Qaia

Is it adviseable to implement a custom garbage collector for a game engine written in C++, or just delete objects immediately as soon as they become unreferenced?

If a GC should be implemented, what criteria / guidelines should be adhered to (in terms of when the GC should be called)?

What's the preferred method to select units with the mouse, i.e., you have a unit on screen that you want to either control or see its stats and you hover your mouse over the unit and click to select it?

I was thinking of reading the (x,y) screen coordinates (not the world coordinates) of the unit and comparing that with the (x,y) mouse coordinates. But I'm not sure if this approach is commonly used or efficient. I was also thinking if it would be possible to associate a hitbox with the mouse and project that into the world.

Are there other approaches commonly used?

What is the most efficient and preferred way to implement static backgrounds involving full 3D camera rotation? For example, space and aquatic sims environments. The background is exactly that, just a backdrop to create the overall encapsulating environment allowing the character to explore the 3D universe unhindered.

I was thinking there are two general approaches:

1) Use a 3D model that encapsulates the universe (like a box) with the inside of the model textured; or

2) Use a 2D sprite sheet that maps coordinates on the sheet to the camera position.

The downside to 1) is that because it is a 3D model (needless) collision calculations would constantly be performed. The downside to 2) is that the engine has to constantly perform the mapping calculations and reload the sprite sheet accordingly.

Which of these two approaches is preferred? Are there any other approaches used?