The way I expect it to work, is that Unicode strings don't get encodings, but byte buffers and I/O handles do. If you give a Buf an encoding, you can treat it as a Str, as long as the encoding is consistent. Using a real Str may perform much better, because Perl 6 can encode it internally however it wishes, and can make many more assumptions (for optimization purposes).

Giving a Unicode string an "encoding" property is the wrong way around, but it can take quite some time and experience before one realises that.