|
|
In Chapter 9, the notion of testability was introduced. It was noted that testability was a multidimensional concept. It had many different attributes. It turns out that the entropy measure is yet another attribute dimension of testability. Low-entropy systems will be very difficult to test. By their very nature, such systems distribute a disproportional amount of their total epochs to a relatively small set of program modules. It is very difficult to impossible, under these circumstances, to create operational (or functional) test cases that will distribute their execution across all modules. The vast majority of test activity will always be spent in a small number of modules. A program module that is executed with a very low probability will perhaps require a large number of epochs to ensure it executes. Object-oriented design, by its very nature, tends to generate very low entropy systems. Testing in this environment can be very costly.
|
|