Automating system testing for programs with graphical user interfaces(GUIs) presents many challenges for the tester. There are an enormousand potentially unbounded number of ways to interact with the softwareunder test and specifications are often limited. Yet test generationtechniques need to adequately cover this space. In this talk wepresent some recent work on a family of model-based GUI coveragecriteria grounded in combinatorial interaction testing that considerscontext in terms of event combinations, sequence length, and absolutepositions. We present an overview of some challenges and potentialsolutions for achieving this coverage in practice, and show theresults of some studies exploring the difference in theireffectiveness and cost. We end with a discussion of a potential newuse for information that is learned during context-based testgeneration.
Monday, May 07, 2012
Free and open to the public