Introduction

Credit: Mark Hammond, co-author of Python Programming on Win32 (O'Reilly)

The first computer I had in my home was a 64 KB Z80 CP/M machine. Having the machine at home meant I had much time to deeply explore this exciting toy. Turbo Pascal had just been released, and it seemed the obvious progression from the various BASIC dialects and assemblers I had been using. Even then, I was drawn towards developing reusable libraries for my programs, and as my skills and employment experience progressed, I remained drawn to building tools that assisted developers as much as building end-user applications.

Building tools for developers means that debugging and testing are often in the foreground. Although images of an interactive debugger may pop into your head, the concepts of debugging and testing are much broader than you may initially think. Debugging and testing are sometimes an inseparable cycle. Testing will often lead to the discovery of bugs. You debug until you believe you understand the cause of the error and make the necessary changes. Rinse and repeat as required.

Debugging and testing often are more insidious. I am a big fan of Python's assert statement, and every time I use it, I am debugging and testing my program. Large projects often develop strategies to build debugging and testing capabilities directly into the application itself, such as centralized logging and error handling. It could be argued that this style of debugging and testing is more critical in larger projects than the post mortem activities I just described.

Python, in particular, supports a variety of techniques to help developers in their endeavors. The introspective and dynamic nature of Python (the result of Guido's we-are-all-consenting-adults philosophy of programming) means that opportunities for debugging techniques are limited only by your imagination. You can replace functions at runtime, add methods to classes, and extract everything about your program that there is to know. All at runtime, and all quite simple and Pythonic.

An emerging subject you will meet in this chapter is unit testing, which, in today's programming, is taking quite a different role from traditional testing's emphasis on unearthing bugs after a system is coded. Today, more and more programmers are letting unit testing guide the way, right from the earliest phases of development, preventing bugs from arising in the first place and playing a key enabling role in refactoring, optimization, and porting. Python's standard library now supplies two modules devoted to unit testing, unittest and doctest, and, in Python 2.4, a bridge between them, which you'll see highlighted in one of this chapter's recipes. If you haven't yet met the modern concept of unit testing, these recipes will just about whet your appetite for more information and guidance on the subject. Fortunately, in this chapter you will also find a couple of pointers to recent books on this specific issue.

In this chapter, in addition to testing, you will find a nice collection of recipes from which even the most hardened critic will take gastronomic delight. Whether you want customized error logging, deep diagnostic information in Python tracebacks, or even help with your garbage, you have come to the right place. So tuck in your napkin; your next course has arrived!