From: Brian <bga...@gm...> - 2009-08-30 20:33:07
|
Hi everyone, I am happy to report I am almost done with Phase 2 of my roadmap. I just had two players play on a LAN:) My next test will be to try 3 players on a LAN. I reworked the architecture so that it does peer-peer network play. This should allow more players to play together on the Internet. The bandwith requirements aren't to bad. There are still some improvements I need to make for better play on the Internet, but I am confident I will get there. If you want to try it out you need the sdl-port version with the latest code. You can run two copies of torcs-ng if you have a fast enough computer(dual core?). You will need to do the following. 1. Create a torcsclient directory in your home folder. 2. Create a torcsserver directory in your home folder. torcs-ng -l /home/brian/torcsngclient torcs-ng -l /home/brian/torcsngclient use 127.0.0.1 for the ip address to connect too. Make sure that you name the human player something other than player. Brian |
From: Annick et Jean-P. <jpm...@fr...> - 2009-08-31 20:43:44
|
Hi, Brian, and all. > I am happy to report I am almost done with Phase 2 of my roadmap. I > just had two players play on a LAN:) Great news ! > My next test will be to try 3 players on a LAN. I reworked > the architecture so that it does peer-peer network play. This should > allow more players to play together on the Internet. What do you mean by "peer-to-peer" here ? Does this means that the communications between the clients and server do no longer follow the start scheme (server at the center of the star) ? Cheers, Jean-Philippe. |
From: Brian <bga...@gm...> - 2009-08-31 22:39:27
|
> > > What do you mean by "peer-to-peer" here ? Does this means that > the communications between the clients and server > do no longer follow the start scheme (server at the center of the star) ? > > Before the race, it is a client-server system with the server choosing the race settings and everyone connecting to the server. During the race, I am using what some people call a hybrid system. Each player(client) does their own physics and sends/broadcasts their car position information to all other players. The host still keeps track of somethings such as the winner of the race and a few other things. In my original plan I was worried the physics wouldn't work right without the server doing all of the physics. My testing has shown that clients doing physics works ok. The peer-to-peer hybrid reduces the latency and bandwidth requirements. All robots are still controlled on the server. Brian |
From: Brian <bga...@gm...> - 2009-08-31 22:49:40
|
If anyone is interested in understanding network gaming here is a very good link. http://www.microsoft.com/downloads/details.aspx?FamilyID=db0854bc-7780-49c2-a507-0b86da927152&displaylang=en <http://www.microsoft.com/downloads/details.aspx?FamilyID=db0854bc-7780-49c2-a507-0b86da927152&displaylang=en>It talks about Microsoft technologies, but everything in applies to network gaming. Including what I am doing. Brian On Mon, Aug 31, 2009 at 6:39 PM, Brian <bga...@gm...> wrote: > >> What do you mean by "peer-to-peer" here ? Does this means that >> the communications between the clients and server >> do no longer follow the start scheme (server at the center of the star) ? >> >> Before the race, it is a client-server system with the server choosing the > race settings and everyone connecting to the server. During the race, I am > using what some people call a hybrid system. Each player(client) does their > own physics and sends/broadcasts their car position information to all > other players. The host still keeps track of somethings such as the winner > of the race and a few other things. In my original plan I was worried the > physics wouldn't work right without the server doing all of the physics. My > testing has shown that clients doing physics works ok. The peer-to-peer > hybrid reduces the latency and bandwidth requirements. All robots are still > controlled on the server. > > Brian > > |