Currently, Ruzzy does not have any automated tests for real fuzzing. There are only tests for bad path behavior, and not any for happy path behavior. We should add automated tests for real fuzzing. This will allow us to move faster and more confidently with changes to the tool.
Open question: how to do it? Should we do something like -atheris_runs from Atheris? Once we call LLVMFuzzerRunDriver we lose control of the process and the ability to make assertions. Will we need some kind of subprocess test model?
Currently, Ruzzy does not have any automated tests for real fuzzing. There are only tests for bad path behavior, and not any for happy path behavior. We should add automated tests for real fuzzing. This will allow us to move faster and more confidently with changes to the tool.
Open question: how to do it? Should we do something like
-atheris_runsfrom Atheris? Once we callLLVMFuzzerRunDriverwe lose control of the process and the ability to make assertions. Will we need some kind of subprocess test model?