Support for monorepo / cross-team repositories

[necojackarc]

Support for monorepo / cross-team repositories

[necojackarc]

Oops, I completely forgot that @seanbecker15 reached out to me on Twitter...
Here's his suggestion - https://github.com/seanbecker15/auto-request-review/tree/per-group-number-of-reviewers

[necojackarc]

Unfortunately, it doesn't support that use case. The biggest blocker is the option for the maximum number of reviewers applied to the whole reviewer selection.

Let's say five reviewers are chosen based on files changed and teams, and you have options.number_of_reviewers = 3. The random three will be picked out of these five. This is because the randomise logic comes at the very end.

auto-request-review/src/index.js

Lines 71 to 75 in 2cfdd67

	core.info('Randomly picking reviewers if the number of reviewers is set');
	reviewers = randomly_pick_reviewers({ reviewers, config });
	core.info(`Requesting review to ${reviewers.join(', ')}`);
	await github.assign_reviewers(reviewers);

Off the top of my head, I can think of something like options.groups.[group_name].number_of_reviewers.

I'm not sure if this is a good idea, but I think your case could be covered with this option:

reviewers:
  groups:
    team_1_maintainers:
      - member_a
    team_1_contributors:
      - member_b
    team_2_maintainers:
      - member_c
    team_2_contributors:
      - member_d

  per_author:
    team_1_contributors:
      - team_1_maintainers
      - team_1_contributors

    team_2_contributors:
      - team_2_maintainers
      - team_2_contributors

options:
  groups:
    team_1_maintainers:
      number_of_reviewers: 1
    team_2_maintainers:
      number_of_reviewers: 1

[seanbecker15]

I think adding that option would work well for this use-case. Teams / groups would be coupled with a number of reviewers and could set the number of reviewers they want added to each PR.

One question I would have is how this option would interact with the global number of reviewers option. It seems like this global option might serve best as a default value if there aren't enough reviewers in the groups.

[necojackarc]

Good question. I think it's important to make it backwards compatible. The global option should limit the total number of reviewers. If there are other ways to use the global options without breaking the current behaviour, that'd be great, but they can be similar but separate options.