Code review comment for lp:~paulliu/unity-mir/logout

> The problem with your last suggestion Daniel would be that we'd (have to)
> block the UI thread during stopAllApplications, as we want to shut them down
> gracefully and only then shut Unity/Mir down. That's why I think the dual-
> signal approach will serve us better, even if we think the responsibilities
> should shift between unity8 and unity-mir.

Ok.

> I'm not entirely sure we want to move it up to the shell, though. If we think
> unity-mir is meant to facilitate building shells, the more we move up there,
> the more the shell needs to do. I can be convinced, though.

hm!? Come on, unity-mir has "unity" in its name. So it's an integral part of the unity shell. But it seems to be treated like a unity8-cpp (where we stuff all the CPP code of unity8) whereas unity8 proper is like a unity8-qml (i.e., we want unity8 to have only QML code).

Also what's being added here is the com.canonical.Unity.Session service, which I suppose will have other methods and signals in the future that unity8 qml will be interested on and not only ApplicationManager.