At first glance it would be more natural to introduce a Buffer::write() method instead of a class that acts on the buffer, unless we want to potentially support multiple implementations of it.
Perhaps the logic in this method would be more clear if we split it out into different stages, e.g.:
1. grow touchpot renderable vector as needed (already there)
2. for each touch: set up and add renderable to scene
3. for unused renderables: remove from scene
Not a complete review yet, just some first notes:
162 +class BufferWriter
At first glance it would be more natural to introduce a Buffer::write() method instead of a class that acts on the buffer, unless we want to potentially support multiple implementations of it.
1577 +void mi::TouchspotCo ntroller: :visualize_ touches( std::vector< Spot> const& touches)
Perhaps the logic in this method would be more clear if we split it out into different stages, e.g.:
1. grow touchpot renderable vector as needed (already there)
2. for each touch: set up and add renderable to scene
3. for unused renderables: remove from scene
1618 + scene-> emit_scene_ changed( );
We should call this out of lock if possible.