finally do not getting reviewed for months, there is a example[1] which I would say sorry to the author that I still do not manage to give it a review till today. Any way to improve current status would be grateful. However, if the proposal from Julian may require more sustainable time, I'm not sure if it is suitable for guys like me who only devotes limited spare time to Calcite. Hence I'm +0 for this proposal. Of course, I would be happy to participate in the schema if we finally decide to do it. [1] https://issues.apache.org/jira/browse/CALCITE-5413 Charles Givre <cgi...@gmail.com> 于2023年4月11日周二 12:43写道： > I for one would very much like to help with reviews. I don't have a lot > of time this month, but next month should have more time. > Best, > -- C > > > On Apr 10, 2023, at 10:56 PM, Dan Zou <zoud...@163.com> wrote: > > > > +1, thanks Julian for proposing this. From my observation, there are > many pending PRs in Calcite and only a few active committers, this puts a > lot of pressure on these committers. For example Julian have reviewed 34 PR > in 2023 Q1, it is an unimaginable number. I am very supportive of achieving > a mechanism to improve the review efficiency of PRs, and also I would like > to make contribution in reviewing PRs. > > > > Best, > > Dan Zou > > > > > > > > > > > >> 2023年4月11日 01:56，Julian Hyde <jh...@apache.org> 写道： > >> > >> I don't enjoy reviewing and merging PRs. And every time I do, I feel > >> like a sucker, because there are over a few dozen committers who are > >> enjoying the project and not doing the work. (There is a small group > >> of committers who regularly review and merge PRs. I don't know how > >> they feel about the task, but I am immensely grateful.) > >> > >> I think I would review more PRs if I saw others doing the same. > >> > >> Can we figure out a fairer way to distribute the load? For release > >> managers (approximately the same amount of work, but compressed into a > >> few hours or days) we have successfully run a rota for several years. > >> Could we do something similar with PRs? > >> > >> I propose the following. For each calendar month, there is a PR > >> manager and 6 - 8 reviewers. The PR manager does not review PRs, but > >> assigns them to reviewers, and politely reminds reviews to keep the PR > >> moving. > >> > >> The PR manager's goals are: > >> * every non-draft PR is reviewed within 3 days of submission, > >> * every PR is merged within 3 days of being done; > >> * rotate duties so that no reviewer is asked to review more than 4 > >> PRs per month; > >> * email a report at the end of the month; > >> * work down the backlog of historic PRs if it's a slow month. > >> > >> The PR manager rotates every month. The reviewers can rotate if they > >> wish, but I suspect most will stay in the pool for several months, > >> because the reviewing load is not very heavy, and because they see > >> others doing the work. > >> > >> Other notes: > >> * Non-committers would be welcome to join the pool of reviews (and > >> that would be a good way to earn the committer bit) and a committer > >> could merge when the PR is approved. > >> * If committers join the pool, that's a good way to earn PMC membership. > >> * Committers who are not in the pool are welcome to review PRs and > >> assign PRs to themselves (but expect to be nagged by the PR manager if > >> you don't review in a timely manner). > >> > >> What do you think? Would you join this scheme if we introduced it? If > >> you agree please +1; also happy to see revisions to this suggestion or > >> other ideas to share the work. > >> > >> Julian > > > > -- Best, Benchao Li
