Re: [dmarc-ietf] Tree walk nits

[Scott Kitterman <[e-mail]>]

On June 22, 2022 3:07:41 PM UTC, John Levine <[e-mail]> wrote:
>It appears that Scott Kitterman  <[e-mail]> said:
>>
>>
>>On June 22, 2022 2:54:28 AM UTC, John Levine <[e-mail]> wrote:
>>>It appears that Scott Kitterman  <[e-mail]> said:
>>>>Not confusing is indeed the tricky part.  I think what's wanted is shortest that's longer than the longest PSD.
>>>
>>>How about this?
>>>
>>>b.a psd=y
>>>c.b.a NXDOMAIN
>>>d.c.b.a blah
>>>
>>>What's the org domain for e.d.c.b.a?  I think it's c.b.a.   Is the policy domain is d.c.b.a or b.a ?
>>
>>I think that because of the b.a record with the pad= tag that's correct for the org domain, although I'm not sure why anyone
>>would do that.  Without the PSD record it would be d.c.b.a.
>
>Imagine that b.a is something like co.uk, abd d.c.b.a is mail.crumpets.co.uk.  I don't think that's a great DMARC setup
>but it doesn't seem ridiculous.
>
>>For policy, it should be b.a, but I'm working from my phone right now and I'm too lazy to look up what the draft actually says
>>(vice what I think it says).
>
>I think we agree about what we want it to say, you stop the tree walk when you see psd=y/n, the policy domain is
>the last record you found,  org domain is the last record unless it's psd=y in which case it's the level below.
>
>I'll do a pull request so we can see if we agree on language.

That sounds right.  Thanks,

Scott K