Embed Notice

@dalias @ireneista @misty @bob So... I don't mean to nitpick, but, um.

Let's skip the whole thing that "Unicode characters" don't exist, and "faithfully represent" is dodgy.

What I meant was that the only way a POSIX-ish API *can* present those names as UTF-8 is if the file system stores not only the bytes in which the names are encoded, but also the encoding.

The other thing is that it may not be a good API to have UTF-8. (By the way, POSIX doesn't, not really.)