RE: CSV (Crocker's draft) good! (evaluation, big suggestion)

> On Mon, May 03, 2004 at 08:48:33AM -0700, Hallam-Baker, Phillip wrote:
> | 
> | Mailing lists are very rarely members of a forwarding relationship.
> | If they are it is because the list has moved and there is a very 
> | clear and explicit understanding of the forwarding relationship.
> | 
> | This being the case I don't see why a mailing list should not 
> | reject all mail that fails Caller-ID checks. The test to see if the
> | email is forwarded becomes very simple.
>
> Ok, that makes sense; it sounds like mailing list managers, when
> receiving, should always impose a directOnly attribute on lookups; is
> that right?  so the tested address is the "From:" no matter what other
> "Sender", "Resent-*", etc headers are present.
We might make certain exceptions:

If the Sender field is present, it validates AND the address is subscribed.
This seems useful to me because we quite often see mails that are forwarded
to lists on behalf of others who could not get their mail situation sorted
out.

I don't see why a postcard site should send to a mailing list though. I
think Meng is right here, mailing list sites should ignore Sender and 
require strict consistency on From. There might be some corner case 
exceptions but those are going to be sufficiently rare they can be 
whitelisted, things like a bulk mailer sending to a mailing list 
legitimately [Say Roving announce the launch of the new Harry Potter
film to a bunch of fan mailing lists on behalf of Time-Warner].


The way this should probably be set out in the spec is first we define 
a set of receivers which should require strict compliance (e.g. mailing
lists) then we define strict compliance. I guess it means the 
authorized forwarder list is the nil set.

                Phill