📎 Webclip
The Art of Software Testing by Glenford Myers
The article treats The Art of Software Testing as an important but partly outdated book. It says most chapters can be skimmed, while Chapter 2 carries the core ideas that still matter for testing and software engineering.
Reading notes#
- The book has three editions, and the author considers it one of the top books in software testing despite some outdated content.
- Most of the eleven chapters are not worth close reading because their material is either covered better elsewhere or is too dated.
- Chapter 3 has an eleven-page checklist for code reviewers, but it is not comprehensive and is no match for Code Complete.
- Chapter 5 covers unit testing, but the article says Growing Object-Oriented Software, Guided by Tests is a better source today.
- The book’s most valuable part is Chapter 2, which focuses on the psychology of testing.
- Myers argues that testing is meant to find errors, not to demonstrate that software works correctly.
- The article highlights two linked ideas: testing aims to execute a program with the intent of finding errors, and a test case fails when it produces correct results without finding any errors.
- The author says these ideas should change how testers approach their work, pushing them to try to break software rather than confirm it.
- Myers also says a program cannot be tested into being error-free and that finding all errors is impractical or impossible.
- For test completion, the article favors using detection of a predefined number of errors as the stopping criterion.
- It says this idea can work in real projects, even if teams often resist it at first.
- The article closes by treating Chapter 2 as the foundation of the book and saying the software testing industry still works against those principles.
