Assign a random team member with r+ priviliges on r? @rust-lang/<team> #187

Activity

1. Centril commented on Jan 5, 2019

   

   Centril

   on Jan 5, 2019

   ContributorAuthor

   More actions

   I'm not familiar with the code-base but... To do this, it should be sufficient to:

   1. update the regex here: https://github.com/rust-lang-nursery/highfive/blob/414fd076a3c67cfa8b0790733c0a734cd5231cb1/highfive/newpr.py#L42
   2. add a rewrite on the form @<org>/<team> in this if statement: https://github.com/rust-lang-nursery/highfive/blob/414fd076a3c67cfa8b0790733c0a734cd5231cb1/highfive/newpr.py#L416
      1. when this form is encountered, fetch the available users in that team from GH.
      2. fetch the available users with r+ rights.
      3. intersect users in i. and ii. -- if the intersection is empty, bail and don't assign anyone (possibly comment on GH with a warning).
      4. pick a user in the intersection at random.
      5. set reviewer to whatever user selected in iv.
   3. update relevant tests somewhere in this file: https://github.com/rust-lang-nursery/highfive/blob/414fd076a3c67cfa8b0790733c0a734cd5231cb1/highfive/tests/test_newpr.py#L1

   cc @pietroalbini to correct mistakes I might have made or clarifications if need be.
2. emilyalbini commented on Jan 5, 2019

   

   emilyalbini

   on Jan 5, 2019

   Member

   More actions

   I'm not sure if it's that useful to intersect with the list of users with r+ rights though. The list seems correct to me, cc @davidalber
3. davidalber commented on Jan 11, 2019

   

   davidalber

   on Jan 11, 2019

   Collaborator

   More actions

   That should cover the case for new comments. You probably want to have this behavior for messages in new PRs, as well. You'll want to encapsulate all of the logic for converting the team alias to a user inside a function, possibly inside find_reviewer.
4. SimonSapin commented on Dec 27, 2019

   

   SimonSapin

   on Dec 27, 2019

   Contributor

   More actions

   Would it be enough to stick to the default directory-based assignment algorithm? #241 could be an alternative if so.

   when someone writes the above is to leave the PR without a reviewer

   That sounds like a bug in itself, highfive should never unassign without assigning someone else.
5. Mark-Simulacrum

   closed this as completedin #249on Jul 12, 2021