On Fri, 09 Mar 2007 17:51:36 GMT, Tony Finch said:
On Fri, 9 Mar 2007, Valdis(_dot_)Kletnieks(_at_)vt(_dot_)edu wrote:
What about SpamAssassin filters, and Sendmail Milters that do filtering?
They are not doing the same job as Sieve.
Exactly. My point is that there's a lot more filtering than just SIEVE being
done, and at a lot of different points along the path the e-mail traverses.
As such, any attempt to equate "filtering" with "SIEVE" is doomed at the
start.
There is no such attempt and we're not talking about filtering in general -
this is one specific type of filtering that we have managed to formalize
and standardize.
Again, I don't care if the diagram says "Sieve" or not. There are certainly
mechanisms that do not enjoy standards status (e.g., .maildelivery or procmail)
which do occupy more or less the same niche in the email ecosystem. But it does
need to be clear that we're not talking about all types of email filtering
here, so I do object to using a totally generic term like "filtering". If we
were talking about filtering in general it would have to appear as a sort of
cloud covering the entire diagram - and I see no point in doing that.
Things might be different if some attempts to formalize and standardize other
mechanisms that provide different and broader sorts of filtering had gone to
fruition, such as the OPES SMTP work. But they didn't. We have what we have,
and if that's a problem the way to solve it is to build a consensus around
additional standardization work. This document is supposed to be documenting
and explaining the formal architecture we have, not creating new formalisms on
the fly.
Ned