I'd prefer a "bayes_hash_tokens" option (default = turned on) over this that affected all storage methods:
* on: hash tokens * off: no hashing or truncation of tokens Simple and easy to understand. Probably would require some hooks in some storage methods to change the format of the tokens, maybe a per-DB format flag. Code would not need to (and probably should not) support either/both formats. -- Daniel Quinlan anti-spam (SpamAssassin), Linux, http://www.pathname.com/~quinlan/ and open source consulting
