RFC: mime boundary tokens

[Jake Di Toro]

On Mon, Jun 02, 2003 at 04:44:19PM -0400, David Relson wrote:
> Greetings,
> 
> This is a bogofilter RFC - Request For Comment.
> 
> According to rfc2046, mime boundary tokens can be 69 
> characters.  Bogofilter is aware of the limit and wont' accept longer 
> strings.
> 
> Some spammers are using longer strings.
> 
> Should bogofilter discard it's length check?
> 

My quick answer is yes.  

I've been a little boggled and how strictly bogofilter is adhering to
standards.  Spammers will do thier best to confuse filters as best as
they can.  If breaking standards in a way that the "common" mail
client will accept get's them past the filter, they will do it.  I
think the answer to a question like this should ALWAYS be yes.

The better question would be does bogofilter want to use this as a
scoring method.  Do we want to be able to add meta tokens like
"Mta:BoundToLong" with a count in the appropiate db's.  If it is only
happeing in spams than the score for this "token" will quickly
approach .99999 and will be a decent indacator.

-- 
Till Later, Jake <karrde+bogofilter at viluppo.net>
-----------------------------------------------
Direct replys are likley to be flagged as spam.
Drop the +addy if you need to reply direct.