I think the whole problem is keeping the character encoding consistent in the applications and their dependencies. Programmers often forget this because they avoid non-ASCII characters in their code.