Precomposed versus decomposed accents? Jamo versus precomposed Hangul characters? The Unicode code point is rarely useful thing to know about on its own, and code which assumes that one code point equals one "character", for whatever definition of a character is in use, is likely to work poorly with UTF-32.
