Hacker News Reader

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

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

I think the number of displayed characters is even font-dependent.

Because even that is not exactly trivial, particularly for noon-Latin languages.