Adding a group of people as reviewers doesn't block others from help and review (and it doesn't change what we do now). I don't see how it's different to today, just having default people reviewing, adding new people. Actually, we clearly have today a bunch of PRs stale just due to lack of reviewers. From a community standard, I'm also concerned that a lot of PR is waiting for review from the same people: that is a concern for community engagement. If we have 3 persons that should review/approve 90% of the PRs, it doesn't scale, it doesn't engage the community, other committers/PMC members might be feeling "untrusted".
So the idea is actually to grow the community: the group of reviewers can invite other people to review (having default reviewers on some modules doesn't block adding others). We have several examples of Apache projects where it works fine (Apache Beam is an example, we increased the community engagement thanks to feedback from reviewer pretty quickly instead of stale for a while and contributors give up due to no response). Anyway, I propose to update my proposal this way: 1. I update the stale PR periodical reminder (every week) 2. I don't add reviewers yml, but if a PR doesn't have reviewer after a week, I send a report on the dev mailing list listing all stale and no review started PRs) Regards JB On Tue, Mar 26, 2024 at 5:03 PM Ryan Blue <b...@tabular.io> wrote: > > Sorry, I'm a strong -1 for having owners or standard reviewers. > > In this community, we've always taken the stance that anyone should be able > to jump in and help. Having assigned owners may seem like a good idea, but it > actually prevents other people from volunteering and getting involved. This > is also why we don't assign issues to individuals -- they often don't end up > submitting a PR and it prevents other people from contributing. Having an > assigned owner gives the impression that the responsibility is on a > particular individual, making other people that are capable of reviewing not > pay attention. I think this will slow down the community and I don't think it > is a good idea. > > Ryan > > On Tue, Mar 26, 2024 at 6:36 AM Ajantha Bhat <ajanthab...@gmail.com> wrote: >> >> +1 for having multiple PR review owners per module/label. >> >> Having module owners can accelerate PR processing. For instance, I'm >> awaiting feedback on a Spark action for computing partition stats >> (https://github.com/apache/iceberg/pull/9437). Currently, only Anton is >> reviewing, which may cause delays if he's occupied. In my opinion, having >> multiple module owners would enable developers to seek feedback more >> efficiently. >> >> - Ajantha >> >> On Thu, Mar 21, 2024 at 11:11 PM Jean-Baptiste Onofré <j...@nanthrax.net> >> wrote: >>> >>> Hi folks >>> >>> Now that we have the proposal process "merged", I will create the PR >>> about reviewers and update stale job. >>> >>> I should have the PR tomorrow for review. >>> >>> Thanks ! >>> Regards >>> JB >>> >>> On Thu, Mar 21, 2024 at 9:55 AM Jean-Baptiste Onofré <j...@nanthrax.net> >>> wrote: >>> > >>> > Hi Dan >>> > >>> > Yes, I saw you merged it, that's great. >>> > >>> > I will move forward on the "stale bot" stuff. >>> > >>> > Thanks ! >>> > Regards >>> > JB >>> > >>> > On Wed, Mar 20, 2024 at 8:48 PM Daniel Weeks <dwe...@apache.org> wrote: >>> > > >>> > > Hey JB, apologies for combining these two things in the same thread, >>> > > but we got enough eyes on the first PR and I went ahead and merged i >>> > > >>> > > If you want to put together the PR for your proposed changes, we can >>> > > get looking at that. >>> > > >>> > > We'll also need to backfill the existing proposals and update the >>> > > website to have a link to the label. (Will work with you and Bits on >>> > > that) >>> > > >>> > > Thanks, >>> > > -Dan >>> > > >>> > > >>> > > >>> > > On Wed, Mar 20, 2024 at 10:01 AM Jean-Baptiste Onofré >>> > > <j...@nanthrax.net> wrote: >>> > >> >>> > >> Hi Fokko >>> > >> >>> > >> I think combining Dan's proposal about "proposal process" and this >>> > >> proposal about "PR flows" would be helpful for the project (to track >>> > >> the proposals and avoid "stale" PRs/proposals). >>> > >> >>> > >> If PMC members are OK, I'm ready to help to set this up :) >>> > >> >>> > >> Thanks >>> > >> Regards >>> > >> JB >>> > >> >>> > >> On Wed, Mar 20, 2024 at 12:27 PM Fokko Driesprong <fo...@apache.org> >>> > >> wrote: >>> > >> > >>> > >> > Hey everyone, >>> > >> > >>> > >> > This is a gentle bump from my end on this thread since I like the >>> > >> > idea. Several people have already approved Dan's PR about >>> > >> > formalizing the proposal process. Are there any questions or >>> > >> > concerns from the PMC before adopting this? >>> > >> > >>> > >> > Kind regards, >>> > >> > Fokko Driesprong >>> > >> > >>> > >> > Op wo 13 mrt 2024 om 13:17 schreef Renjie Liu >>> > >> > <liurenjie2...@gmail.com>: >>> > >> >> >>> > >> >> Hi, JB: >>> > >> >> >>> > >> >> Your proposal looks great to me. We should definitely have a vote >>> > >> >> for a proposal impacting the spec, and the model is great. >>> > >> >> >>> > >> >> On Tue, Mar 12, 2024 at 10:55 PM Jean-Baptiste Onofré >>> > >> >> <j...@nanthrax.net> wrote: >>> > >> >>> >>> > >> >>> Hi >>> > >> >>> >>> > >> >>> I think a vote would be necessary only if we don't have consensus >>> > >> >>> on a >>> > >> >>> proposal. If anyone is OK with the proposal (no clear "concern" in >>> > >> >>> the >>> > >> >>> doc and/or the GitHub issue), a vote is not required. >>> > >> >>> That said, any proposal impacting a spec should be voted (as part >>> > >> >>> of >>> > >> >>> the spec proposal). >>> > >> >>> >>> > >> >>> I think it's fair to identify a proposal vote as a "code >>> > >> >>> modification" vote. >>> > >> >>> It means that it follows this model: a negative vote constitutes a >>> > >> >>> veto , which the voting group (generally the PMC of a project) >>> > >> >>> cannot >>> > >> >>> override. Again, this model may be modified by a lazy consensus >>> > >> >>> declaration when the request for a vote is raised, but the >>> > >> >>> full-stop >>> > >> >>> nature of a negative vote does not change. Under normal (non-lazy >>> > >> >>> consensus) conditions, the proposal requires three positive votes >>> > >> >>> and >>> > >> >>> no negative votes in order to pass; if it fails to garner the >>> > >> >>> requisite amount of support, it doesn't. Then the proposer either >>> > >> >>> withdraws the proposal or modifies the code and resubmits it, or >>> > >> >>> the >>> > >> >>> proposal simply languishes as an open issue until someone gets >>> > >> >>> around >>> > >> >>> to removing it. >>> > >> >>> >>> > >> >>> We can link to https://www.apache.org/foundation/voting.html. >>> > >> >>> >>> > >> >>> Regards >>> > >> >>> JB >>> > >> >>> >>> > >> >>> On Tue, Mar 12, 2024 at 2:21 AM Renjie Liu >>> > >> >>> <liurenjie2...@gmail.com> wrote: >>> > >> >>> > >>> > >> >>> > Hi, Daniel: >>> > >> >>> > >>> > >> >>> > Thanks for this summary. >>> > >> >>> > >>> > >> >>> > I think one thing missing is that do we need a vote for the >>> > >> >>> > proposal to be accepted or rejected? If required, what should >>> > >> >>> > the voting process be? >>> > >> >>> > >>> > >> >>> > On Tue, Mar 12, 2024 at 9:04 AM Daniel Weeks <dwe...@apache.org> >>> > >> >>> > wrote: >>> > >> >>> >> >>> > >> >>> >> Hey everyone, I synced up with JB about the proposal process >>> > >> >>> >> and wanted to see if we could make some initial progress. >>> > >> >>> >> >>> > >> >>> >> Based on some of the earlier discussions, we want to leverage >>> > >> >>> >> as much of the informal process as possible, but improve >>> > >> >>> >> discoverability and a little structure. This probably means >>> > >> >>> >> using github for tracking, google docs where possible for the >>> > >> >>> >> early proposal implementation comments, and the dev list for >>> > >> >>> >> discussion threads, awareness and voting. >>> > >> >>> >> >>> > >> >>> >> That said, I propose we adopt the following: >>> > >> >>> >> >>> > >> >>> >> 1. A simple issue template for initiating a proposal and >>> > >> >>> >> applying a 'proposal' label to the issue >>> > >> >>> >> 2. Use a github search link to document current proposals >>> > >> >>> >> (based on the 'proposal' label) >>> > >> >>> >> 3. Continue using google docs for proposals >>> > >> >>> >> documentation/comments (referenced from the github issue) >>> > >> >>> >> 4. Continue to create DISCUSS threads on the dev list for >>> > >> >>> >> communication >>> > >> >>> >> 4. Backfill current proposals by creating issues for them >>> > >> >>> >> >>> > >> >>> >> I've created this PR to capture the initial template and docs. >>> > >> >>> >> >>> > >> >>> >> I think we want to introduce this with as little overhead as >>> > >> >>> >> possible. Please follow up with questions/comments so we can >>> > >> >>> >> close this out. >>> > >> >>> >> >>> > >> >>> >> Thanks, >>> > >> >>> >> Dan >>> > >> >>> >> >>> > >> >>> >> >>> > >> >>> >> On Sun, Mar 10, 2024 at 11:30 PM Jean-Baptiste Onofré >>> > >> >>> >> <j...@nanthrax.net> wrote: >>> > >> >>> >>> >>> > >> >>> >>> Hi Manu >>> > >> >>> >>> >>> > >> >>> >>> Yup, it's on my TODO. Thanks for the reminder, I will be back >>> > >> >>> >>> on this >>> > >> >>> >>> one this week :) >>> > >> >>> >>> >>> > >> >>> >>> Regards >>> > >> >>> >>> JB >>> > >> >>> >>> >>> > >> >>> >>> On Mon, Mar 11, 2024 at 4:07 AM Manu Zhang >>> > >> >>> >>> <owenzhang1...@gmail.com> wrote: >>> > >> >>> >>> > >>> > >> >>> >>> > Hi JB, >>> > >> >>> >>> > >>> > >> >>> >>> > Are you still working on this nice proposal? >>> > >> >>> >>> > >>> > >> >>> >>> > Regards, >>> > >> >>> >>> > Manu >>> > >> >>> >>> > >>> > >> >>> >>> > On Thu, Jan 4, 2024 at 3:35 PM Fokko Driesprong >>> > >> >>> >>> > <fo...@apache.org> wrote: >>> > >> >>> >>> >> >>> > >> >>> >>> >> Nice! I fully agree with the abovementioned. I originally >>> > >> >>> >>> >> set up the stalebot for the issues because I noticed that >>> > >> >>> >>> >> there were many issues around old Spark versions that >>> > >> >>> >>> >> weren't even maintained anymore. I feel it is better to >>> > >> >>> >>> >> either close or take action on an issue. For me, it makes >>> > >> >>> >>> >> sense to extend this to PRs as well. >>> > >> >>> >>> >> >>> > >> >>> >>> >> Same as Amogh said, always feel free to ping me when either >>> > >> >>> >>> >> a PR or issue lingering and you need some eyes on it. >>> > >> >>> >>> >> >>> > >> >>> >>> >> Kind regards, >>> > >> >>> >>> >> Fokko >>> > >> >>> >>> >> >>> > >> >>> >>> >> Op do 4 jan 2024 om 07:42 schreef Jean-Baptiste Onofré >>> > >> >>> >>> >> <j...@nanthrax.net>: >>> > >> >>> >>> >>> >>> > >> >>> >>> >>> Hi >>> > >> >>> >>> >>> >>> > >> >>> >>> >>> That's also the purpose of the reviewers file: having >>> > >> >>> >>> >>> multiple >>> > >> >>> >>> >>> reviewers per tag. >>> > >> >>> >>> >>> >>> > >> >>> >>> >>> Thanks guys for your feedback, I will move forward with >>> > >> >>> >>> >>> the PR :) >>> > >> >>> >>> >>> >>> > >> >>> >>> >>> Regards >>> > >> >>> >>> >>> JB >>> > >> >>> >>> >>> >>> > >> >>> >>> >>> On Thu, Jan 4, 2024 at 6:38 AM Ajantha Bhat >>> > >> >>> >>> >>> <ajanthab...@gmail.com> wrote: >>> > >> >>> >>> >>> > >>> > >> >>> >>> >>> > +1, >>> > >> >>> >>> >>> > >>> > >> >>> >>> >>> > Some of my PRs have been open for a long time and >>> > >> >>> >>> >>> > sometimes it doesn't get the attention it requires. >>> > >> >>> >>> >>> > Notifying both the reviewer and the author can help >>> > >> >>> >>> >>> > expedite the review process and facilitate quicker >>> > >> >>> >>> >>> > handling of new contributions. >>> > >> >>> >>> >>> > I think having more than one committer assigned for PR >>> > >> >>> >>> >>> > can also definitely help in speeding up the process if >>> > >> >>> >>> >>> > one of the committer is busy or on holiday. >>> > >> >>> >>> >>> > >>> > >> >>> >>> >>> > But we also need to think on the next steps. What if we >>> > >> >>> >>> >>> > still don't receive the necessary response even after >>> > >> >>> >>> >>> > sending notifications? >>> > >> >>> >>> >>> > Should we have a slack channel for those PRs to conclude >>> > >> >>> >>> >>> > by discussing (or some guidelines on how to take it >>> > >> >>> >>> >>> > further). >>> > >> >>> >>> >>> > >>> > >> >>> >>> >>> > We can have a trial run for some days and see how it >>> > >> >>> >>> >>> > goes. >>> > >> >>> >>> >>> > >>> > >> >>> >>> >>> > Thanks, >>> > >> >>> >>> >>> > Ajantha >>> > >> >>> >>> >>> > >>> > >> >>> >>> >>> > On Thu, Jan 4, 2024 at 8:19 AM Amogh Jahagirdar >>> > >> >>> >>> >>> > <am...@tabular.io> wrote: >>> > >> >>> >>> >>> >> >>> > >> >>> >>> >>> >> +1, I think this is a step in the right direction. One >>> > >> >>> >>> >>> >> other consideration I wanted to bring up was dependabot >>> > >> >>> >>> >>> >> and if there's any unique handling we want to do there >>> > >> >>> >>> >>> >> because I've noticed that PRs from dependabot tend to >>> > >> >>> >>> >>> >> pile up. I think with the proposal we won't really need >>> > >> >>> >>> >>> >> to do anything unique and just treat it as a normal PR >>> > >> >>> >>> >>> >> (it would be a build label with its own set of >>> > >> >>> >>> >>> >> reviewers) and we'll get notified the same way. >>> > >> >>> >>> >>> >> >>> > >> >>> >>> >>> >> I'll also say for reviews (speaking for myself, but I >>> > >> >>> >>> >>> >> think many others probably feel this way as well), >>> > >> >>> >>> >>> >> always feel free to ping on Slack and follow up :) But >>> > >> >>> >>> >>> >> overall I do like having more of a mechanism. > > > > -- > Ryan Blue > Tabular
