Hi all, would it make sense to have a "negative" test that ensures that no tbaa markers are emitted under "-O0" or "-Ox -fno-strict-aliasing" (x >= 1)? Perhaps there is one already or this case is covered in some other way. Thank you, Roger