When I use UTF-8 encoding for my database. upper and lower() functions break (no longer process accented chars correctly). This is with the correct encoding fr-FR(at)utf-8 I think, for CTYPES et al.