replace-nonascii-characters stopped working?

David Relson relson at
Tue Mar 22 02:11:10 CET 2005

On Mon, 21 Mar 2005 19:05:31 -0600
Bill McClain wrote:

> On Mon, 21 Mar 2005 19:39:43 -0500
> David Relson <relson at> wrote:
> > The file looks like all the accented european characters.  The
> > relevant line is:
> > 
> >    Content-Type: text/plain; charset=Windows-1251
> > 
> > Running the 0.92.8 and 0.94.1 versions of bogolexer give the same
> > results.  The design purpose of replace-nonascii-characters was to
> > limit the tokens generated by asian spam. Since windows-1251 is a
> > 'normal' charset, what bogofilter's doing is appropriate.
> Ok, I believe it. It is a new type of spam for me. The first one arrived
> just a few days ago the same time I upgraded bogofilter. Or more
> exactly: 8-bit tokens first appeared in my wordlist then. I see that
> messages with that Content-Type have arrived in the past, but weren't
> registered because of thresh-update.
> Thanks for investigating this.
> -Bill

If it's some sort of marketing message, it's lack of readability makes
it an abysmal failure.

Glad to be of assistance.  'Tis good to have people paying close
attention to what's going on.  That's how bugs are found and is a
crucial step towards stomping them.

Bogofilter mailing list
Bogofilter at

More information about the Bogofilter mailing list