> Note that in Python 2.x you don't have such issues because
> there, most tools for text processing will happily work on
> any sort of buffer, so you don't need a string sub-type
> in order to implement e.g. references into another string
> (the buffer type will allow you to do this easily).

The new buffer API has a provision for type flags, although none of them
had a chance to be implemented by the original author before he ran
away...
There could be a type flag for unicode characters and then its support
could be implemented in the memoryview object.