i have to write simple tcp messanger on linux. It has to have some basic functionality - when connecting via telnet, it must react to basic commands as put, get and exit and so on and when connecting by my own client every message must be delivered to everyone talking without any commands (real time).
So what do i need to write it? The server must surely have some shared memory for it's subproccesses to communicate - but what would be the best - PIPE or unnamed PIPE? I know that all these processess are "family" so unnamed PIPE would be normally best. But how can i reach for previous messages (telnet issue) when using unnamed PIPE? Is it possible at all?
One more thing: named PIPE stops everything when opened. So this subprocesses must be divided to 2 more - one for reading and one for writing?