On 22/01/2026 22:52, Wes Hardaker wrote:
"John Levine" <[email protected]> writes:
I'd look at RFC 9718, about publising the DNSSEC root keys since I'd expect it
to be published at roughly the same place. It might as well use a similar
method. The key file is XML rather than JSON for historical reasons, and there
is a detached signature which it appears nobody uses in favor of trusting the
https certificate when you download it from data.iana.org.
I think it may be critical to have a signature which is separate from
the HTTPS cert because you want IANA to be the ultimate authority over
the contents with zero dependency on another agent. Our current WebPKI
doesn't really protect against malicious parents (or even malicious
aunts and uncles except checking after the fact whether or not the cert
you used was invalidly issued by the wrong authority).
But, it certainly could be that the average implementation would never
check that more decentralized signature in favor of just trusting their
TLS stack. But the ability to trust an IANA controlled key itself is
probably critical (IMHO) for absolute verification.
Unbound (unbound-anchor) uses the detached signature as described in RFC
9718, Section 3.2.
See also some documentation:
- https://unbound.docs.nlnetlabs.nl/en/latest/manpages/unbound-anchor.html
- https://nlnetlabs.nl/documentation/unbound/howto-anchor/
Best,
-- Benno
--
Benno J. Overeinder
NLnet Labs
https://www.nlnetlabs.nl/
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]
