[Users] To queue, or not to queue was: Re: Error while sending....

[Pierre Fortin]

On Mon, 26 Jan 2015 00:35:13 -0800 Ian Zimmerman wrote:

>On Sun, 25 Jan 2015 13:56:36 -0500,
>Pierre Fortin <pf at pfortin.com> wrote:
>
>> Huh? Besides taking my comment out of context...  Whether I put a
>> letter in the mailbox, or leave it in my pocket to post later, is a
>> user (MUA) issue.  I know of no post office (MTA) that holds (queues)
>> mail -- except when waiting on authorization to deliver to the
>> destination.
>
>Huh squared :-)  Queuing is really the essence of MTAs, as SMTP is a
>store-and-forward protocol.  A MTA holds on to a message as long as it
>is unable to send it to the next hop, subject to a reasonable upper
>limit.  I'm sure you know this, so there must be a
>fundamental misunderstanding.

No misunderstanding: as in "except when waiting on authorization to
deliver to the destination" which recurses at each step. :)  

This list is about CM (MUA) and the subject (error while sending) is
within CM, so MTA-MTA issues are outside the scope of what I was talking
about.

>> The "error while sending" has nothing to do with MTAs because the mail
>> in question has yet to leave the MUA (CM) on its way to an MTA.
>
>It is a failure to complete a SMTP transaction with a remote MTA.  What
>I'm saying is that a local MTA, by its nature, is better equipped to
>deal with such failure, because that is what it does all day.

A local MTA, other than operating over a virtually error-free loopback
interface (127.0.0.1) operates no differently than a remote one; but I
was addressing the "error while sending" within CM (devs: feel free to
correct me :) which results in the message being queued _within CM_ -- I
have no problem with that; but I do have a problem with it being tagged
send LATER.  It should just be queued until the MUA-MTA link is free/up.

>> As to your fedora 20 thread -- there, you admit to using a local MTA...
>> so what?  Physical location does not change the way mail works.  Just
>> because you don't see something because of speed, and a very reliable
>> link does not make MUAs and MTAs operate differently.
>
>Well, yes and no.  I'd say a local socket connection really is very
>different from a remote one, and the difference goes beyond speed.  But

Virtual v. real links...  :)

>in a way I agree - I would prefer claws to not talk SMTP at all, and
>fork a copy of /usr/sbin/sendmail instead.  Unfortunately this runs into
>the Windows problem.

sendmail is an MTA which is often replaced by postfix nowadays. IMO, it's
unreasonable for an MUA to expect/require a local MTA.  No differently
than elm, mutt, pine, kmail, Eudora, Thunderbird, Outlook, etc; all of
which I've used over the years.  :)

Cheers,
Pierre