Code review comment for lp:~vanvugt/mir/frontend-server

Andreas wrote:
>> Alans proposal is more convincing than just renaming.
>> But I am not yet sure about scene vs server vs something better:
>> object_of_bespoke_interface->add(session_ptr)
>> // is add now supposed to decide whether the sessions surface should be visible, maybe focused or stay >> invisible?
>> Then both Scene and Shell seem to be legitimate names

I agree. Going on, it's possible to think of the Shell as interpreting requests from the frontend to the Scene...which isnt a total lie with the current object. Sure, the implementation object is not convincingly a shell...but that's not so important, it's the interface which exposes the shell services to the frontend. I'm not so upset with Shell.

I think scene is a little overly general in this case.

Given some refactoring in SessionMediator and SessionManager we could imagine this interface as a sort of "mf::ProtocolInterpreter"

Daniel wrote:
>> The common problem is the "frontend" component. While it exists, at least in that vague guise, it will
>> continue to cause confusion around the purposes of the things within it.

>> As far as I understand "frontend" means the protocol server "end". And by that understanding there should >> probably be no "Shell" or "Scene" within that. Maybe my understanding is amiss somewhere?

Sure there should be no Shell or Scene implementation there, but it may need an interface by which to interact with the Shell or Scene. We declare interfaces in the namespace they are used to improve radial encapsulation.

Chris Wrote:
>> Given that we're all agreed that this class needs to be refactored, I don't think there's much benefit in >> churning the name, but I don't strongly object either.

+1!