On Thu, 5 Feb 2026, Yaron Sheffer wrote:

[ speaking very much as only an individual, not as AD ]

I cannot respond for the working group of course, but as far as I can tell the 
community is committed to migrating TLS to a PQC world, and this draft 
addresses one
potential problem (rollback attacks) with this migration.

That's not really answering Viktor. For reference, EKR and Richard
Barnes fiercly opposed any kind of pinning in browsers, that had
the exact same property: prevent a downgrade attack.

The concern at the time was that pinning would cause more damage to
operations than it would protect. Seeing that we are far from actual
quantum computers, it seems the same balancing act applies here.

(for background of those who weren't around. This escalated to the
point I could not even continue my presentation on the stage at that
TLS WG meeting for being continuously interrupted and over talked that
I had to step off the stage. The TLS WG was also very unhappy with
this and no longer wanted it to take up time and unadopted the document
not for technical reasons, but for not wanting to be involved reasons.
The draft with pinning against downgrade solution then had to go to ISE
to get published as RFC9102)

I think it is a _very_ fair question now to ask why having pinning as
a feature is suddenly acceptable. Because now it does appear that the
original pushback was not about pinning, but about supporting an
alternative trust model to the CAB/Forum (webpki) using DNSSEC(DANE).

(And if pinning is okay now, whether the previous opponents of pinning
will volunteer to bring the DANE extension via the TLS WG to Standards
Track RFC 😜 )

Paul


Thanks,
      Yaron


From: Viktor Dukhovni <[email protected]>
Date: Wednesday, 4 February 2026 at 11:44
To: [email protected] <[email protected]>
Subject: [TLS] Re: PQC Continuity draft (DANE chain déjà vu)

On Mon, Feb 02, 2026 at 04:05:11PM +0000, Kampanakis, Panos wrote:

> [1] https://datatracker.ietf.org/doc/draft-sheffer-tls-pqc-continuity/

The proposed "validity_period" component looks rather similar to what
was proposed for the DANE chain extension some years back, which
encountered fierce pushback from EKR and others and ultimately derailed
that RFC.

What's different this time?  Why would it be OK for a server to commit
to ongoing support for PQC certs when it wasn't OK for it to commit to
ongoing support for DANE chains?

--
    Viktor.  🇺🇦 Слава Україні!

_______________________________________________
TLS mailing list -- [email protected]
To unsubscribe send an email to [email protected]



_______________________________________________
TLS mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to