mySQL function behaves differently on 2 servers which are essentially the same - mysql

Question

Allright, here is a hard one...
I have a development server with mySQL 5.1.73 on which I wrote a function to normalize a string for searching purposes.
When moving the function to the production environment, same mySQL version, same major OS version (CentOS 6.5) newest patches, same major kernel version etc. The function stopped working.
Here is the function
CREATE DEFINER=`user`#`%` FUNCTION `normalize`(str VARCHAR(255)) RETURNS varchar(255) CHARSET utf8
BEGIN
DECLARE normstring VARCHAR(255);
DECLARE i INT;
SET i = 0;
SET normstring = '';
SET str = lower(str);
loop1: WHILE i < length(str) DO
CASE substring(str,i,1)
WHEN 'ä' THEN SET normstring = concat(normstring,'ae');
WHEN 'ö' THEN SET normstring = concat(normstring,'oe');
WHEN 'ü' THEN SET normstring = concat(normstring,'ue');
WHEN 'ß' THEN SET normstring = concat(normstring,'ss');
WHEN '/' THEN SET i = i + 1; ITERATE loop1;
WHEN '.' THEN SET i = i + 1; ITERATE loop1;
WHEN '-' THEN SET i = i + 1; ITERATE loop1;
WHEN '(' THEN SET i = i + 1; ITERATE loop1;
WHEN ')' THEN SET i = i + 1; ITERATE loop1;
WHEN ' ' THEN SET i = i + 1; ITERATE loop1;
WHEN '\'' THEN SET i = i + 1; ITERATE loop1;
WHEN '\\' THEN SET i = i + 1; ITERATE loop1;
ELSE SET normstring = concat(normstring,substring(str,i,1));
END CASE;
SET i = i + 1;
END WHILE;
RETURN normstring;
END$$
DELIMITER ;
On the development server this converts 'Mönßtär' to 'moensstaer', but on the production server it converts it to 'mönßtä'
Changing
SET i = 0; and WHILE i < length(str)
to
SET i = 1; and WHILE i <= length(str)
corrects the missing last character, so the result is 'mönßtär' but one server should not start counting with 0 the other one with 1, right?
And the production server leaves all special characters untouched.
I have compared all global variables, not only those explicitly set in my.cnf, and except timezone, password and symlink setings they are equal (yes I should correct those differences, but that should have nothing to do with my problem, right?)
Are there some compile-settings which can influence this behaviour, or some external libraries that mySQL uses?
I'll probably have to find a workaround for the problem - I plan to normalize in the application rather than the database - the function is too slow in large queries anyway - but it would have been nice to convert the existing data in the database. But I'm really curious as to what causes such strange behaviour.
Character-Set settings on both servers (from the running environment):
character_set_client........................ utf8
character_set_connection.................... utf8
character_set_database...................... utf8
character_set_filesystem.................... binary
character_set_results....................... utf8
character_set_server........................ utf8
character_set_system........................ utf8
collation_connection........................ utf8_unicode_ci
collation_database.......................... utf8_unicode_ci
collation_server............................ utf8_unicode_ci

Answer 1

I believe you have different character set between your local environment and production. Look into these articles on how to detect How do I find out the default server character set in mysql? and how to change Change server variable character_set_server.