Partial configuration of utf8 collations and charsets on MySQL 5.6 - mysql

Question

I've been trying various configurations to force charset and collations to utf8. However, everything I tried has failed and some settings reflect the latin charset or collation.
This is MySQL 5.6 on CentOS 6 from IUS repository.
My config on /usr/share/mysql/my.cnf
# The MySQL server
[mysqld]
port = 3306
socket = /var/lib/mysql/mysql.sock
skip-external-locking
key_buffer_size = 16K
max_allowed_packet = 1M
table_open_cache = 4
sort_buffer_size = 64K
read_buffer_size = 256K
read_rnd_buffer_size = 256K
net_buffer_length = 2K
thread_stack = 128K
init_connect='SET collation_connection = utf8_unicode_ci'
init_connect='SET NAMES utf8'
character-set-server=utf8
collation-server=utf8_unicode_ci
skip-character-set-client-handshake
Settings through Heidi
From the MySQL client
mysql> show variables like 'char%';
+--------------------------+----------------------------+
| Variable_name | Value |
+--------------------------+----------------------------+
| character_set_client | utf8 |
| character_set_connection | utf8 |
| character_set_database | latin1 |
| character_set_filesystem | binary |
| character_set_results | utf8 |
| character_set_server | latin1 |
| character_set_system | utf8 |
| character_sets_dir | /usr/share/mysql/charsets/ |
+--------------------------+----------------------------+
8 rows in set (0.01 sec)
mysql> show variables like 'coll%';
+----------------------+-------------------+
| Variable_name | Value |
+----------------------+-------------------+
| collation_connection | utf8_general_ci |
| collation_database | latin1_swedish_ci |
| collation_server | latin1_swedish_ci |
+----------------------+-------------------+
3 rows in set (0.00 sec)
I did not have any issues in 5.1 though.
Update #1
I've put default-character-set = utf8, in client, mysqld and mysql section of my.cnf and rebooted the MySQL service. Still the same thing, a mix of latin and utf8.
mysql> show variables like 'char%';
+--------------------------+----------------------------+
| Variable_name | Value |
+--------------------------+----------------------------+
| character_set_client | utf8 |
| character_set_connection | utf8 |
| character_set_database | latin1 |
| character_set_filesystem | binary |
| character_set_results | utf8 |
| character_set_server | latin1 |
| character_set_system | utf8 |
| character_sets_dir | /usr/share/mysql/charsets/ |
+--------------------------+----------------------------+
8 rows in set (0.00 sec)
mysql> show variables like 'coll%';
+----------------------+-------------------+
| Variable_name | Value |
+----------------------+-------------------+
| collation_connection | utf8_general_ci |
| collation_database | latin1_swedish_ci |
| collation_server | latin1_swedish_ci |
+----------------------+-------------------+
3 rows in set (0.00 sec)
Update #2
My issues related to migrating a database using Migration Wizard through MySQL Workbench. The resulting database has collations and charsets that I don't want. If I run the produced script through Heidi, I get the same result as the migration result.

Answer 1

init-connect is ignored when you connect as root (or other SUPER user).
In your my.cnf (my.ini) files, add this under [mysql] or [client]:
default-character-set = utf8
(I don't know if Heidi will override that -- find out how it does the "connect".)
Addenda
character_set_database and _server are not necessarily an issue. _client, connection and _results are used to say how the bytes in the client are encoded. You say what the table contains by adding CHARACTER SET utf8 in CREATE TABLE.
Aside from the settings being like you show them, what symptoms do you have of messed up text? Do this with some column with non-English text:
SELECT col, HEX(col) FROM ... WHERE ...
to show what is actually stored.

Answer 2

I ended up hard coding the collation and character set that I wanted in the generated scripts.
ie. when creating the database and adding columns, specifying the collation and character set where necessary, utf8_unicode_ci, utf8...