Code review comment for lp:~kdub/mir/RFC-surface-arrangements

> [side-comment] I still find this RFC branch useful for discussion, easier to
> change stubs than an implementation.
>
> Changed the api RFC to need no dynamically-sized client-allocated memory.
> In light of lp:~alan-griffiths/mir/first-pass-of-surface-spec-modification,
> this or that branch will probably needs some changes, as I think the ideas
> behind 'what is a MirSurfaceSpec' is a bit different between the two branches,
> and should be unified.

I don't think there are any fundamental incompatibilities between the approaches - in both cases a MirSurfaceSpec is a collection of surface attributes the client asks the server to apply.

In this branch the attribute in question is a collection of buffer streams in the other it the attributes are name, size and height. (There are some existing issues with buffer streams that are likely to cause more churn - like "class Surface : public BufferStream"!)

Or is there a difference in thinking I'm not yet aware of?