unset LC_ALL
LC_CTYPE=en_US.UTF-8
export LC_CTYPE

# a umlaut
s=$(printf '\303\244')
# euro sign
s=$s$(printf '\342\202\254')
# some sort of 't' outside BMP
s=$s$(printf '\360\235\225\245')
set -- "$s"
[ ${#s} = 3 ] && [ ${#1} = 3 ]