On Wed, Jul 26, 2017 at 3:27 AM, Keith Herron <kherron(a)wikimedia.org> wrote:
Hi Platonides,
Please see my comments inline below :)
What will be the behavior when a final user
presses reply to one of
those emails from the list?
This would result in multiple addresses in the Reply-to: header like this:
Reply-to: sender(a)example.com, list(a)lists.wikimedia.org
Pressing reply should then open a message with both addresses in the To:
field.
Should a bofh enable the reject option, will the
rejection message
properly explain that «they are sending an email to a public
mailing list
and their domain is configured to not allow that, and should they have any
issue to complain to their email provider»?
If desired dmarc_moderation_notice can bet set on a per-list basis to
explain why the message was rejected.
Quite sad, but it's actually what everybody
should be doing instead of
working around it by overwriting the author of the
message with the sender
and clobbering the Reply-to (cf. rfc 5322 section 3.6.2).
(and leaving everyone else with badly emails)
Does it at least leave a pattern that allows a faithful reconstruction?
This will not change the from addresses displayed in archives and unless
you have a static contact book entry for the list address the sender name
should appear in the thread as “First Last via list".
Hmm, I thought the mailman archive was also affected.
If by archives you mean the received emails in your MUA, keeping the name
(provided by the user!) is not "all the user cares about".
For instance, this thread started with a post by someone signign as Keith
Herron. I don't think I came accross that user before*
However, an email from «Keith Herron <kherron(a)wikimedia.xn--org>-yna is quite
distinct than an email from «Keith Herron via Announcements for mailing
list admins»
And of course, such munging breaks basic features like «search all emails
received from this user (and only those from this user)»
* Probably because you joined about two months ago ;)
https://wikimediafoundation.org/w/index.php?title=Template:Staff_and_contra…
This is probably a sidestep, though. By «Does it at least leave a pattern
that allows a faithful reconstruction?» I meant, «Is there an algorithmic
way to get back the unmunged message?» I see for instance that Google and
Yahoo groups leave a X-Original-From: header that could be used.
Could deliveries to the gmane archive be exempted?
I’m not sure about gmane, but in this case it might be more
straightforward to reject DMARC senders and avoid inconsistencies between
archives.
Well, rejecting posts from senders using a strict DMARC policy would of
course solve the munging issue.
Best regards