[email protected] since the protocol has been adopted there...

On Sat, Jun 25, 2016 at 11:27 PM, Murray S. Kucherawy <[email protected]>
wrote:

> On Sun, Jun 19, 2016 at 1:34 AM, Brandon Long <[email protected]> wrote:
>
>> I recently wrote an authres parser and it seemed fairly easy to support
>> both authres and aar, but I also wrote the parser with that expectation,
>> not sure how easy/hard it would be to change that now.
>>
>>
> The authres syntax is fairly irritating, an over abundance of allowance
>> for CFWS and then using actual WS as a delimiter... that and the /version
>> stuff made it hard to take short cuts.
>>
>
> AR's CFWS use and use of WS as a delimiter is pretty much the same as most
> other things in email as I recall, with the notable exception of DKIM.
>
> And, I'd think you can recycle it, you just pass it the part after the
>> first semi-colon assuming the first stanza consists of i=N.
>>
>
> The section defining AAR doesn't actually say that it's "i=N;" (with the
> semi-colon).  It should.  Note though that the rest of that header field is
> not pure tag=value syntax.
>
> So here's a bit of hand grenade: You could separate the AAR parts from the
> AR parts by doing something like:
>
> ARC-Authentication-Results: i=1;
> ar=<base64-or-qp-encoded-copy-of-AR-you-want-to-preserve>
>
> That makes AAR into a purely tag=value syntax.  If there's no need for any
> part of ARC to care what the preserved thing is, this seems mighty clean.
>
> -MSK
>

I'm OK with this sort of strategy. Presumably the ar tag value would be
canonicalized before encoding?

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

Reply via email to