--On 10 January 2006 15:14:22 +0000 Tony Finch <dot@???> wrote:
> On Tue, 10 Jan 2006, Ian Eiloart wrote:
>>
>> I'd like to be able to fakereject a random sample of spam, in order that
>> I can examine the kind of stuff that we're rejecting. For example, I'd
>> like to see a random sample of about 0.1% of the spam that we're
>> rejecting from DNSBLists.
>
> I've used Exim's hashing functions to get pseudo-random samples of
> incoming email. See
> http://www-uxsup.csx.cam.ac.uk/~fanf2/hermes/doc/talks/2005-02-eximconf/p
> aper.html#5.2.%20Scanner%20testing
>
> Tony.
Hmm, that's cool. I don't want my chance of selecting a message to depend
on its content, though. I guess I could hash something like
$tod_log$message_headers rather than the body.
--
Ian Eiloart
Servers Team
Sussex University ITS