Abstract:
Software is widely used technology of this era. Applications are being developed, solving
tasks ranging from simple calculations to complex banking transactions and so on. Use
of software in critical systems leaves no space for even simple errors which can lead to
sever nancial loss or threat to human life. Software must be well tested to verify its
functioning. Manual testing is tedious and time consuming job. Object oriented programs
use classes and their interactions to perform tasks. Integration testing requires to test
the interfaces. Path explosion make it worst to test the integration of classes. Small
work done in integrated test case generation of OO programs. Coupling criteria is used
for generation of automated test cases in proposed approach. Search space is reduced
by selecting only coupling methods that are directly involved in integration. Objects
must be in proper state before testing process started. Further def-use analysis helps in
achieving the desired object states for proper interfaces testing, representing methods in
intermediate tree. Method sequences are generated considering data
ow of involved state
variables. Test can't be accomplished without proper input data. We propose a tness
function for test data generation considering coupling path coverage. Genetic algorithm
is used to optimize the solution based on proposed tness function. Eleven randomly
selected project are used from SF100 (Software Testing Benchmark) to show the strength
of proposed approach. Results showed relatively high coverage as compared to random
testing.
