Generic and Extensible Automatic Test Data Generation for Safety Critical Software with CHR

We present a new method for automatic test data generation (ATDG) applying to semantically annotated control-flow graphs (CFGs), covering both ATDG based on source code and assembly or virtual machine code. The method supports a generic set of test coverage criteria, including all structural coverage criteria currently in use in industrial software test for safety critical software. Several known and new strategies are supported for avoiding infeasible paths, that is paths in the CFG for which no input exists leading to their execution. We describe the implementation of the method in CHRv and discuss difficulties and advantages of CHR in this context.