Best Kept Secrets of Peer Code Review: Modern Approach. Practical Advice. (Modern Approach. Practical Advice.)

This is a book with an agenda: selling Smart Bear Software's multi-user peer code review application. So you have to keep in mind as you read: how much of what's written makes sense on its own, and how much is carefully worded to lead you to the desired conclusion--that Smart Bear are the experts and their product is the greatest thing since sliced bread? That said, the book, or rather, collection of essays, is a good discussion-starter about code review. The chapter on Five Types of Code Review is a bit too assertive for me. I thought the chapter on Personal Software Process and code review was excellent. And the selling point of the book is the data they've gathered. Everything they could find on the Internet about actual industrial code review data (not much) plus their own real-life study at one of their customers--CISCO. So if you're ready to do code review, then just do it, and then read this book to compare notes. If you're not ready yet, this book will certainly make you wonder why not.

First, let me say that I have decidedly mixed experience of code inspections. At their best, I've seen them bring many diverse skills to bear on subtle problems, creating a product better than any one of the very capable developers could have built on their own. At worst, inspections range downwards from useless, not just worthless in themselves but divisive and demoralizing to the development group. I want to believe the process can be used safely and effectively, but I have some hard-won doubts. The fact that the publisher sells a technical product for managing the review process added other doubts. I'm happy to say that my doubts are addressed - not banished utterly, but faced head on. Despite a bit of jargoneering (is "trial" really a verb?), the authors present updated techniques that appear to reduce some of the problems working against effective inspections. And no, the techniques don't demand that you buy their product first. The techniques dovetail well with modern ideas, including the Law of Demeter (p.46, on module coupling), SEI's PSP, and the rich IDEs common on developers' desktops today. Finally, the last chapter - and only the last chapter - addresses the company's product. Despite the book's commercial origin, it has lots of great information beyond the writers' product. In fact, it reminds me of "Planecraft," written over 70 years ago. That book was written as a sales tool for the greater corporate glory of a company selling woodworking planes, but has enough worthwhile technical content that it was reprinted a few years back. Maybe "Best Kept Secrets" won't last 70 years, but I was struck by the analogy. The book's discussion discussion is wide-ranging, readable, and helpful. A few editing glitches put potholes in the reading path, but only a few. On the whole, I came away somewhat more willing to throw myself back into the fray of inspections, as long as they're run carefully, in accordance with many of the ideas shown here. //wiredweird, reviewing a copy given out as conference swag