perhaps you could write up a one-paragraph summary of the result
Good thought.
The PSL is intended for web browsers, mostly to prevent "supercookies" and
secondarily to help browsers display domain names more intuitively. It is
basically a coincidence that it more or less works to find DMARC
organizational domains. It is manually maintained, chronically out of
date, and there are org domains it cannot describe at all, the ones that
are vanity TLDs like .BANANAREPUBLIC. I have great respect for the PSL
maintainers, several of whom I know personally, but they are quite clear
that they're just a handful of volunteers, and browsers are its main
target.
The goal of switching to a tree walk is to allow domains make their own
statements about what their org domains are and find those org domains
more reliably. Given that software changes slowly, we need to keep in
mind that there will be DMARC records published under the current rules,
and software interpreting them using the PSL for a long time, so a new
design should produce results no worse than what we get now. They need
not be identical, since the PSL can get things wrong, which is why I say
no worse. The tree walk MUST (in the English sense, not the RFC sense)
work wthout reference to the PSL in view of the limits described above,
and if we were to tell people to keep using the PSL, what's the incentive
to change anything?
I believe that the tree walk design that Scott and I are closing in on
will produce good results efficiently. The only change to the DMARC
record format is psd=u/y/n which will be needed in only an infinitesimal
fraction of DMARC records, the rest being unchaanged from what they are now.
Regards,
John Levine, [email protected], Taughannock Networks, Trumansburg NY
Please consider the environment before reading this e-mail. https://jl.ly
_______________________________________________
dmarc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmarc
