[KLUG Members] exim? postfix? why not?

[Adam Williams]

>Subject: Re: [KLUG Members] exim? postfix? why not?
>>True. They may be pretty much over.  And they probably all have a
>>sufficient user base to survive into the foreseeable future;
>That, I am sure, is another reason the roar died down: when sendmail was
>all there was, no one knew any better; when the second SMTP MTA was
>written, all sendmail's idiosyncrasies (and flaws) came to the surface;
>when the third was written, then people really had to agree what was
>what. ;-)

Probably a theory with alot of validity.  

So if we create a third desktop environment project would that spur on 
both GNOME and KDE to new heights?  It would be an intersting study to see 
how many popular and sustained projects each niche can support.  I suspect 
the further back-and-up the chain (away from the user) you move, the more 
diversity is sustainable.  Anyone have any URLs about anything like this?

>Also, period two above came about when sendmail was in that 'hung,'
>unsupported state: even worse.

I remember,  before the Open Source quasi-revolution.

>>Right, but that illustrates the primary problem I found with other
>>MDAs.   If one thinks the docu for sendmail is bad (and I do), the
>>docu for the others is, in a word, horrible.  
>Oh, no, Exim has great documentation:
>http://www.exim.org/exim-html-3.30/doc/html/spec.html

Indeed.  That is very nice.

>It's just that when I was setting up Cyrus, I read through the docs
>until I saw 'lmtp transport' where I halted and read. Had I continued, I
>would have seen that the smtp transport also speaks LMTP. :)

Let me know how switching to LMTP over the SMTP transport works out.

>>That would be a HUGE draw back, IMHO.  Looking at our mail stats, not
>>performing the error in session would result in a significant increase
>>in our mail volume.
>>I'd be surprised if Exim/Qmail/et al. didn't really support milters
>>however.  I assume it is `just' a matter of assembling a HOWTO;
>>requiring about 12 hours with Google, a crystal ball, and
>>psychological profiles of the lead developers.
>If you don't mind running embedded Perl in your mailer:
>http://www.exim.org/FAQ.html#SEC207

Nah, I'll pass.  Seems like something it would be very simple to render 
insecure.  And I'm not much of a Perl monkey.

>No, Exim has many settings to set/unset bouncing a message before it is
>officially accepted (RBL et alia, reverse lookups, recipient matching,
>header syntax verification, ad nauseum) but the filter language
>(decidedly Sieve-like, I think) is definitely post-SMTP-acceptance.
>Approximately how much mail do you refuse based on content filters?

On a very slow day, over a hundred.  If a new worm is hatched, then we get 
smashed hard.  It is just a hazard of our industry, most shops are run 
with no IT staff at all, predominately Win32 PC's connected dirtectly to 
the internet and running Outlook.  One of our competitors actually got 
effectivley knocked off the 'net for more than 24 hours a couple of worms 
ago (I forget what the name was).  That was nice for us,  but I don't want 
it to be me next time (especially since I made so much fun of them).  So 
efficiently saying "NO!" is very important.  I'm quite certain it is only 
going to get worse.  With sendmail and milter-vilter and McAfee I see an 
occasional hump on the OpenNMS load average and traffic charts, and thats 
about it.