On 2008-03-11 at 20:52 +0000, David Cantrell wrote:
> I beg to differ.  While I agree that it's complete and comprehensive, it
> is largely impenetrable unless you already have a good idea about how to
> drive exim.  The O'Reilly book for exim 3 suffers from this bug too IIRC.
> No idea about the CUP book for exim 4 yet though.

*shrug*  I learnt from it.  Biggest problem is people skipping past the
early stuff thinking that it's the usual lightweight fluff that you
commonly find padding out books these days.  You can rapidly tell which
people have bothered reading "3. How Exim receives and delivers mail"
and which people are ignoring it.

If you (in the general case, not you David) don't want to learn a
mail-system, don't; use Debian's stuff, set the couple of variables,
you're about at the level of the Sendmail users who say that Sendmail's
not complicated because you just set a couple of macros.  The Debian
maintainers do a fairly decent job of making sure that the results are
usable and not dangerous to you or others.

If you want to do something not covered by the basic common questions,
then most of the time, judging by what I've seen, it's because people
have some seriously bad ideas.  Sometimes, people do understand mail and
just want to do things their way, or do things more optimally.  At that
point, there's no substitute for damned well reading how the chosen MTA
processes mail and learning "Routers are tried in order, until one
accepts the message, deciding where to send it.  Transports are an
unordered collection, referenced from the Routers.  ACLs let you make
ordered decisions on inbound mail, hooking into various steps in the
SMTP (or non-SMTP) (non-)dialogue.  There's a queue.  There are hints
databases to cache results.  And *gasp* you can run the MTA with
debugging turned on, to various depths, to see actual useful data about
why the MTA did what it did."

I think that I've about had my fill (again) of people unable to put in
the effort to understand even that, whilst demanding that other people
help them, and the leaches ("pay a genius to do it for you" or however
it goes) who can't accomplish even the most basic tasks without
demanding that others do the work for them and never contributing back
whilst bitching if someone calls them on it.

My previous statement on how many people who Hates Software help out on
exim-users might be about to drop back by one.  Not because Exim sucks,
but because it's sufficiently easy that most (but by no means all) of
the people who actually need help are too wilfully clueless to even read
the documentation, even after repeated prodding.

My magical fault-diagnosis crystal ball is almost drained and needs
another two year recharge.

-Phil

Reply via email to