[Emailcore] Re: [Last-Call] Re: Re: Re: Re: Re: Re: Re: draft-ietf-emailcore-as-28 ietf last call Secdir review

[Rob Sayre <[e-mail]>]

On Thu, May 7, 2026 at 3:17 PM Barry Leiba <[e-mail]> wrote:

> > > We know from Richard Clayton's recent message that 16% of the hosts
> that talk to
> > > Yahoo's giant mail system don't do TLS,
> >
> > >  Why are we even having this argument?
> >
> > Because you don't understand that the large ones will want to accept
> cleartext email while narrower installations will not.
>
> But that's exactly the point the document is making:  It needs to be
> configurable to accommodate BOTH.


Ah. that might be the disagreement. It does not need to be configurable to
be a permissive service. That requirement is an attack surface.

Eliot has a point I agree with:
> We should not tell coders what is configuration and what is code.
https://mailarchive.ietf.org/arch/msg/emailcore/1s4_JWP359hyaRw51DauS9fGcGc/

I've hit this one myself in HTTP, I wish it just didn't have the cleartext
path:
https://mailarchive.ietf.org/arch/msg/emailcore/bxiDbosHvI_TWCsxAlWsARQ3BOk/

There, a non-conformant client exercised a codepath that shouldn't have
been there. That's all real code run by billions of people, and I found it.

And I went to the trouble of doing a Postfix implementation with no
cleartext configuration. [0]

All of this just to say we should delete the last paragraph of Section 6.5
and publish. I'm not asking for strict requirements to be added.

thanks,
Rob

[0]
https://github.com/sayrer/postfix/commit/4d12d04ea69f89e021c74d70a9af464411d0a27a#diff-48e549783fcfddf2329fcbe3549df5d60982698764a4234b330f4066719e6526R29