An initial study identifying trends in test cases

Abstract

The paper presents a study where an attempt was made to understand the trends in test cases, generated for testing software programs. The main focus was to identify redundancies in the test cases that are generated, both, manually and in an automated fashion along with other significant trends. Three projects were considered for the analysis-two of which are widely used open-source projects and a course project. The analysis did reveal redundancies in test cases that could be easily avoided. It was attempted to equate redundancy with time. One of the projects was considered for the experiment, where the time to run the original test cases and the test cases after removing the redundancies was measured and compared. An improvement by about 40 % was observed. The analysis also revealed that the test cases were not designed to support all the environments/platforms the actual application does. This initial study shall, definitely, give a jumpstart for a full-fledged empirical study that needs to be performed to understand the actual impact of the redundancy in terms of time and cost. The study has made it clear that there are some significant concerns that need to be addressed when generating test cases for software programs. © 2013 Springer Science+Business Media.