The complexity of software testing

The futility of using a general-purpose metric to characterise `the´ complexity of a program has been argued to support the design of specific metrics for the different stages of the software life-cycle. An analysis of the module testing activity is performed, providing evidence of the absurdity of all-purpose metrics, as well as a methodical means with which to measure testing complexity. Several standard metrics are seen to serve as component measures for the intricacies of testing. The methodology is applied to compare traditional and adaptive means of testing. It is shown that previous informal arguments asserting the superiority of adaptive methodologies are formally confirmed