31

utf16 or utf32? I'm trying to store content in a lot of languages. Some of the languages use double-wide fonts (for example, Japanese fonts are frequently twice as wide as English fonts). I'm not sure which kind of database I should be using. Any information about the differences between these four charsets...

Improve this question

2 Answers 2

Reset to default
43

MySQL's utf32 and utf8mb4 (as well as standard UTF-8) can directly store any character specified by Unicode; the former is fixed size at 4 bytes per character whereas the latter is between 1 and 4 bytes per character.

utf8mb3 and the original utf8 can only store the first 65,536 codepoints, which will cover CJVK (Chinese, Japanese, Vietnam, Korean), and use 1 to 3 bytes per character.

utf16 uses 2 bytes for the first 65,536 codepoints, and 4 bytes for everything else.

As for fonts, that's strictly a visual thing.

"The Absolute Minimum Every Software Developer Absolutely, Positively Must Know About Unicode and Character Sets (No Excuses!)"

See also MySQL documentation for Unicode support.

Improve this answer
5
  • 6
    Just to be extra-clear, the comment about utf8_general applies to all the other utf8_* collations too; all will be using MySQL's utf8mb3 aka utf8 charset.
    – John Flatness
    Commented Jul 18, 2012 at 2:37
  • @JohnFlatness Thanks. Your comment is just what I was going to ask about next. I thought that UTF-16 used 2 bytes for Mandarin characters, for example, though? I'm looking at the documentation you gave me, hoping that it covers what 65,536 means. XD
    – Wolfpack'08
    Commented Jul 18, 2012 at 3:05
  • Chinese characters are within the Basic Multilingual Plane (the first 65,536 codepoints).
    – Ignacio Vazquez-Abrams
    Commented Jul 18, 2012 at 3:06
  • It seems like the latter 2 are the better options, space-wise.
    – Wolfpack'08
    Commented Jul 18, 2012 at 3:08
  • 2
    utf8 and utf8mb3 do not cover all CJK characters, some of which are 4-byte wide.
    – Meng Lu
    Commented Aug 27, 2014 at 0:09
0

utf8mb4 is the best.

utf8mb4 supports 4 bytes per character compared to utf8's 3 bytes per character, so it covers a wider range of uses without error.

With utf8mb4 you can support emojis, for example. If you try to insert an emoji in an unsupported character set you will get errors.

utf8mb4 is the more modern version of the 2 and will replace the older version eventually.

Improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.