Re: [DISCUSS] [Contributing] (3) - Review Tooling
In Beam, we have a bot that regularly nags people about inactive PRs and also closes them after long inactivity.
And we use the github feature for assigning reviewers in github.
Sometimes it is hard for people to judge how "valuable" a PR is. Maybe some knowledgable people could mark PRs as valuable if they think it's a good addition but if they don't have the review bandwith. Other people can then search for valuable PRs that don't yet a reviewer and review/merge them.
> On 22. Sep 2018, at 04:25, vino yang <yanghua1127@xxxxxxxxx> wrote:
> Hi Jin Sun,
> Earlier this year, I also had these questions when I started contributing
> code to Flink. In fact, the timing of a PR being reviewed will be related
> to the priority of the problem solved by the PR.
> And when you indicate the module to which it belongs in the title of the
> PR, like "[FLINK-xxxx][module] XXXX", the person in charge of the relevant
> module or the contributor who is familiar with it will find it easier.
> To Stephan:
> Maybe we can open a separate mail thread (after all, the current discussion
> thread is about a specific topic) to hear the contributors about PR review
> related questions and doubts. Perhaps some of their feedback will help the
> community improve the way they review.
> Thanks, vino.
> Jin Sun <isunjin@xxxxxxxxx> 于2018年9月22日周六 上午6:40写道：
>> As a new contributor I cared about how to make my contribution accepted by
>> the community, some questions:
>> 1) When will it get reviewed? Is there a rule about review timeline?
>> 2) There are long backlog of pull requests, What happened if a pull
>> request not get noticed, do we have some mechanism to make it moving
>> forward, like a pull request will be assigned a owner of reviewer? Or we
>> have a review queue and a pull request will be get handled fairly.
>>> On Sep 20, 2018, at 12:56 PM, Stephan Ewen <sewen@xxxxxxxxxx> wrote:
>>> Hi all!
>>> This thread is dedicated to discuss the tooling we want to use for the
>>> It is spun out of the proposal *"A more structured approach to reviews
>>> *Suggestions brought up so far*
>>> *Use comments / template with checklist*
>>> - Easy to do
>>> - Manual, a bit of reviewer overhead, reviewers needs to know the
>>> *Use a bot *
>>> - Automatically add the review questions to each new PR
>>> - Further details?
>>> *Use GitHub labels*
>>> - Searchable
>>> - possibly not obvious to new contributors
>>> - Any restrictions? Do members need to apply at ASF infra to have
>>> permissions to edit github labels?