Re: [Bimi] BIMI/DMARC & PSL vs Tree-walk

[John Levine <[e-mail]>]

It appears that Ken O\'Driscoll <[e-mail]> said:
>The DMARC WG moved away from using the PSL for very valid reasons that
>probably don't need to be rehashed here. Those reasons are equally valid
>for BIMI. I think that BIMI should follow suit and use whatever mechanism
>DMARC uses to determine the organisational domain.

It seems hopelessly confusing to do anything else.

>Also, why do you think the current spec (dmarcbis-27) would give a
>different answer for the organisational domain than a PSL query?

We spent an absurd amount of time arguing about the tree walk, due
to a few people who obsessed about the handful of domains in the PSL.

In the vast majority of cases, the tree walk and the PSL give the same
result. We found a few cases where they didn't -- a domain that sends
mail but doesn't have a DMARC record, an org domain with a DMARC
record, and another domain between those two with a DMARC record with
a different policy. When we looked at them it was not at all clear
which result the domains expected.

How about telling people that if they want to use BIMI, it would be
a really good idea to publish DMARC records for the domains they use
to send mail?  Then the problem goes away.

R's,
John