[Dovecot] vpopmail

[Matthias Andree]

(cutting down Cc list)

On Mon, 05 Jun 2006, John Peacock wrote:

> >The current setup:
> >- Pine configured to send mail via localhost (MTA)
> >- MTA on localhost send mail to SMTP server
> >- SMTP server allow/reject connections based on POP-before-SMTP
> >- SMTP server performs spam and virus checks
> 
> This isn't that unreasonable a setup - because the laptop may not be 

The setup is unreasonable, we already agreed that POP-before-SMTP
alone is unreasonable and dangerous in dynamic-IP environments.

> connected at all times, he wants to have a local MTA on the laptop to 
> deposit mail into while offline.  That MTA is configured to relay all 
> mail via his primary MTA, which scans incoming/outgoing mail and 
> delivers it to the appropriate remote MTA.  As such, whether he is 
> running postfix or qmail (or Sendmail) *on the laptop*, he wanted to 
> know why POP-before-SMTP broke on his primary server.

This was already answered a few posts ago, with the side note that
POP-before-SMTP opens security loopholes.

I understand the intentions and the situation (and, as yet another
sidestab, "this can happen with betas"), and the chosen MTA that
is incapable of SMTP AUTH is a fact that matters here, since the
limitations of the MTA cause such crutches as POP-before-SMTP -- and
they are quite besides the track given that the chosen MTA claims
security.

> It's not that I disagree with you that POP-before-SMTP is suboptimal and 
> should be replaced with AUTH; it's that you are being such an a$$h0le in 
> your anti-qmail fervor, that any wisdom you may be imparting is lost in 
> the vitriol.

There's much less emotion ("fervor") in my calling qmail decrepit than
you suspect. It would be unwise to not talk about qmail being the agent
that caused the POP-before-SMTP dangers.

And, yet another side note, that POP-before-SMTP may fail after software
upgrades, is just another reason to replace it.

-- 
Matthias Andree