I think the Tree Walk change presents a greater problem for v=1. The question is how often we have validation tests that match Elizabeth's example: - the examined domain name is two or more levels below the organizational domain, - the examined domain does not have a DMARC policy, - an intermediate domain has a DMARC policy, and - the organizational domain has a DMARC policy.
This may be rare, and a subset of that number may produce the same results with either policy used. The most likely error will be that the intermediate DMARC policy prevents the examined domain from testing as aligned with the organizational domain. DF On Fri, Feb 24, 2023 at 6:43 AM Alessandro Vesely <[email protected]> wrote: > On Thu 23/Feb/2023 21:17:09 +0100 Scott Kitterman wrote: > > > > If the consensus is that dropping pct requires a version bump, then I > think the correct solution is to not bump the version and add pct back to > the specification. > > > I think the consensus is to not bump version. > > Yet, it's true that the new I-D breaks pct=. It is still mentioned in > Appendix > 7, but it's not specified. I wish for a more gradual change. The tag > should > be handled by verifiers but not set in new DMARC records. > > When I'll upgrade my DMARC software, I'll add handling of t= but won't > remove > that of pct=, not until it disappears from DNSes. > > The inaccuracy of pct= is documented here: > https://en.wikipedia.org/wiki/Bernoulli_sampling > > > Best > Ale > -- > > > > _______________________________________________ > dmarc mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/dmarc >
_______________________________________________ dmarc mailing list [email protected] https://www.ietf.org/mailman/listinfo/dmarc
