Thread: Network programming theory

Hi!

1. I've been working on an OpenGL-based action game, and I've just arrived to the network part.
Initially this is what i've planned: I'm using one server, x clients, and the server is looping through a list of the clients (and the server itself if its not dedicated). The problem is that I don't know exactly what they will send to each other.
If the clients send wich units (players, AI, powerups, firing bullets, everything) they're moving, creating or destroying, what should the server answer? Either it's just sending back the time since last time it listened to the client and let it calculate everything, or should it calculate the changes made itself and send back the whole result? If the latter, the latency can become huge on bigger maps, but I guess it will be a lot safer.

2. One more question: Should I use WinSock?

Thanks,
/KJ

2 Things.

1 - If its an action game you are going to want to use UDP rather than TCP (read the guide
Salem suggested)

2 - No, you shouldn’t use WinSock if you want your game to be 100% portable.*

*That being said it’s basically the same implementation as Berkeley sockets which is on most systems so porting your game isn’t to hard and every Windows system has some version of WinSocks so you wont need to distribute extra libraries.
So yes I would recommend using it.

Thanks, I checked up those tutorials, but I didn't really find what I was looking for. The problem is that I don't know what the server will send to the clients.
Either the server recieves changes from every client, calculate (if a player has moved, if a bullet is proceeding in its direction) and send it to everyone, or if the server just send back the changes from all the other clients. I believe that sending every new situation will take to long time but incremental updates sounds unsafe to me.