Book Image

Hands-On Network Programming with C

By : Lewis Van Winkle
Book Image

Hands-On Network Programming with C

By: Lewis Van Winkle

Overview of this book

Network programming enables processes to communicate with each other over a computer network, but it is a complex task that requires programming with multiple libraries and protocols. With its support for third-party libraries and structured documentation, C is an ideal language to write network programs. Complete with step-by-step explanations of essential concepts and practical examples, this C network programming book begins with the fundamentals of Internet Protocol, TCP, and UDP. You’ll explore client-server and peer-to-peer models for information sharing and connectivity with remote computers. The book will also cover HTTP and HTTPS for communicating between your browser and website, and delve into hostname resolution with DNS, which is crucial to the functioning of the modern web. As you advance, you’ll gain insights into asynchronous socket programming and streams, and explore debugging and error handling. Finally, you’ll study network monitoring and implement security best practices. By the end of this book, you’ll have experience of working with client-server applications and be able to implement new network programs in C. The code in this book is compatible with the older C99 version as well as the latest C18 and C++17 standards. You’ll work with robust, reliable, and secure code that is portable across operating systems, including Winsock sockets for Windows and POSIX sockets for Linux and macOS.
Table of Contents (26 chapters)
Title Page
About Packt

Chapter 13, Socket Programming Tips and Pitfalls

  1. Is it ever acceptable just to terminate a program if a network error is detected?

Yes. For some applications terminating on error is the right call. For more substantial applications, the ability to retry and continue on may be needed.

  1. Which system functions are used to convert error codes into text descriptions?

You can use FormatMessage() on Windows and strerror() on other platforms to obtain error messages.

  1. How long does it take for a call to connect() to complete on a TCP socket?

A call to connect() typically blocks for at least one network time round trip while the TCP three-way handshake is being completed.

  1. What happens if you call send() on a disconnected TCP socket?

On Unix-based systems, your program can receive a SIGPIPE signal. It is important to plan for that. Otherwise, send() returns -1.

  1. How can you ensure that the next call to send() won't block?

Either use select() to make sure the socket is ready for more data or use non-blocking sockets.

  1. What happens if both peers to a TCP connection try to send a large amount of data simultaneously?

If both sides to a TCP connection are calling send(), but not recv(), then they can be trapped in a deadlocked state. It is important to intersperse calls to send() with calls to recv(). The use of select() can help inform your program about what to do next.

  1. Can you improve application performance by disabling the Nagle algorithm?

It depends on what your application is doing. For real-time applications using TCP, disabling the Nagle algorithm is often a good trade-off for decreasing latency at the expense of bandwidth efficiency. For other applications, disabling it can decrease throughput, increase network congestion, and even increase latency.

  1. How many connections can select() handle?

It depends on your platform. It is defined in the FD_SETSIZE macro, which is easily increased on Windows but not on other platforms. Typically, the upper limit is around 1,024 sockets.