CodeGuru : Thinking in C++

Object-Oriented Design with Applications 2 ^nd edition (??) by Grady Booch, Benjamin/Cummings, 1993 (??).The Booch method is one of the original, most basic, and most widely referenced. Because it was developed early, it was meant to be applied to a variety of programming problems. It focuses on the unique features of OOP: classes, methods, and inheritance. This is still considered one of the best introductory books.

Object Analysis and Design: Description of Methods , edited by Andrew T.F. Hutt of the Object Management Group (OMG), John Wiley & Sons, 1994. A summary of Object-Oriented Analysis and Design techniques; very nice as an overview or if you want to synthesize your own method by choosing elements of others.

Before you choose any method, it’s helpful to gain perspective from those who are not trying to sell one. It’s easy to adopt a method without really understanding what you want out of it or what it will do for you. Others are using it, which seems a compelling reason. However, humans have a strange little psychological quirk: If they want to believe something will solve their problems, they’ll try it. (This is experimentation, which is good.) But if it doesn’t solve their problems, they may redouble their efforts and begin to announce loudly what a great thing they’ve discovered. (This is denial, which is not good.) The assumption here may be that if you can get other people in the same boat, you won’t be lonely, even if it’s going nowhere.

This is not to suggest that all methodologies go nowhere, but that you should be armed to the teeth with mental tools that help you stay in experimentation mode (“It’s not working; let’s try something else”) and out of denial mode (“No, that’s not really a problem. Everything’s wonderful, we don’t need to change”). I think the following books, read before you choose a method, will provide you with these tools.

Software Creativity , by Robert Glass (Prentice-Hall, 1995). This is the best book I’ve seen that discusses perspective on the whole methodology issue. It’s a collection of short essays and papers that Glass has written and sometimes acquired (P.J. Plauger is one contributor), reflecting his many years of thinking and study on the subject. They’re entertaining and only long enough to say what’s necessary; he doesn’t ramble and lose your interest. He’s not just blowing smoke, either; there are hundreds of references to other papers and studies. All programmers and managers should read this book before wading into the methodology mire.

Peopleware, by Tom Demarco and Timothy Lister (Dorset House, 2 ^nd edition 1999). Although they have backgrounds in software development, this book is about projects and teams in general. But the focus is on the people and their needs rather than the technology and its needs. They talk about creating an environment where people will be happy and productive, rather than deciding what rules those people should follow to be adequate components of a machine. This latter attitude, I think, is the biggest contributor to programmers smiling and nodding when XYZ method is adopted and then quietly doing whatever they’ve always done.

Complexity, by M. Mitchell Waldrop (Simon & Schuster, 1992). This chronicles the coming together of a group of scientists from different disciplines in Santa Fe, New Mexico, to discuss real problems that the individual disciplines couldn’t solve (the stock market in economics, the initial formation of life in biology, why people do what they do in sociology, etc.). By crossing physics, economics, chemistry, math, computer science, sociology, and others, a multidisciplinary approach to these problems is developing. But more importantly, a different way of thinking about these ultra-complex problems is emerging: Away from mathematical determinism and the illusion that you can write an equation that predicts all behavior and toward first observing and looking for a pattern and trying to emulate that pattern by any means possible. (The book chronicles, for example, the emergence of genetic algorithms.) This kind of thinking, I believe, is useful as we observe ways to manage more and more complex software projects.

Analysis & Design