[Top] [All Lists]

Re: Strict RFC x821 Compliant: MAIL FROM:

2005-07-06 08:22:05

--On Wednesday, 06 July, 2005 10:55 -0400
Valdis(_dot_)Kletnieks(_at_)vt(_dot_)edu wrote:

On Wed, 06 Jul 2005 12:42:33 +0300, Matti Aarnio said:

commonly, without brackets, etc.    Some machines are clearly
behind some NAT box:  "EHLO []"  (yup, it had
correct brackets)

It's things like that which caused the RFC prohibition on
bouncing mail due to the EHLO value - that machine was being
as strictly compliant as it could.

On the other hand, I could *easily* make the case that you
should reject on site any 'MAIL FROM: <..' crud from certain
vendors as a security measure. That sort of blatant inability
to code per spec indicates a site that should be blacklisted
because the next time you hear from them, it's likely to be Yet
Another Worm/Virus (if the one you're looking at isn't one

For whatever it is worth, it seems to me that this is _exactly_
the right way to draw the distinctions in the standard.
Behavior that is the best possible out of an SMTP client or
server should not be banned.  That is consistent with
robustness, maximum conformance, and general good sense.  And,
if a host has no global DNS name and only a NAT-supplied private
space address, I'd rather see the domain literal above than
either a lie or 
   EHLO Jones-Family-801.11b

Conversely, if the spec is clear and imposes a requirement that
has been there for 20 years, and that requirement is easy to
conform to if only one manages to read the spec, punishing
non-spec-reading by, at least, making the behavior
non-conformant, seems entirely rational.  As you imply, such
clients, and the sites that run them, will probably exhibit
worse behavior, and behavior that is more indifferent to others,
on the next message.