Re: [Exim] Forgive my ignorance ...

Page principale
Supprimer ce message
Répondre à ce message
Auteur: Phil Pennock
Date:  
À: exim-users
Sujet: Re: [Exim] Forgive my ignorance ...
This does have much content relevant to Exim and to migration issues.
Those avoiding "off-topic" mails might wish to read this one, since I
give some explicit pointers to how a migration to Exim caused problems.


On 2002-03-28 at 11:53 -0600, dman wrote:
> | His employer is my ISP.
>
> That has nothing to do with Phil himself. If he doesn't do as his
> boss tells him, he won't have any food on his plate and thus won't be
> around (here or there) to help with exim.


I'm going to do the unthinkable.

I'm going to come to the defense of my boss!

Engineers the world over will now doubtless shun me. I'll survive,
somehow.


I refer those who are confused by the propaganda, to my earlier post on
the topic, from my work account, where I explicitly said that we're
working on migrating to Exim. To clarify: We do want Exim. We want
more features, and we're planning on providing them by moving to Exim.
Exim Rocks!

But in the Real World, this is non-trivial. At least one previous
attempt has been made, within my employer, to transition to Exim. An
interaction with another system had been overlooked, and it led to the
transition being backed out. This was not Exim's fault. The
mail-system of a large and established ISP will contain some complexity,
especially when interfacing with a lot of in-house code.

Hrm, going and digging out the public announcement, it doesn't
immediately look like it was Exim; if it wasn't, then it's an even more
salutary tale as it would have been an update within the same MTA, just
between versions. I _think_ (but am not 100% sure) that this is,
however, something that happened with an Exim migration. It was fixed,
without rolling back, but _other_ issues led to a later rollback. I
think. Drav?

Those interested in a diagnosis of what, exactly, can go wrong and how
lessons are learnt the painful way might wish to read:
<URL:http://www.demon.net/helpdesk/announce/1999/da1999-08-24a.shtml>


Back to the current day ... Systems which are relatively easy to
transition, have been transitioned. Check the hosts which respond to
the name of post.demon.nl -- they're all running Exim. :^)

My employer's current customer-bound mailsystem, MMDF, is not
fully-featured, but it is not broken. We would like an 8-bit clean
mailsystem (Exim by choice), but the one we have _does_ work and is not,
to the best of our current knowledge, corrupting mail which is sent
according to the IETF standards, draft standards, proposed standards, or
any other RFC which I know of. If people violate the standard then they
should expect to get bitten.

Given the severity to an ISP of that accusation (not many ways to more
quickly scare off customers) it's hopefully clear we do take such
allegations very seriously and do investigate. No evidence has been
provided in this case and the accusation appears to be frivolous.


The Boss does want us moved to Exim. With regards to food on my plate,
though, I should point out that for the most part, I'm not involved with
the Exim migration. Nor do I have time to be, unless a number of other
duties are reassigned.


quoting Tony:
> | Demon is NOT rfc1652 compliant.


Actually, it is. RFC 1652 has a pre-requisite of supporting EHLO, and
requires that the server advertises 8BITMIME. The customer-bound
mail-systems do neither, and therefore is compliant. The RFC is in fact
quite explicit in explaining this scenario, as others on the list have
pointed out.

> Nor is exim.


Exim is, by default. It provides a tunable knob, because people wanted
it. The documentation for the knob states quite clearly that it will
break things. If people want to break things, that's their choice. I'd
rather live in a world where people can break things and screw up than a
world where they're not allowed to. That doesn't mean that those people
should expect everyone else to bend over backwards to do whatever's
necessary to handle their breakage.


I'd just like to thank those on the list who've taken the time to try to
explain reality.  It is appreciated.  :^)  After all, Galileo got in
trouble for explaining reality to those who didn't want to hear it.
Reality shouldn't be argued with.  Standards are a human invention.
Sticking to standards isn't the same as defending an incorrect
worldview.     ;^)
--
Ignorance is only a sign of idiocy the second time around.