I know it's kinda late to make any changes to the 1.9 roadmap, so I'm 
not hoping to change anything, but there's this *tiny* thing I'm worried 
about with m17n: depending on the string's encoding, methods like 
String#index will return either byte position or character positions. 
Since a Fixnum cannot carry information about where it came from, isn't 
there the risk of opening a whole new *class* of bugs where a 
character-position integer is fed to a byte-encoded string method and 
vice-versa? Has this been encountered or has there been too little 
experience with m17n strings or am I just too paranoid?

Daniel

Disclaimer: ultimately I am a curmudgeon who believes m17n strings are a 
(too-)ambitious technical solution to a social problem ;-P