Code review comment for lp:~robertcarr/mir/client-focus-notifications

> There are a few open questions:
>
> Should focus be modeled as a Surface attribute?
> Is there any reason for the Surface to keep track of if it has focus.

makes sense to me it would be a surface attribute. as to why it might want to keep track....optimizations if there is no focus? and possibly, other client requests that won't work/don't make sense if focused or not.

just curious...is this some convention ?
mir_surface_focus_arraysize_
didn't understand why it was in the surf attrib for focusstate

in general...might be my lack of understanding. there are some mutex/surface locks in the proposed code....but i don't see unlocks (do those happen automagically as the code goes out of scope?)