Comment UTF-8 (Score 4, Insightful) 165
The answer is UTF-8. It's pretty much going to be the de-facto character set now. It has backwards compatibility with ASCII, and can easily be extended in the future to support possible U+200000 - U+7FFFFFFF codepoints, as the original UTF-8 specification used to include that anyway.
Any important point is to not mess things up and end up with CESU-8 like MySQL did. There are completely valid 4-byte UTF-8 characters, so don't think of it as some special alternate UTF-8 by artificially capping UTF-8 at a max of 3 bytes per character.