The selfcheck catches a lot of bugs, but not all. We should have a testing framework that allows us to quickly add minimal tests for various features. In addition to the selfcheck, the equivalence of input and output should be empirically tested by wrapping it into an executable stub, running it (in an emulator) with random inputs, and comparing the results.
The selfcheck catches a lot of bugs, but not all. We should have a testing framework that allows us to quickly add minimal tests for various features. In addition to the selfcheck, the equivalence of input and output should be empirically tested by wrapping it into an executable stub, running it (in an emulator) with random inputs, and comparing the results.