Talk Protocols
Talk protocols allow two or more users to communicate, using a telnet-like client, with separate send and receive windows.Unfortunately, a number of versions have proliferated, including: Talk, nTalk (New Talk), yTalk, oTalk, etc.
As a result, Talk protocols have been overshadowed by IRC, MUDs, IPhone, et. al. -- however, many Unix systems still support them.
Overview
Talk can be implemented as a separate talk server daemon and client, or a combined application that works as both server and client.The server side opens a UDP port (517 for talk, 518 for ntalk), and listens for incoming requests to start a talk session.
The client side generally initiates a talk session by asking for a particular user on a host: user@host.dom -- the server resolves the user name to an existing TCP port, or notifies the target user that a request has come in.
Once the session has started, everthing a user types in their send window is sent the the other user; you receive everything typed by the other user in a scrolling receive window.
Many talk implementation support special escape key sequences that allow controlling your (or the other user's) client.
References
Talk MAN Pages
nTalk MAN Pages
Internet Relay Chat (IRC) - RFC 1459
Multi-User Domains (MUD)
For comments/correction/additions regarding this reference, email specs@graphcomp.com.
