Code review comment for lp:~kdub/mir/display-groups

173 + * This may wait for based on hardware conditions and platform requirements */

Something missing.

178 +private:

(Nit) No need for private, protected will do.

629 + for (auto& compositor : compositors)
630 + {
631 + target.ensure_current(std::get<0>(compositor));
632 + std::get<1>(compositor)->composite(scene->scene_elements_for(comp_id));

(Non-blocking) As discussed on IRC, this approach serializes rendering of multiple outputs on Android (i.e., it's R1,R2,...,Rn,Post vs the ideal Parallel(R1,R2,...,Rn),Post). This could lead to some performance loss for the multimonitor case on Android, so hopefully there is a way to improve this if needed.

1711 +// display->for_each_mock_buffer([](mtd::MockDisplayBuffer& mock_buf)
1712 +// {
1713 +// EXPECT_CALL(mock_buf, make_current()).Times(1);
1714 +// EXPECT_CALL(mock_buf, view_area())
1715 +// .WillOnce(Return(geom::Rectangle()));
1716 +// });

Not needed?

1769 +#if 0
1770 TEST(MultiThreadedCompositor, makes_and_releases_display_buffer_current_target)

?