Some systems are more difficult to test than others. Software testers contend with undefined or partially defined requirements; outdated, incomplete, or nonexistent documentation; complex logic; a mixture of languages; or worse. All of these factors make a system dirty, or virtually untestable.

In Testing Dirty Systems, authors William Perry and Randall Rice teach testers a six-step process for approaching such systems: system diagnosis; test planning; test execution; test analysis; report development; dirty system repair.

Because of the unknown characteristics of the dirty system, the traditional validation of comparing actual processing results against the expected processing results is often inadequate. Analysis of a dirty system must go much further into describing the expected operational characteristics of the system, including

• probability of failure based on failures during testing
• expected difficulty of making changes based on inadequacy of documentation
• estimate of defects that remain in the system
• operating conditions that will lead to failures
• coverage levels based on code or test cases
• complexity levels based on coding structure

Project leaders, independent testers, quality assurance personnel, and IS auditors will benefit from this book, as well as end-users and customers with a vested interest in the success of their systems.