In tr_TR.UTF-8 locale tolower('I') != 'i' (this is not a bug in glibc,
tolower('I') is suppossed to return dotless i that is 2-byte long).
Because of that strcasemp("i", "I") returns that the strings are
different.
Signed-off-by: Piotr Caban <piotr@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>