Re: Network protocol for Expert QotW #23
Daniel Martin <martin-+m399P62/[email protected]>
| Newsgroups | gmane.comp.lang.perl.qotw.discuss |
|---|---|
| Message-ID | <[email protected]> |
Rick Measham <[email protected]> writes: > The protocol as defined by Daniel looks good. TCP is the go. Lets use > port 47898 which is an unassigned port at IANA (unless someone has > good reason for some other port. If you wonder where I pulled that one > from, grab your cell phone and type 'tetri' backwards) We actually need three port numbers - player one, player two, and spectators (unless you intend that those roles are assigned in order of connection) Of course, I think that the server and client should accept the port numbers as options on the command line, though we could define default ports. I envision the possibility of having a tetris meta-server program that people could connect to, request a game, and have the tetris meta-server start up a game and say "here are the port numbers to point your client at". The DOS possibility via REFRESH is, I think, a good argument that the server should delay the response to REFRESH until it would send a response anyway. Also, it should be specified that the first diff. sent to any client (especially to spectators) should be a full board diff.