[Emailcore] Re: [Last-Call] draft-ietf-emailcore-as some mail stats

["Martin J. Dürst" <[e-mail]>]

I fully agree with Victor. As an applicability statement, we have to 
make sure we are in sync with reality, and describe all relevant aspects 
of this reality. This is most important for minor or new 
implementations, not for the big ones that already have the necessary 
information and connections. RFCs are often criticized as requiring to 
"read between the lines"; we want to avoid that wherever possible.

Reality clearly is that some emails get sent as plain text, because the 
sender doesn't have the capability to use TLS. Saying that accepting 
plain text emails is no longer needed would mean to define a new 
protocol, let's call it TLS-only-SMTP. But this is not what the WG was 
chartered to do, not for 5321bis and even less for the AS.

Regards,   Martin.

On 2026-10-02 11:55, Viktor Dukhovni wrote:
> On Thu, Oct 01, 2026 at 06:59:46PM -0700, Rob Sayre wrote:
> 
>> What's the problem with deleting that one last paragraph in Section 6.5?
> 
> Deleting it would skew the balance of the applicability statement.
> Section 6.5 is doing two things.
> 
>      - It says that TLS support is required, and that that TLS MUST
>        be used when offered.
> 
>      - It says that nevertheless continued support for cleartext is
>        also required, though ...
> 
>      - Local policy ultimately determines what any sending or receiving
>        system will choose.
> 
> Nobody should be left with the impression that in the context of the A/S
> (real-world requirements for an interoperable inter-ADMD SMTP relay)
> either TLS or cleartext support is optional to implement.
> 
> If you want to delete the required cleartext support, you'll need to
> delete all of section 6.5, which on balance is a significant step
> forward in terms of requiring TLS (when mutually supported).  But
> if discarding the bathwater is more important than keeping the baby,
> by all means, delete 6.5.  Postfix TLS support will evolve sensibly,
> with or without any related language in the A/S.
>