From: rewriting on newlib@sourceware.org

[Maciej W. Rozycki]

On Wed, 16 Jul 2025, Frank Ch. Eigler via Newlib wrote:

> > > Honestly, the newlib ML works like this for a long time, and I don't
> > > see why the onus is now falling to us applying patches from the list,
> > > just because some admins can't do it right.
> 
> Understood, it's just that email hygiene mandates seems to be steadily
> increasing.  Other admins don't seem to be doing anything wrong (e.g.,
> RFC-violating).  Rather, it is a sourceware mailman2-admin-side
> decision to allow transmission of imperfectly signed emails.  The main
> sourceware-side knobs available to handle that are the from-munge (or
> email-wrapping) options.

 If bouncing "imperfectly signed emails" is not an RFC violation, then IMO 
it's IETF rather than admins that's doing something wrong, though I guess 
one doesn't preclude the other.  If someone wants authenticated e-mail 
(and say PGP/GPG is not enough), then that ought to be a distinct protocol 
on its own rather than an SMTP abuse by taking the "embrace and extend" 
approach.

 I'd reject a postal system right away that requires one to authenticate 
oneself to drop a letter into a postbox.  I can imagine some regimes do 
have such a system in place.  Though perhaps it is not the best analogy, 
as it's actually as if only certain letterboxes rejected unauthenticated 
letters which have passed through the system already, which makes the case 
even more twisted.

 I think the key question is who gets unsubscribed when a relevant bounce 
happens.  If it's the person using such a crippled system for their e-mail 
delivery, then I'd consider it fine, and an incentive to move to a proper 
e-mail service provider.

 FWIW,

  Maciej