OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: Bug in local delivery? (does username case matter?)
From: Jonathan Bartlett (johnnybwolfram.com)
Date: Fri Jun 23 2000 - 04:24:54 CDT

Also, postfix doesn't have to be everything to all people. There _should_
be situations where you just say "You'll have to use another mailer".
Either that, or have a generic plugin architecture. It would probably be
a lot easier to only code the 90% solution, and then write another mailer
to cover the 10% situations, rather than trying to fit all 100% of the
cases into the same delivery system.

At the place I work, we came to this conclusion:

If you want a mail gateway between different mail systems
        use sendmail
If you want a super-configurable mail system
        use exim
If you want a super-fast and secure mail system
        use postfix

Where I work, we are getting ready to transition to using postfix for our
external mail gateway, and exim for our internal mail server (our
users have really nutty configuration requirements). Anyway, the same
system doesn't have to work everywhere. And, being that the code is
free, people can customize it for their own solution if postfix is
_almost_ correct for what they want. Anyway, I think for most people,
speed, security, simplicity, and robustness are more important than
infinite features. Anyway, the more features you have the less simplicity
and speed you have too. So, before trying to handle every possible
request, make sure it fits in with what you are trying to accomplish.

Jon

On Fri, 23 Jun 2000, Wietse Venema wrote:

> Wietse Venema:
> > Postfix maps usernames to lowercase. Not only the local delivery
> > agent does that, so does the canonical and virtual mapping.
>
> At least, this is the present situation. I expect such
> things to become configurable on a per-table basis.
>
> Time to market (1), 10%-90% rule (2), and all that (3).
>
> Wietse
>
> (1) Completely finishing Postfix before release would have taken
> too much time. Therefore the released code is work in progress.
>
> (2) In some cases, 90% of a problem can be solved with 10% of the
> code that makes up a complete system.
>
> (3) There was a time when a system could be complete. That was when
> systems changed very little over time. That is a very long time ago.
>