Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!cica!sol.ctr.columbia.edu!lll-winken!elroy.jpl.nasa.gov!decwrl!shelby!portia!bugboy From: bugboy@portia.Stanford.EDU (Michael Frank) Newsgroups: comp.unix.questions Subject: Re: Socket stuff Message-ID: <9498@portia.Stanford.EDU> Date: 24 Feb 90 01:16:53 GMT References: <4119@rouge.usl.edu> Sender: Michael Frank Reply-To: bugboy@portia.Stanford.EDU (Michael Frank) Distribution: usa Organization: Stanford University Lines: 33 Peter C. Barhs says: >First, does anyone have a reference to a comprehensive text or TR dealing >with sockets (socket, bind, connect, accept, read, ERRORS) besides >the two standard UNIX Tutorials on IPC (sorry, I forgot the authors). > >If not, how about somebody put together a good ipc TR that uses all of >the features via source code examples (something that looks more legible >than our rwho.c code). I, too, would like a reference to the text, and good example source code for ipc. I'm trying to create a server-client system. It's on a SPARCstation, so I could use RPC, but sockets seem (?) more portable and more flexible. What I want to do, specifically, is this: I want my client processes to be able to find the server easily (at least on the host machine), and send various messages to it, and receive messages back (but not neccessarily in a 1:1 correspondence to the messages sent, as RPC seems to assume). I want the clients to be able to send and receive messages either with or without blocking, so that each client may have whatever kind of dynamics are appropriate to it. I want the clients to be able to time out on a blocked send or receive if they want to. And I want the server to queue up the messages from all the clients and handle them one at a time, and send messages back to various clients in response. The server should never block on a message send. And if a client dies, the server shouldn't break. Maybe it's in the manuals, but it's too much to swallow all at once without some good examples. Sockets are extraordinarily complex. Anyway, thanks for any help. Mike Frank