In this chapter, we saw that programming with UDP sockets is somewhat easier than with TCP sockets. We learned that UDP sockets don't need the listen(), accept(), or connect() function calls. This is mostly because sendto() and recvfrom() deal with the addresses directly. For more complicated programs, we can still use the select() function to see which sockets are ready for I/O.
We also saw that UDP sockets are connectionless. This is in contrast to connection-oriented TCP sockets. With TCP, we had to establish a connection before sending data, but with UDP, we simply send individual packets directly to a destination address. This keeps UDP socket programming simple, but it can complicate application protocol design, and UDP does not automatically retry communication failures or ensure that packets arrive in order.
The next chapter, Chapter 5, Hostname...
