On 21/Oct/11 17:09, Hilda Fontana wrote:
> 
> I'd really like to get a final consensus of the changes for the next draft.
> So far I have the minor changes, including the updates needed to the
> example.  

Is it ok for the example to start at the first column?  Usually lines
start with three spaces...

> Is it fair to say we have consensus on the one report per failure?  

There are different cases, e.g:

*straight case*
The sending domain signs twice, e.g. to experiment different
parameters, but it uses the same selector.  Hence the r= address is
the same for both failures.  If both signatures fail, the report
generator can tell how many messages it is going to send.

*possibly unrelated*
The sending domain defines SPF, DKIM, and ADSP.  If some of them fail,
the report generator would have to compare the relevant email
addresses in order to tell if reports are destined to the same operator.

*modular case*
The receiving domain performs multiple checks, but they are all
handled by different software modules, and thus result in multiple A-R
fields.  Possibly, generators should be configured to send reports to
an internal mailbox where they are reassembled, modified, or counted.

Such report-tweaking internal mailbox can also be used for setting
fields like Delivery-Result.  For example, the MX server that carries
out the checks can be configured to send the reports to the MDA that
decides about delivery; the MDA then forwards the report to the final
destination, which is still written in the included DNS records.

> What is the final verdict on the DKIM-Canonicalized-Body?  How should it be
> phrased in the doc?

I'm happy with the text Murray proposed.
_______________________________________________
marf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/marf

Reply via email to