Comment by derefr

> and a code point is still not a user-perceived character

How about indexing, measuring and slicing operations based on user-perceived characters, then?