Re: [Emailcore] MDAs (was: Re: Delivered-To issues)

[Dave Crocker <[e-mail]>]

Ned,

On 1/8/2021 11:56 AM, Ned Freed wrote:
>> On 1/8/2021 7:30 AM, Ned Freed wrote:

>> Except that a mailing list operates /after/ delivery.  (And before 
>> re-posting.)
> 
> First of all, I was not talking about the list itself, but rather the
> tools that maintain it.

For reference, I still do not see how mailing lists are relevant to a 
discussion about transport rules.  But see my next comment.


> There's also the case where the list does similar comparisons in
> order to implement restrictions like "you have to be on the list to
> post" - and these often implement some understanding of subaddress
> semantics - and yes, those do operate after delivery, but this is all
> beside the point. The fact is that mailing lists and mailing list
> managers routinely violate the rule against interpretation and
> application of semantics only by "the host specified in the domain
> part of the address".

We have a fundamental disconnect, here, which well might mean we do NOT 
have a fundamental disagreement.  (And the end of your note seems to 
have anticipated or discerned the former, if not the latter.)

It might be that I needed to be clearer:  My focus, about the opacity of 
the left-hand side, pertains only to classic, MHS-level -- MTA -- 
behavior.  NOT with possible, idiosyncratic behaviors at the MUA level, 
which would include MLMs.

I can imagine a view that it should also include MUA level, as well as a 
view that it should not.  But I'll suggest that it is an entirely 
unrelated discussion, which isn't needed for the current work.  (Neither 
5321bis nor Deliver-To:)



>> I specify the address [e-mail].  My message travels to
>> an ietf.org email server and it resolves emailcore, handing the
>> message over to the mailing list system.(*)
> 
>> The message has been delivered to the address I specified.
> 
>> What the MLM does after that is secondary, for the purposes of this
>>  thread.
> 
> That might have been reasonable had you not justified the rule by
> making the point that "Having the Local-part be globally opaque is an
> important feature of Internet Mail." But you did.

Would "opaque to the Mail Handling Service" be acceptable, since that's 
the scope I'd intended?


> And that is in fact what justifies having the rule, not the far
> lesser claim that the pure transport infrastructure alone should not
> interrpet local parts. That rule by itself is almost worthless, since
> then all I need to do to justify whatever mangling I want to do of
> other people's address is say, "But I'm an MDA!"

I've read the above a few times, and I'm probably missing the point. 
Saying that the transport service treats the LHS as opaque, until 
reaching a place under the administrative control for the RHS, doesn't 
seem worthless to me.  Lesser, yes, but far from worthless.

As for the local domain's LHS namespace, it doesn't strike me as taking 
a liberty or all that odd to say it can interpret strings for its own 
namespace however it wants.  As in: of course.

As for declaring the MDA role whimsically, it seems to me not a small 
thing to note that that declaration means the message is no longer in 
the MHS and has moved into user-space.

What am I misunderstanding?

d/


-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net