On 2011/08/11 18:39, Karsten Bräckelmann wrote:
On Thu, 2011-08-11 at 14:40 -0700, jdow wrote:
>>> KB_DATE_CONTAINS_TAB is misfiring with great consistency on the LKML.
>>> This is not good.
It's a known issue. There is at least one ISP and LKML rewriting *MUA*
generated headers like that. Frankly, *this* is not good. If you're an
MTA, add your own headers, but bloody don't mess with existing headers.
None of your business.
As an aside, you'd better not filter your legit mailing lists through
SA. Even less so the seriously above negligible bandwitdh that LKML is.
This rule once was written to and highly efficient at catching some
sneaky German stuff (at least German only in my in-stream back those
days). Probably a candidate for pulling -- getting your deserved
vacation, err, retirement. ;)
I have a suitable rule that "amplifies" the Bayes scores for mailing lists.
Some spam still makes it through the lists. A double check is a good thing.
If Bayes does not say it's really spammy, I knock off some score. It Bayes
says it's really spammy I add some score. I am about ready to pull LKML from
that purgatory. It's doing a good anti-spam. Google Groups and Yahoo Groups
are not doing very good anti-spam efforts. With the magnitude of the task I
can see why, so I partially filter them.
This new rule broke my long chain of very good results.
(Let's hear it for meta rules!)
>> what MLM is doing that?
> I am not sure if that is the MLM's fault or the user's tool's fault:
Neither.
> And the MLM does not identify itself. It's typical LKML idiot syncratic.
> (sic)
>
> Ah, it identifies itself in the message in a helper line:
> majord...@vger.kernel.org
Without proof at hand right now, I believe it is *not* majordomos fault,
but some MTA configuration that actually rewrites MTA submitted headers.
Enh - something on a legitimate source of mail fiddles the header that way.
Ideally one would convince them not to. In practice, this is the LKML we're
talking about. {^_-}
{^_^}
