As a participant, I fully disagree with the second paragraph of this. The justification for changing the mechanism is that in cases where the mechanisms differ, the tree walk produces results that are more likely to represent the intent of the sending side than consulting the PSL does. This has been borne out by cases we have actually seen, which says to me that the change is the right thing to do.
It is certainly possible that we will find cases where the PSL gives a better answer than the tree walk does. We can construct such cases in theory, but haven't found real-world ones yet, and that's telling. Moreover, if, as I think we have done, we have made the situation better overall, we have done the right thing, and we simply need to document what organizations can do with their DMARC policies to fix things if they fall into such a tiny corner case. Barry On Wed, May 3, 2023 at 6:46 AM Douglas Foster <dougfoster.emailstanda...@gmail.com> wrote: > > I have opened issue 113 to formally document my strong objections to the > current tree walk: > > Current DMARC policies are configured based on RFC7489 and the PSL, and > evaluators obtain results based on those implementation decisions. Domain > owners may have many reasons to want an alternative to the PSL: (1) The PSL > may contain errors that impact the domain owner's mail flow. (2) The PSL is > implemented in different iterations by different evaluators. (3) The RFC7489 > / PSL algorithm does not allow for partitioned alignment within an > organization. > > Nonetheless, an evaluator has no justification for implementing an algorithm > which produces different results unless the domain owner indicates that he > prefers usage of that different algorithm. This can be accomplished by > tagging his DMARC policies to indicate which of the four possible roles > applies to a particular policy: Org Top, Subdomain, Org Top-and-Bottom > (single label registry), and Org Bottom (bottom layer of a multiple-layer > private registry), and DMARCbis should define those tags The current > upward-walk proposal will cause damage by directing evaluators to apply an > undesired and often incorrect re-interpretation of domain owner intent and > associated alignment boundaries. > _______________________________________________ > dmarc mailing list > dmarc@ietf.org > https://www.ietf.org/mailman/listinfo/dmarc _______________________________________________ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
