Interesting. I guess there are no well working normalization functions as there are for the different code points that can make umlauts (äöü)?