Daniel Tiefnig wrote: > Tony Finch wrote:
>> This seems to be more effective than you might expect, even without
>> any kind of fuzzy-matching in the checksum.
>
> Sounds reasonable at least for the anti-virus scan, I'm not sure whether
> I'd like that for SpamAssassin though, as headers are important there.
> And these mailcounting systems like DCC would not work with it of course.
> For anti-virus, I'd sure give it a try, if someone can't resist to
> implement it. :o)
But our content-scanner can extract individual MIME parts and hash
*them*. Maybe only do this if the size of each part is larger (or
smaller) than some threshold(s). This sounds quite useful indeed.
