John R Levine wrote:
I think we have confirmed that none of us really know what we'd do with
multiple signatures. Is that a problem, or should we just say that we'll
try to support them with unspecified semantics and hope they turn out to
be useful?
The one thing we know for sure at this point is that
more -- and reliable -- input for spam filters is better
than less. So from that standpoint alone, unspecified
semantics are preferable to removing information. I
expect that there may be some use cases where we
can catalog what the correct behavior is, but I don't
think we need go much farther than that.
I'm reminded about much of the dialog in the SIP WG:
there's a huge propensity from the, shall we say,
Old Bell World, to try to conflate service and
mechanism. We need to provide mechanisms that services
can be built upon. We can not and should not define the
service itself.
Mike
_______________________________________________
ietf-dkim mailing list
http://dkim.org