Request for comments

[Alex Broens]

On 7/20/2007 7:36 PM, Matt Kettler wrote:
> Alex Broens wrote:
>> On 7/20/2007 6:55 PM, Steven Andrews wrote:
>>> Why not?  I know specious argument, but this would work well so you
>>> could apply a penalty or a credit to a certain domain.
>>>
>>> Blackberry devices are just an example, they always trigger certain
>>> rules that push their scores up.  Are they going to change that fact?
>>> Nope.  Do I want to lower the value of those rules?  Nope.  They catch
>>> other traffic.  Do I want to whitelist blackberries entirely...no way.
>>> If I had a mechanism to punish or credit a certain domain, that would
>>> allow such a situation where I can keep rules intact but adjust the
>>> spamminess of a domain.
>> header BLACKBERY_PASSTHRU   Received =~
>> /smtp[0-9]{2}\.\w+\.\w+\.blackberry\.com\b/
>> score  BLACKBERY_PASSTHRU   -5.0
>>
> 
> Even better, use X-Spam-Relays-Untrusted. It's a fake header generated by SA
> that contains pre-parsed Received: headers. Its format is constant and isn't MTA
> specific. The first entry is the host delivering to your last trusted server.
> ie: if your trusted_networks isn't broken the last trusted server, making the
> machine dropping mail off at your network the first untrusted.
> 
> 
> This little trick starts at the begining of the text (hence the first ^) and
> scans ahead for blackberry.com, but will sto if it encounters a ] (which would
> be the closing bracket of the end of the first entry)
> 
> header BLACKBERY_PASSTHRU X-Spam-Relays-Untrusted =~
> /^[^\]]+rdns=smtp[0-9]{2}\.\w+\.\w+\.blackberry\.com\n/

DOH!

used that for other stuff.. dunno why I didn't think of it for the 
"blueberrries"

thanks for the hint

Alex