Date: 7 Mar 1996 19:27:13 GMT
Message-ID: <19960307192713.18104.qmail@koobera.math.uic.edu>
From: djb@koobera.math.uic.edu (D. J. Bernstein)
To: drums@cs.utk.edu
Subject: Re: re PIPELINING

> you will see that I stated my assumption about what you are doing and 
> concluded that if, in fact, you were doing things that way, it was a 
> ridiculous thing to do.

That assumption was correct. Your conclusion was nevertheless wrong.

The problem was your unstated assumption that ``foreach'' had to be
done in serial. This is what I pointed out in my response.

If you had realized this at first---which you should have, given my
comment about latency---you could have said ``I presume that you're
referring to the possibility of handling multiple recipients in several
parallel SMTP connections,'' etc.

Btw, I apologize for referring to your comment as ``bullshit''; an
_honest_ mistake doesn't qualify as bullshit.

> I still believe there are technical problems with your approach

Such as?

> In addition, I may now be forced to put a protection mechanism in my own 
> server to limit the number of simultaneous connections from any one host, 
> just in case it is running qmail.

You're not running a load-limiting SMTP server? Yikes. That's more basic
than a security problem---that's a reliability problem. Lots of programs
(not the mailer, of course) will crash and burn if you run out of memory.

---Dan