[Emailcore] Re: [Last-Call] Your (Roman's) DISCUSS on draft-ietf-emailcore-as and status of the document

[Viktor Dukhovni <[e-mail]>]

On Fri, Sep 25, 2026 at 08:21:52PM -1000, Richard Barnes wrote:

> > ... there is no mandate ...  The A/S merely records that a generally
> > interoperable system MUST be able to allow them.
> 
> That's a mandate!  It says that the (admittedly optimistic) secure-only
> implementation is not a valid behavior.  If it were a simple statement of a
> sad fact, it would not involve normative language.

A "mandate" that requires "to be able to", with actual behaviour then
subject to local policy, is a rather weak mandate.  An implementation
can ignore the mandate, and not be broadly interoperable, or can provide
mechanisms for the operator to make the same choice.  The cost of making
rejection be conditional on a configuration setting rather than set in
stone for the

    - MAIL
    - RCPT
    - DATA and BDAT

commands when TLS has not yet been engaged is entirely trivial.  There's
not even a requirement for a specific default behaviour.  A conformant
MTA could default to requiring TLS to accept inbound mail, with the
operator needing to choose a non-default setting to change that.  That's
still conforms with "able to".

What does make the IETF look silly is not the content of the A/S, but
that its process sadly involves endless futile debates over minutia,
often to the detriment of the larger goals of those holding firm.  The
IETF certainly does not look good in light of the acrimony over
non-hybrid PQ algorithms.

Whose intrasigence is at fault is in the eye of the beholder, but we
collectively lack the means to effectively reach practical compromises.
I've foundnd myself repeatedly at odds with some of the same "opponents"
on previous documents, it is almost a given that any time I speak up for
opportunistic security the usual suspects will be found on the other
side.

Adherence to the A/S promotes use of TLS in SMTP transport, the floor is
not lowered, it simply does not change, but the ceiling is raised by
requiring conformant systems to support TLS, which may lead to some of
the long tail deployments finally adding support.  Standing in the way of
this does not advance the cause of email transport security.

-- 
    Viktor.  🇺🇦 Слава Україні!