History: CharacterEncodingTrouble

All the data in Tiki are supposed encoding with UTF-8

If you are looking at this site and you see some strange character when you choose a non english language, it is perhaps that your browser is interpreting the UTF-8 encoded output as Latin-1 which is the default in Western Europe. (the characters look like 'æ' for 'æ', 'ø' for 'ø', and 'Ã¥' for 'å')

The UTF-8 encoding is backwards compatible with ASCII, byte for byte, but not with anything else. The Universal Character Set defined by Unicode is a superset of Latin-1 which is a superset of ASCII, but when characters from UCS are encoded in UTF-8, then everything except for plain 7-bit ASCII end up as between two and six bytes. That's why 'æ', 'ø', and 'å' turns into two-letter combinations when an UTF-8 encoded text is viewed as Latin-1.

If Tiki on your server doesn't look fine:
Some server (like with the default Debian config) adds a charset=iso-8859-1 to the Content-Type header. The browser (ex: Mozilla) first looks for a charset value in the Content-Type header and then for META tags, so the header overrides the META tag inserted by TikiWiki.