Incremental identification of categories and choices for test case generation: a study of the software practitioners' preferences

Abstract

Test case generation is a vital procedure in the engineering of test harnesses. In particular, the choice relation framework and the category-partition method play an important role, by requiring software testers to identify categories (intuitively equivalent to input parameters or environment conditions) and choices (intuitively equivalent to ranges of values) from a specification and to systematically work on the identified choices to generate test cases. Other specification-based test case generation methods (such as the classification-tree method, cause-effect graphing, and combinatorial testing) also have similar requirements, although different terminology such as classifications and classes is used in place of categories and choices. For a large and complex specification that contains many specification components, categories and choices may be identified separately from various kinds of components. We call this practice an incremental identification approach. In this paper, we discuss our study involving 16 experienced software practitioners and three commercial specifications. Our objectives are to determine, from the opinions of the practitioners, (a) the popularity of an incremental identification approach, (b) the usefulness of identifying categories and choices from various kinds of specification components, and (c) possible ways to improve the effectiveness of the identification process.