That's kind of funny - Oracle has a similar issue. If you want industry-standard UTF-8, you have to specify "AL32UTF8" as your encoding. "UTF8" is kind of crazy - it's this monstrous abomination called CESU-8 (https://en.wikipedia.org/wiki/CESU-8), which isnt' UTF-8 at all - it's actually this weird "UTF-16 complete with surrogate pairs wrapped in a UTF-8 shell" thing.
That would be a serious problem if Oracle's idea of "UTF8" wasn't roundtrip compatible with UTF-8, but it is fortunately, so approximately no one should notice the difference.
