Alessandro Vesely <[email protected]> writes:

> Correct, the client MUST --according to RFC 5321-- give a resolvable
> FQDN, if possible, or else fall back to an address literal. However,
> that RFC further specifies that
>
>   An SMTP server MAY verify that the domain name argument in the
>   EHLO command actually corresponds to the IP address of the client.
>   However, if the verification fails, the server MUST NOT refuse to
>   accept a message on that basis.

Eh?  What's the point of having a standard if you aren't allowed to
refuse to support things that don't comply with it?  Do we also have
to accept messages with multi-thousand-character header lines?  No.
We should try to be liberal in what we accept when we can, but when a
particular violation of the standard creates too many problems, we
tell the other host to go jump in a lake (and then they either fix
their implementation to comply with the standard, or they communicate
only with a subset of the internet, their choice).  In the case of
email, not being able to check the identity of the sending host can be
a fairly significant source of problems.

Besides that, whether to accept mail from any given host is a matter
of site policy.  The role of the RFC is to provide the communication
framework so that everyone is speaking the same language, so that when
the receiving server says "Yes, we will deliver that message for you"
or "No, we won't deliver that" or "Not now maybe later", the sending
server can understand the response and know things like whether it
should try again, notify the user of a failure, or consider the
message delivered.  Deciding who we should accept mail from is most
definitely NOT the role of the RFC.  That's between me, my users, and
the people we want to communicate with by email.  If we want to only
accept mail from specific whitelisted hosts, that's our business.

>> in the Glorious Future when the software does everything...
>
> That's also the stance of RFC 5451, which defines the
> Authentication-Results header field that a server taking the
> responsibility to accept a message would add. The header describes
> the result of various checks, given in a uniform syntax. Besides SPF
> and DKIM, it provides for an "iprev" authentication method,
> described as matching [r]DNS queries, but without going down to the
> actual implementation details.
>
> In the Glorious Future, mail clients will check that header field...

I'm not totally in full agreement with that, either.

If the message is going to be rejected, it should be rejected as early
as possible.  If a message can be rejected before the DATA stage, it
should be.  Even if it goes to the DATA phase, if you then know that
the user isn't going to be looking at the message, it should be
rejected as soon as you know this.

Among other things, issuing an error response during the SMTP session
allows the sending mail server to immediately notify the user that the
message was not delivered.  In the event that a real user was actually
trying to send the message, this is something they want to know.
(Third-party bounce messages are no longer worth anyone's time to read
these days because five nines of them are spam, but when your own mail
server tells you that a message you tried to send couldn't go out,
that's different.)

I'm not saying such a header can't be useful.  It can, obviously.
Sometimes the same mail server has to support both users who get a
trainload of spam and want as many ways to reduce the flood as
possible, and also users who are paranoid that they might miss an
important message.  And it isn't always practical to let the users
install policies on the server.  ISPs for home users are a good
example here.  In such cases, a header can be useful, as it allows
individual users to be more strict in what they'll read than the
server can be in what it'll accept.  Sure.

But not all mail servers are in that situation.

-- 
Nathan Eady
Galion Public Library


------------------------------------------------------------------------------
Download Intel&#174; Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
courier-users mailing list
[email protected]
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users

Reply via email to