The Pragmatic Programmer revisited

The Pragmatic Programmer is an excellent book. Top notch.

The book shaped how I think about writing software and established a mental framework for getting code developed in an effective manner. I recently decided to revisit the book since seven years passed from my initial read-through. Now that I have much stronger feelings about software, forged from my personal experience, I was curious how the advice described in the book jived with my own thoughts. I also wanted to see how well the advice fit into the software world of 2017.

After so much time has passed, does this book from 1999 still matter? Yes, it does.

The key to this continued relevance is a focus on principles over specific practices. There are specific practices and code samples sprinkled throughout the text, but they exist to support a principle topic instead of being the main focus. The emphasis on principles creates a timelessness to the material in spite of the text being full of references to older technology like CVS. These principles cover a sprectrum from generating code to metaprogramming to working with client expectations.

After so many years away from the material, my favorite principle continues to be the section entitled Tracer Bullets. When I first read the book, I was working for a large defense contractor. The culture at work dictated a heavy process with an emphasis on a waterfall methodology. Tracer bullets made my brain explode (not literally, thankfully) when considering how software could be developed. The section is best read to get a full appreciation of the idea, but the core concept is that software can be implemented in an a way that generally points toward the client’s goal and honed in until it strikes the target. This method is in contrast to the meticulous planning and calculating that would be required to reach the goal on the first shot. As someone who spent his time considering the mission software of satellite systems, the idea of quick and cheap iteration was mind bending. Maybe this idea is self-evident for someone working in a web startup space, but it was ground breaking for a young guy steeped in a process heavy culture.

Where is the book lacking today? In my mind, the biggest difference is in the pace of development. For instance, we encounter the suggestion to do nightly builds. This was great advice for the time. Integration was hard in older version control tools and compilation could take much longer than on today’s hardware. In modern development, we use Continuous Integration services like Travis CI to make sure our code is integrated far more than nightly.

I think I would also actively tell other developers to be cautious with the appendix of resources. There are definitely great suggestions in there (like Python). On the other hand, many suggestions are likely eclipsed by more modern tools. You might want use Lua as an alternative to the book’s suggestion of Tcl for an embedded programming language, for example.

The Pragmatic Programmer isn’t aging perfectly, but it still does an amazing job of showing developers a way of thinking about development that is, well, pragmatic. It is a book that you should be sure to read. Read it. Seriously.