User Comments

Tip: Don't accept the defaults. If you're going to enter text data that is written in a Western alphabet, choose UTF-8, but if you're going to be entering mainly Asian or other non-Western text, choose UTF-16.

Here's why: Latin 1 (a.k.a. ISO-8859-1) is officially obsolete as it doesn't include the Euro symbol. Latin 9 (ISO-8859-15) is the replacement (it includes € - the Euro symbol). However, it still doesn't represent certain accented letters common in Slavic names, such as Č (C with a v-shaped hat) and Ł (L with a stroke across it) and Ž (Z with a hat), which someone will almost certainly try to input one day, if you live anywhere in Europe. UTF-8 can represent all those Slavic letters and anything else you throw at it, but it is compact when storing Western text. English language text previously stored as US-ASCII or Latin-1 can be transferred in without conversion, so there's no reason not to, really.

UTF-8 can represent Thai, Chinese, Arabic, Hebrew, and Hindi and Gujerat letters, too, but it is less compact than UTF-16 for representing such text. UTF-16 stores all such letters as two bytes, but UTF-8 encodes them into strings of up to six bytes for each letter.

My suggestion, therefore: Western text = UTF-8, any other text = UTF-16 (unless, perhaps, this would involve you in converting text from something else, such as BIG5 or shift-JIS).

As an aside, Windows prefers two-byte characters, and so do Java and C#, so if you're using those platforms or tools, UTF-16 is the way to go, regardless of what language your text is in.

Add your own comment.