> -----Original Message-----
> From: John Levine [mailto:[email protected]]
> Sent: Wednesday, October 12, 2011 11:15 PM
> To: [email protected]
> Cc: Murray S. Kucherawy
> Subject: Re: [marf] New Version Notification - 
> draft-ietf-marf-authfailure-report-03.txt
> 
> Hmmn.  We already allow multiple groups each consisting of an auth-res
> header each followed by some other stuff.  I suppose we could define
> subgroups consisting of auth-failure, then dkim-domain, then other
> stuff.  But I worry that people will get it wrong.  Presumably an a-r
> could report multiple failures but you only send a report for
> the one that is likely to be of interest to the report target, so
> we need to be sure that the target can tell which failure all the
> subgroup stuff refers to.

The other idea I had borrows from a MIME extension:

Authentication-Results*0: ...
DKIM-Domain*0: ...
DKIM-Selector*0: ...

Authentication-Results*1: ...
DKIM-Domain*1: ...
DKIM-Selector*1: ...

Not pretty, but it would work.

I'm a little worried about the "send one report per authentication failure" 
because if I send a message with twenty bogus signatures bearing your domain 
name, that's an amplification attack.

> I see that the ABNF in section 4 of the draft doesn't update the
> feedback-report ABNF in section 3.5 of RFC 5965.  It better do that
> or there's no place in an ARF report where the new lines can occur.

It's covered by "ext-field" in Section 3.5 of RFC5965, isn't it?
_______________________________________________
marf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/marf

Reply via email to