This removes the implicit choice of the "all" group in the explicit reviewer pick. This prevented repositories that have an "all" group from being able to r? someuser, since it would always just pick from "all".
This was regressed in #377, but instead of trying to bring back the previous behavior (where "all" was only included when a group was matched), this just removes "all" altogether. There was a comment at https://github.com/rust-lang/highfive/pull/249#discussion_r666504539 about this, but the explanation didn't make sense to me. If you say r? somegroup, I think it should only pick from that group and not include "all".
This removes the implicit choice of the "all" group in the explicit reviewer pick. This prevented repositories that have an "all" group from being able to
r? someuser
, since it would always just pick from "all".This was regressed in #377, but instead of trying to bring back the previous behavior (where "all" was only included when a group was matched), this just removes "all" altogether. There was a comment at https://github.com/rust-lang/highfive/pull/249#discussion_r666504539 about this, but the explanation didn't make sense to me. If you say
r? somegroup
, I think it should only pick from that group and not include "all".Fixes #380