> the complement operation would need to take the complement inside the set of valid UTF-8 strings rather than the set of all byte-sequences Anything else woul

> the complement operation would need to take the complement inside the set of valid UTF-8 strings rather than the set of all byte-sequences Anything else would be a bug in any case.