[Dovecot] Sending email with IMAP instead of SMTP?

[Amelia A Lewis]

On 2005-08-16 13:43:46 -0400 Marc Perkel <marc at perkel.com> wrote:
> Andrew Hutchings wrote:
>> On Tue, 2005-08-16 at 10:30 -0700, Kenneth Porter wrote:
>> 
>> Would it really be all that useful though?  I can't see it making
>> dovecot 1.0 at least.  SMTP is standard for mail relaying for a very
>> good reason.
>> 
> This is all just theoretical. The idea isn't to eliminate SMTP. Servers 
> would 
> still talk SMTP to each other. The advantage is that when you configure a 
> client you would only have to configure the IMAP. It wouldn't require an 
> additional step for setting up outgoing email. And on the server end you 
> wouldn't have to set up SMTP authentication because the IMAP would already 
> do 
> that for you. IMAP would act as a transport to get the outgoing mail from 
> the 
> client to the SMTP server.

So, just a particular IMAP mailbox, then?  And what would happen if you 
copied a random received email into that box?  Say, one to the dovecot 
mailing list?

How do you do blind carbon copy?  Despite the pseudo-header displayed by 
mail clients (and typically saved as a Bcc: header in a sent-mail box, if 
your MUA is configured to store sent mails), this is not supplied to the 
outgoing SMTP server as a header.

Or, in other words, how much header munging does the IMAP-send-enabled MUA 
do, and how much is left to the IMAP-send-enabled imapd?  If imapd sets SMTP 
MAIL FROM:, does it also munge From:?  Sender:?  Resent-From:?  List-From:?  
X-Secret-Message-From:?

Not saying that these issues can't be resolved, just that they do exist, and 
consequently MUST be resolved; depending on how much complexity 
implementation requires of the server versus the MUA, you'll see different 
implementation results.

The two use cases suggested seem to be: reduce transmission time for 
sending/storing as sent-mail; reduce configuration complexity.  For 
reduction of configuration complexity, it will depend on support for the 
capability in clients commonly used by less technical users.  Reduction of 
transmission time alone seems marginal as a use case, given possible 
complexities and interoperability issues.

Amy!
-- 
Amelia A. Lewis                    amyzing {at} talsever.com
The less I seek my source for some definitive, the closer I am to fine.
                 -- Indigo Girls