I read a draft for ballot[1] about changes to CABForum BR domain
validation rules, and noticed that it drops the "required website
content" bit, only retaining the Random Value and Request Token
bits. Now, request token parts are irrelevant: HTTP-01 clearly does
not qualify as request token.

This leaves Random Value. There there is also a requirement that the
Random Value does not appear in the request. Now, obviously HTTP-01
does put the Random Value into request. So it seemingly would not
comply.

Having to drop HTTP validation without replacement would be very
bad to say the least. Estimated number of domains using it is in
millions, and for vast majority, there is no viable replacement
(besides disabling HTTPS, perhaps using commercial cert to ride
HSTS timeout)..


Maybe the WG could define HTTP-02 that doesn't send the same
random value back? It would also solve the known issue that one
can mint responses without knowing about the challenge
(TLS-SNI-01 had a similar issue).



As to why they want to drop the "required website content" bit,
historically, many CAs have invented many quite bad validation
methods, some based on methods from Baseline Requirements.


[1] AFAIK, not yet in pre-voting discussion, but has the two
endorsers needed to start official pre-voting discussion.


-Ilari

_______________________________________________
Acme mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/acme

Reply via email to