Branko Čibej wrote on Mon, 12 Feb 2018 23:11 +0100:
> Backporting experimental features isn't crucial, surely? Yet if there
> are changes in trunk, they _might_ affect other code; in that case, more
> eyes is better.
So perhaps something like 'two +1s and one "I verified no non-
experimental code is affected"'. There were some shelving-related
nominations in the past weeks that I would have easily been able to cast
the latter vote for, but not the former.
The thinking here is to have a three-man rule for the "Stable features
aren't affected" part but a two-man rule for the "Changes to an
experimental feature" part.