> Change all those little LC_ and LANG vars over now convinces st to behave.
The problem seems that st suppouse the input keyboard is a utf8 device, and in the case you use a non utf8 locale it detects non ascii characters as first byte of a multibyte utf8 character (if you have good luck, because it is possible it detects it as an incorrect utf8 number). This is a new bug in st (at least the temporary work around is easy).
