I fully support Olivier's point of view about UTF-8 being the optimal
way to store/transmit text data.

For people worried about being able to store plain, old bytes I would
recommend that they take a look at the Oracle hack for this:
character lengths can be declared in CHAR units (default) or in BYTE
units. The latter gives a guarantee about the maximum size needed for
buffers, the former a guarantee about the number of characters that
can be stored.

Paul