Test Run: Randomness In Testing

摘要

Creating and using random test case data is an essential software testing skill. Although most test case data consists of specific inputs to the system under test and specific expected values/states, you will almost always want to subject your system to random test case inputs, too. Typically you do this to see if you can cause a crash or evoke an exception by throwing a large variety of inputs at the application. In this month's column, I'll explain four common tasks when dealing with random test case data in a Microsoft~® .NET Framework environment: 1. Generate pseudo-random numbers (Knuth algorithm) 2. Analyze a pattern for randomness (Wald-Wolfowitz test) 3. Shuffle a list of items (Fisher-Yates algorithm) 4. Generate Gaussian numbers (Box-Muller algorithm).