On 13/Feb/12 07:20, Scott Kitterman wrote:
> 
> As I read RFC 5965 2.d, either the message body or rfc822-headers must be 
> included as the third part of the message.

Yup.

> For SPF auth-failure reports, I'm not sure what, if anything, it makes to 
> include in this third MIME part of the message.  Since it's during the SMTP 
> transaction, there are SMTP exchanges of identities, but no headers.

AFAICS, the third part is needed to let the sender understand what
message failed to authenticate.  For example, if a user of theirs sent
a message via an unauthorized server, the envelope is possibly not
enough to take action (e.g. tell the user to not do that.)

> Since Mail From is already included as an explicit field in the second MIME 
> part, I'm not sure what, if anything should be required for the last part.  
> In 
> the case of the work I'm doing, I'm using the Postfix policy delegation 
> interface (see http://www.postfix.org/SMTPD_POLICY_README.html if you're 
> interested) and I don't have access to anything in the body of the message 
> (nor is there a reason for me to have it).

Could it be worked around designing a two-tier process?  That is,

1) generate non-ARF data (essentially, ARF w/o 3rd part), and then
2) finalize, possibly redact, and send.

The outcome of the first tier can still be useful for debugging or
internal auditing.  If the message gets accepted, the second tier has
to be able to retrieve the file.

> Is this MUST in 5965 being ignored or am mis-interpreting something?

2.d implies that only accepted or rejected-after-data messages can
actually be reported, while SPF rejections occur before data.  That
point could be overridden in spf-reporting or, if it's not too late,
in authailure-report.  A use case is for domains that need to collect
as many SPF *fail* reports as possible, e.g. to gather evidence on
which third parties habitually infringe on their registered name.

I'd be favorable to adding such possibility.
_______________________________________________
marf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/marf

Reply via email to