Sufficiently Large Number for Testing C++ Standard Template Library Modules

One of the most crucial questions in software testing is when to stop testing. The problem was proved to be an undecidable problem in general. However, for a set of collection programs in C++ called standard template library (STL), we are able to show that there exist a sufficiently large number N associated with the test data set size of each module, such that testing with a data set size larger than N does not reveal more faults in the module under test. This paper presents the concepts, theorem, and the experimental results that confirm the finding of the sufficiently large number.

 Keywords: Sufficiently large number, Data Coverage Testing, C++ Standard Template Library, Automated Test Generation, Testing of Collection Programs

E-mail: cast@kmitl.ac.th