The Practice of Programming

This book is basically all the "common sense" stuff that you learn after programming for years and years.. most proffessional programmers already know this stuff--or should! This is a perfect book for a college graduate who is good, but needs some pointers in the real world, or for those who just want to freshen up their skills, etc. Beware though, this book is not "Programming for Dummies" you must already be well familiar with C/C++ and some algorithms and data structures to fully understand what the author is trying to convey, at times. But even if you don't, it's still a great book... it is (or will be) one of those classics like The C Programming Language, which every programmer has on their bookshelf. Pretty much any book written by Kernighan or published by A-W (Professional series) is a worthwhile read...In a nutshell, as they say in the book: this is what most people should have learned in college, but couldn't or didn't. Get it if only to read the first chapter on style... I for one hate rewriting or reading people's code who didn't know how to write it well/clear, or comment it well.

I am bemused by the disparaging comments of my fellow readers in regard to this book. Kernighan and Pike make clear their intent in the first paragraph of the preface to this book; it is about simplicity, clarity, and generality.To be sure, there is tinder here for short tempers and delicate egos. If you're under the gun, trying to duct-tape together the fifth release of some huge, unwieldy application, this book does not contain the short-term quick fixes you've been hoping for. If you're righteously convinced of your own sound practices and don't care to look at someone else's methods, this book may irk you.Kernighan and Pike have written a book about the most basic habits and outlook that a programmer should have. They have not tried to address all facets of programming. Instead, they sacrifice scope to make their points stand out all the more clearly.Would this be a better book if they had cast their net wider? Hardly. If you start off by applying the carefully thought out, methodical approaches described clearly throughout this book, your code will still hit abstruse bumps and strange circumstances. But most problems will succumb to the same analytical ways of thinking and tools that Kernighan and Pike have laboured to describe with such clarity.But don't imagine that I think this book is perfect; the authors have been doing many things in the same ways for a long time. Most often, this is because their methods are effective, but sometimes they are far too close to being cop-outs. For example, the idea that it's OK to just print an error message and bomb out if something goes wrong is laughable outside of the Unix command line environment, and is rarely appropriate even there.Fundamentally, though, if you can't solve the problems at the ends of the chapters (they're easy), or you think you can't possibly benefit from reading a book that troubles to describe quicksort (which you probably learned in CS-201), then there is likely nothing for you here. At least until you think to question your perspectives. Until then, I'll confess to a sense of relief that I don't have to work with you :-)

In computing, the learning curve is doubly steep. Not only do we have to learn very complex operations, but we have to learn them at a pace unrivaled in any other field. Furthermore, the equipment improves at a rate that simply boggles the mind. In this frantic environment, we rarely have time to read our code twice, much less read a book about code. Therefore, when we do read, we must make every minute count. This is one book where your count of wasted minutes would be a very small one. Some of the tips in this book are obvious in retrospect, yet ones that you probably would not think of. My favorite is the fact that due to the changes in processors, a double precision floating point arithmetic operation can be faster than the equivalent one for integers. In the "old" days, the gospel was that you must avoid floating point operations unless absolutely necessary, to avoid the degradation of performance. Other tips, such as methods to assure you comment what is necessary, taking a few minutes to learn simple performance features, debugging and testing guidelines; portability issues and basic algorithm analysis should cause you to pause for a moment. Even in our hectic development environments, stopping and analyzing your code is a necessity. It is difficult to conceive of someone who will not find a tip in here that will justify the cost of the book. Unless of course, you are one of the authors. I listed it as one of the top books of the year in my On Books column that appeared in the September, 1999 issue of _Journal of Object-Oriented Programming_.

I've long recommended Pike's "Notes on Programming in C" on my web page. This book includes most of the content from that essay and much more, but is still thin and concise. What I like most about this book is that they justify all of their recommendations, show both good and bad examples, and keep the discussion grounded in actual code (rather than abstract principles).Other things I liked: - begins with a discussion of programming style and aesthetics - they critique some of the designs that they have been involved in, such as C's stdio and string handling libraries - they discuss the unique design issues presented by library design - they give examples in C, C++ and Java, and give an honest appraisal of the tradeoffs involved in each language. - FINALLY, excellent single chapter descriptions of systematic approaches to debugging and testing! - they face up to some of the tough design choices that must be made outside the UNIX Ivory Tower (rare for these authors). For example, they sacrifice UNIX consistency in one application so that the application will behave consistently across UNIX and Windows.Minor gripes: - still skirts around tough design issues in error recovery and reporting; they advocate the "print a diagnostic and exit" approach (which is totally inappropriate for library code), and don't discuss the tradeoffs - a few of the principles they cover will be trivial or obvious for experienced programmers

The book covers topics such as style, design, interfaces, notation, debugging, testing, performance tuning, and portability. Each of these topics gets its own chapter and is covered pretty throughly.The writing is generally very interesting though at times it does get a little too much into details. It must have been interesting. I have finished reading it in less than 24 hours after having received it. This is something I've never done with a computer book before!!The book does live up to its goal. I think any programmer (even with 20 years of experience) who reads this book will gain something from reading it. Before you think that I think the book is perfect, I will point out some areas where I think it could be improved. First, I think the book is too focused on C. This is not surprising given that Kernighan is one of the authors. Personally, I have always believed that C is a language that encourages bad programming practices. In a few places in the book, I see some of those in the code that is presented. The other area I was disappointed in was that there was not much on object-oriented programming practices. Beyond that, I saw something I do not believe should be in any good OO program. That something was a public instance variable. A few places I disagreed with the authors on style issues. However, all of these problems are fairly minor and will hopefully be corrected in the next edition.In conclusion, I think this is a book that all programmers should read at least once. I know I will keep my copy on my shelf for those times when I have questions like "how can I optimize this code?". It really is the best general programming book that has been published in years.