I've been thinking about compression and Olivier's stunning suggestion
to switching the engine to all utf-8. If we were to beef up compression
to handle multibyte sequences, it would simultaneously handle multibyte
characters and improve ascii compression. A dumbed down version of LZW
(patent expired in June, 2003) or some other adaptive compression scheme
might be very interesting.

--

Jim Starkey
Netfrastructure, Inc.
978 526-1376