[antispam-f] Re: Recent Id: and Ref: Spams
- From: Frank de Bruijn <antispam@xxxxxxxxxx>
- To: antispam@xxxxxxxxxxxxx
- Date: Sat, 27 Jan 2007 08:02:11 +0100
In article <4eab71c9b5steve.pampling@xxxxxxxxxxxxx>,
Steven Pampling <steve.pampling@xxxxxxxxxxxxx> wrote:
> On 26 Jan, Frank de Bruijn <antispam@xxxxxxxxxx> wrote:
> [Snip]
> > The current implementation of this in AntiSpam 1.60 [1] using the NSQ
> > [2] module is that every response code from a block list zone has an
> > associated value, its 'weight'. All the weights are added together and
> > the end result can be tested at EOH (end of headers).
> > The (PBL) codes 127.0.0.10 and 127.0.0.11 coming from SpamHaus have a
> > weight of zero when NSQ is used with a 'deep parser' like AntiSpam.
> Why not simply pass out the specific loopback address numbers 127.0.0.x ?
That's what I did with the early versions...
> (Although I've come across mentions of use of 127.x.y.z where the x y and
> z have significance those seem to be people trying to extend a standard so
> the least significant octet seems to be the most important here.)
Exactly. While investigating providers I became less and less convinced
there really IS a standard. The only address they all seem to use is
127.0.0.2 and even that can have a slightly different meaning.
> With those values the end user can decide which to use and which not.
The address is also returned by the relevant SWI, so if an end user
wants to use it he can.
> [Snip]
> > > Any chance of the working combination (module and appropriate AS) to
> > > test. If I can use this to produce a headered list into my spam test
> > > bin without a false positive the wife for one will be happy when I
> > > change it to delete.
> > Not just yet. I'm still adding bells and whistles to the module and
> > although the bells are ok, some of the whistles seem to be slightly out
> > of tune since the last change. I'll try to get that sorted out this
> > weekend.
> No pressure, late Saturday will do :-)
Hehe. Next Saturday is more likely.
Regards,
Frank
Other related posts:
