https://github.com/michaelrj-google commented: The tests look good overall, but make sure you're careful when using randomness in tests. A test failure isn't helpful if you can't figure out what caused it. https://github.com/llvm/llvm-project/pull/94156