On Thursday, April 30, 2015 06:45:25 PM John Levine wrote:
> >I recall some earlier discussion about encoding information in From local
> >part, but if for no other reason, automatic addition of new email addresses
> >to contacts by some MUAs makes that highly problematic.
> 
> This trick also has patent problems.
> 
> >DNS query to message-id.yamfsidlocalpart._dmarc.domain.  The sender then
> >replies pass/fail/error.
> 
> I think you will find that his has impossible scaling problems,
> particularly if you care enough about security to use DNSSEC to deter
> the usual poisoning attacks.
> 
> There have been a variety of proposals over the years that publish
> per-message data out of band, so the recipient can check and see if
> the message is OK, and they all have the same scaling problem.
> 
> Besides, you can get the same effect by taking whatever would be in
> that DNS record and putting it in the message, with a crypto signature
> to prove that it's real.  The signature's validation key may come from
> the DNS, but that's OK since one validation key can be shared over
> many messages.  We've just reinvented DKIM, perhaps with an extra
> field or two to include the yaimfs stuff.  If you're worried about
> message mutations breaking the signature, don't sign the stuff that's
> likely to mutate.

Yeah.  If you want something signed, then it becomes equivalent to the fs= 
proposal.

To be effective, it would have to have something based on database queries, not 
a static zone, so I can imagine DNSSEC problems.  Is having a non-DNSSEC 
subdomain that has no real hosts in it problematic?  RBLs seem to scale 
reasonably well, although I realize per message is more of an issue than per 
sending IP.

Scott K

_______________________________________________
dmarc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmarc

Reply via email to