Variable integrity testing of software

This paper provides a method of variable integrity testing to detect the faults of software. By checking the definition domain of the variable itself and the restriction condition of consistency between variables, we only need to determine the bound of each output data, rather than the exact value. Although the output of testing case cannot be easily caught, this method can improve the efficiency of detecting faults. Furthermore, the checking point is not limited to the final output place, but spreads over the reference places in the program code to key variables just like those at the checkpoints under debugging. As a consequence, the testing becomes more precise.