Reviews for GNU/Linux Application Programming, Second Edition (Charles River Media Programming)

1. Lazy authoring and dodgy code

   Rated 3 out of 5 stars, October 12th, 2007

   Bought this book as it had so many good reviews on Amazon.com. I really should have paid more attention to the one review that said "Disappointing" (Paul Floyd of Grenoble, France).
   
   I haven't yet read the bits about history, tools or shell scripting. What I have read is the section covering application development. This skims through areas such as sockets, threading, semaphores, mutexes, message queues, memory-mapped files etc. All useful areas, but this book does little more than tell you what the man pages tell you. That's where the first bit of lazy authoring comes in. The second bit of lazy authoring is the complete absence of an explanation of how to use these areas together, or an example of using them together. Given the list of topics, an example would have been useful that starts a worker thread to handle a TCP connection, that thread waiting on file descriptors and a timeout using select or poll, using mutexes to protect data, a message queue to communicate between the main thread and the worker thread, and possibly a memory-mapped file to create a circular log of the last N actions performed. Unfortunately the author didn't attempt that. But we should possibly consider ourselves lucky that he didn't, as the example code he does provide contains some major failings, most notably putting code inside assertions that is required even in optimised release builds. When this code gets compiled out in an optimised release build the examples fail. Obvious to anyone that knows about assertions, but not necessarily to everyone reading the book.