[ On Saturday, February 22, 2003 at 15:16:43 (-0500), Greg Louis wrote: ]
> Subject: Re: [Exim] MX Record points to non-existent host
>
> I sure do expect that the FAX machine at the other end will dumb itself
> down to my level if it finds I can't smarten up to its.
The "dumbing down", as you put it, is still entirely controlled by the
official International standards defining how these machines talk to
each other. Without the basic "dumb" standards there is no possiblity
of it ever interoperating with your standards- only machine. Period.
That so-called FAX machine at the sending side isn't ever going to talk
to your FAX machine.
Internet e-mail standards have already been dumbed down. That's how
they started out, right from the very beginning. It can't get much
simpler.
> > If you operate a strictly RFC-compiant infrastructure then you are
> > already allowing anyone to contact you if they wish.
>
> No. We would be allowing anyone to contact us if their bosses can
> afford, and choose to afford, technical folks who have the knowledge
> and the resources to do it right. That's not the same thing at all,
> unfortunately.
Unfortunately you are so completely wrong that you don't even know it. :-)
Do you expect your customers to be able to send you e-mail over the
Public Internet even if they spell your domain name wrong? Sure you can
go out an register all the obvious mis-spellings, but that just makes
the problem worse because by doing so you set expectations that you can
_never_ meet.
Do you expect your customers to be able to send you e-mail even if they
operate a broken, or worse, proprietary, TCP stack? They might be able
to communicate with anyone else running with the same bugs, but are you
going to run the same buggy software as they do just so that they can
send you e-mail? What are you going to do when the next different kind
of breakage comes along? Are you going to stop talking to one set of
customers just so you can talk to the next new set of customers? (I've
had this very same conversation with "security officers" who run
firewalls that break the Path MTU Discovery protocol that the very same
servers they're firewalling demand everyone else on the outside use.)
Where do you draw the line? I'll tell you where you must (from the
point of view of all the rest of the members of this community) draw the
line: right where the RFCs tell you to draw the line.
If you do not do that then you create a huge amount of animosity amongst
the rest of the Internet community -- among those of us who are not
willing to put up with an infinite number of subtly incompatible systems
that are clearly violating all the standards we have agreed upon.
> If much of the rest of the world operates the proprietary equipment,
> they will stop trying to send us FAXes, all right. They will just
> smile disdainfully and go buy from someone else.
Sure, but here the real problem is exactly the other way around. Only a
few of the customer you're trying to attract are likely to be running
with a broken setups. You are effectively proposing that all the rest
of the world should break their software too just so that you don't hurt
their poor ignorant IT guy's feelings.
Sure, you can do whatever the heck you want with your systems. However
if you even suggest to someone that because their broken systems can
send e-mail to your broken systems then they should be able to send
e-mail anywhere, and if someone like myself finds out about this, you're
the one who's really going to get "corrected". It's bad enough that you
WILL be giving them this impression indirectly if you adjust your
software to deal with their brokenness and if you don't help them fix
their bugs.
The global public Internet really is a community effort. If you want to
be a member of this community in good standing then you'll seek out and
help those of your customers who are stumbling along with broken setups.
You don't have to take on the world to do it -- just help them one by
one. You don't really want to be the kind of guy who'll sell your
product to everyone, even those with bad credit and/or fraudulent
identities etc., do you? You can give good service to good customers
without having to drag everyone else into the gutter too.
> I used to say just that until it became apparent that it was utterly
> futile.
It's obviously not futile. However by doing what you're doing you are
becoming part of the problem, not part of the solution.
> They just say "everyone else gets our mail." Worse, my users
> say, "everyone else gets their mail."
Then they are all wrong, perhaps by omission, even if they don't know
it. You need to correct them and educate them and help them. Do not
ever perpetuate or tolerate such broken perceptions. You can do it with
a clue-by-4 or you can do it with a carrot -- your choice.
> I'm not devoid of sympathy with that sentiment. But from their point
> of view, if it works with lots of other domains, why is it a problem?
> That's the perception that makes it really hard to convince them.
You can dial random numbers on your telephone and most of the time
you'll end up talking to someone. However if you always very carefully
dialed exactly and only the right numbers but frequently you didn't end
up connected to the location you thought you dialed then I bet you'd be
pretty un-impressed with the telephone system.
The only difference here with e-mail is that everyone runs their own
equivalent of a local telephone system. Everyone has to manage their
own number registries (DNS) and their own signalling systems (mailers)
so that they can talk to all the other systems in the world and vice
versa.
Sure there are some kinds of bogosities that will be tolerated by some
systems some of the time so that some of the time some of the people
using those broken systems can communicate with each other. That's not
the problem -- the problem is where it doesn't work. It's the
unreliability of error handling that really causes the deepest problems.
Just because something works some of the time doesn't mean it will work
all of the time.
As for people who buy and use broken software out of ignorance, well the
same thing applies: They need to be made aware of their poor choice and
its consequences; and the software company should be informed of the
problem too (and if they refuse to fix it then perhaps public pressure
can be brought to bear upon them).
> What I would like is for my MTA to tolerate as much error as possible
> on the other end while behaving perfectly in every other respect.
That's what the RFCs already allow for. They're already the most
lenient and forgiving communications standards the world has ever seen
(at least for any kind of modern technology driven form of
communications).
If you really want to allow your mailer to understand something like an
IP# instead of a canonical hostname in MX records then I strongly
suggest you first write an RFC describing how this be done and propose
it to the IETF. In the mean time you'll be doing us all a favour if you
operate your mailer so that it _only_ honours the existing standards.
--
Greg A. Woods
+1 416 218-0098; <g.a.woods@???>; <woods@???>
Planix, Inc. <woods@???>; VE3TCP; Secrets of the Weird <woods@???>
