Yea, also found that limitation of the regex library pretty annoying. But conceptually this has nothing to do with string slices being different from byte slices... It might encourage it though.
FWIW, I think getting the underlying string representation right (and relatedly how it interacts with byte buffers/slices) is really hard. So far, I think Perl 6 probably has the most well thought out structure with regard to that, but they took a decade and a half and lots of testing to come up with what they did. Then again, how it works in Perl 6 is also a function of how the language functions, so it wouldn't be a a direct mapping of concepts, but there are some really good ideas there.
