I can't answer for Dan regarding implementation issues, but from a (computer) language point of view, consistency is better than correctness on this issue, because there is no single definition of "correct" until you specify what you mean by "correct". So at the first three Unicode support levels in Perl 6 (bytes, codepoints, and graphemes), sorting is by default always "Unicodabetical" regardless of the actual encoding of the string. You may, of course, always be more explicit in your sort command. Alternately, you may go to the fourth level, "letters", by specifying a particular language for the current lexical scope, in which case the default sort order for that language applies everywhere in that lexical scope. Arguably this is getting up into the library layers rather than Parrot internals, and I think Dan has it right to concentrate on the first three levels with pure Unicode semantics (whatever those are this week :-).
Larry
