Add Spectator Mode

[Cheeseness]

We'd like to allow late joiners to be able to watch an in-progress game.

[Cheeseness]

Following some IRC discussion, it looks like we're considering some larger scale changes for spectator cameras than seem necessary.

I'm note sure that we need to bother with syncing spectator or player camera locations across the network (unless we want to make spectators visible to players/each other, but is there much value in that?)

My suggestion would be to allow spectators to control a freely moving camera with no synchronisation. It looks like this could be implemented without any significant networking changes.

If we did want to allow a spectator to follow a player, we could position the camera relative to already synced cart position/orientation and use HOTween to interpolate camera movement client side (we're already using it for other stuff, so we can quickly implement using that and see what kind of impact it has). We'll end up with a little camera lag, but that might end up looking pretty neat anyway.

[thegsm]

I actually think this will make some things simpler for spectator view to sync it and the change may not prove to be that significant.
Anyway, at this stage I recommend delaying spectator view altogether until we have closed the other elements related to networking (but keep that in mind).

[Cheeseness]

I actually think this will make some things simpler for spectator view to sync it

Can you expand some more on the value of camera/spectator positions here? Involving network synchronisation of that seems like it would make things less simple than not doing so. If we're already passing cart position orientation and orientation around the network, that's all we need for knowing where the cart's camera position is (that's what the cart's camera is positioned relative to in the first place).

Anyway, at this stage I recommend delaying spectator view altogether until we have closed the other elements related to networking (but keep that in mind).

Are you referring to a code tidy/refactoring of the networking system structure? If so, I think I agree. It'd be nicer to clean things up before potentially clouding them up more.

Right now, having unjoinable in-progress matches is costing us a little momentum and enthusiasm from people who're interested in testing though, and that's worth considering.