utf8

[Tom Allison]

I'm not sure how/if this is done.
But I was wondering if anyone has looked into decoding all the  
charsets into utf8 for bayesian analysis.
octets is not readily visible to the user the way it's done today.