> On Mar 29, 2016, at 8:17 AM, Laura Atkins <la...@wordtothewise.com> wrote:
> 
> 
>> On Mar 28, 2016, at 8:20 PM, Vick Khera <vi...@khera.org> wrote:
>> 
>> On Wed, Mar 23, 2016 at 11:15 AM, Steve Atkins <st...@blighty.com> wrote:
>> 
>> > On Mar 22, 2016, at 9:35 PM, <frnk...@iname.com> <frnk...@iname.com> wrote:
>> >
>> > Are you taking that approach because the workaround is less than ideal?  
>> > Otherwise the current “workaround” could be the new standard.
>> 
>> The workaround is terrible and breaks basic email functionality.
>> 
>> What's so terrible about setting the visible From address to something you 
>> control, and setting reply-to to the original address? I thought that was 
>> the acceptable workaround, at least as discussed at sessions at M3AAWG.
> 
> Filters. - Hard to selectively filter particular mailing list users. 
> 
> Searches - nearly impossible to search for mail from a particular user. 
> 
> Reply-To - many lists want to set reply-to: list rather than reply to author. 

Also, email headers have well-defined meanings.

The From: header specifies the author of the message, the email address of the 
person responsible for writing the message.
The Sender: header specifies the email address of the agent responsible for 
sending the message, when that's different to the author.
The Reply-To: header specifies the email address to which replies should be 
sent, when that's different to the author.

They have different meanings, and they're all useful, and used.

The problem with DMARC is that it insists that the From: field contain an email 
address in a domain from which it was sent. If that's not naturally the case - 
mailing lists and ESPs are two examples of that - then you cannot comply with 
both RFC 5322 originator field semantics and DMARC requirements.

All the hacks that mailing list operators have put in place to allow people to 
use their mailing lists contrary to the DMARC-published requirements of the 
domain owner are based on violating the RFC 5322 semantics.

You need to put an email address in the From: field that's in the same domain 
as the mailing list manager - the email address that should be in the Sender: 
field has to go in the From: field. That destroys any record of who the author 
of the email is. That breaks quite a lot of functionality - filtering and 
searching, as laura mentioned, but even more obviously any ability to easily 
reply to just the author.

To work around that last problem a list operator may put the real authors 
address in the reply-to field. That destroys any explicit reply-to that the 
author may have set. That in turn breaks other expectations. If a user wants to 
subscribe to a mailing list with one address, but wants individual replies to 
go to another one they can say "From: li...@blighty.com; Reply-To: 
st...@blighty.com" - this sort of DMARC mitigation breaks that.

Also, many lists already set the Reply-To to the list submission address - so 
they can't put the author address in that field, leaving them with no choice 
other than to change (long established) expectations of how the mailing list 
works, or to have the author email address appear nowhere in the mail where it 
can be used mechanically by any variant of a reply command.

All of these break user expectations for how email should work - all users, not 
just those at domains that are publishing DMARC p=reject records. Mailing list 
operators can make operational changes to mitigate the inconvenience for 
different classes of user, but they can't fix it entirely.

There is *no way* to fix this without fundamental changes to either RFC 5322 or 
DMARC as they are inherently incompatible.

ARC will make those fundamental changes to DMARC, but only for one specific use 
case - traditional email discussion lists. It'll be an improvement (though not 
a total fix) for that case, but it makes no improvement for other use cases (of 
which ESPs catering to smaller users are one major one, but there are others).

Cheers,
  Steve


_______________________________________________
mailop mailing list
mailop@mailop.org
https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop

Reply via email to