Identifying test-critical dependencies

Regression testing is a major cost driver during software maintenance. An important approach to reduce maintenance costs is therefore to improve soft-ware testability. The improvement efforts should concentrate on those parts of a software system that cause the most severe problems during testing and maintenance. This paper describes a new approach to testability improvement focusing on system structure. We identify dependencies that are critical for testing, i.e. test-critical dependencies, based on a set of testability metrics. The results of four case studies show that 1) a small subset of the dependencies within a system has an exceedingly, high impact on particular testability metrics, 2) conventional coupling metrics are not good predictors of these test-critical dependencies, 3) dependencies automatically identified to be test-critical are good indicators of design that needs improvement.