Require 2/3 majority for FCP #315
Conversation
|
Note that for large teams there's still the "at most two unchecked" rule that means a 10-person team still needs 8 people, more than the ⅔ here. (I do think we could consider loosening that somewhat with the new supermajority check instead of the previous bare majority check.) |
|
FWIW the PR in the form proposed by me (>= 2/3) would affect only teams with 5 members, which are the following: lang, libs-api, opsem, release. We could have FCP with those 4 teams to change this, maybe? The form proposed by Jakob (> 2/3) would also affect teams with size 3 or 6: libs, docs-rs, crates-io, community-localization. (This is only looking at teams that have rfcbot enabled according to their data in the teams repo.) |
|
Updated to require >= 2/3. I think asking the affected teams if this makes sense to them seems reasonable. I don't know where the best place to hold such an FCP is though... |
|
I've kicked of the FCP process in rust-lang/rust#114986. |
| @@ -395,7 +395,7 @@ fn evaluate_pendings() -> DashResult<()> { | |||
| proposal.id, why)); | |||
| } | |||
|
|
|||
| let majority_complete = num_outstanding_reviews < num_complete_reviews; | |||
| let majority_complete = num_outstanding_reviews <= 2 * num_complete_reviews; | |||
There was a problem hiding this comment.
This seems wrong: with 2 complete reviews and 4 outstanding reviews, this evaluates to true. ^^
There was a problem hiding this comment.
Ah yes, the multiplication should be on the left-hand side.^^
Closes #293 .
Both the opsem and lang teams felt that a simple majority, which in their case meant 3/5 team members, did not sufficiently reflect team consensus. This PR updates rfcbot to require that more than 2/3 of all reviewers approve before an FCP begins.
There are a couple of alternatives to this approach:
I'll nominate this PR for discussion by the council. They seem like the right group to decide how to come to a decision here.