http://bugzilla.spamassassin.org/show_bug.cgi?id=2266





------- Additional Comments From [EMAIL PROTECTED]  2004-04-27 22:10 -------
Subject: Re:  New features: Tokens in report, status of Bayesian classification.

On Tue, Apr 27, 2004 at 10:00:03PM -0700, [EMAIL PROTECTED] wrote:
> http://bugzilla.spamassassin.org/show_bug.cgi?id=2266
> 
> ------- Additional Comments From [EMAIL PROTECTED]  2004-04-27 22:00 -------
> I could think of two ways that it might make sense with hashed tokens.
> 
> One is if hashing tokens is a configuration option. Someone with a low enough
> volume of mail might prefer to keep the ability to look at the tokens.

Considered, but it complicates lots of things (setup, backup/restore,
dump, etc).  If we're gonna go some of the way, might as well go all
the way.

> The other is if there were a utility similar to sa-learn that creates a 
> database
> of tokens keyed by their hashes. Then if someone cared to find out what the
> original token is for a hash they could look it up.
> 

Ugh, I guess, but doesn't sounds pretty.


If folks really want to see which tokens were used to score the msg, I
suppose we could push the token hashing stuff down into the storage
layer.  I'll look into that.

Michael





------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to