> I have heard that writing a script to use qmail-remote
> directly and only use qmail-send when the initial
> delivery fails can work wonders. (can anyone share
> code?) But is it enough to beat Lyris?
One question is: where does the content come from? This script that
calls qmail-remote directly has to maintain state, which, if the
content is unique, includes the content. Normally we call this state a
mail queue.
If the content is customizable/derivable - and I use these terms to
differentiate it from truly unique content - then what you have is a
specialized mail engine which can therefore use a specialized queue.
All other things being equal, a specialized queue will deliver a lot
faster than a generalized queue, such as qmail.
While qmail is fast, the design does trade off some speed for
flexibility, reliability and security in a shared Unix environment. If
you can forgo some or all of these trade-offs, then you can do a lot
better than qmail.
Regards.
