An automated program testing methodology and its implementation

• Authors: Dorothy M. Andrews;Jeoffrey P. Benson
• Venue: ICSE '81 Proceedings of the 5th international conference on Software engineering
• Year: 1981

This paper describes an automated testing methodology and an experiment performed to determine its effectiveness. The method is to insert in the program to be tested a number of “executable assertions,” statements about the program that trigger error signals whenever they are evaluated to be false (violated). A test-case is then developed for the program using actual values of the input variables. When the program is run, a plot is generated of the number of assertions violated versus the input variable values used. The resulting function is called the “error function”. Heuristic search algorithms can then be used to maximize this function and thereby automatically locate input values which cause the most errors to occur. The experiment included developing assertions for the program to be tested, choosing and inserting representative errors into the program, and implementing search and data collection algorithms for testing. The results indicate that combining executable assertions with heuristic search algorithms is an effective method for automating the testing of computer programs.