On Tue, Dec 07, 2004 at 08:01:54AM -0500, Nico Kadel-Garcia wrote:
I already explained that I do not expect nor want a bounce from
you3. I may not even be able to communicate (for _whatever_ reason)
with you3.
And Alex, you're a nice guy, but you're just plain weird. In almost all
such cases people want to know about the bounce ASAP, because it means
their email didn't get through. They don't care much whether the bounce is
at the forwarding SMTP server or the final target SMTP server.
So, please describe to me how a server in China should hand me
the bounce when I decided to block all of the address space in
China that I know of. I don't send messages there, I can't receive
bounces from there.
Your point is that I want to know delivery failed. I don't get
to know about this.
If the final target SMTP server returns the message to the forwarding
SMTP server, and if the forwarding SMTP server returns the message to
me, all is well.
If I send a real letter to you, and you leave a forwarding address but it
can't be delivered because your house burned down, should that mail wind up
sitting at the first post office? No. The chain of SMTP servers along the
way to the recipient is relevant to we who keep them working, but most
senders simply don't care. They just want their email to get there, and by
implication they want to know when it fails. Your approach to forwarding
badly breaks that information and dumps responsibility for it on the people
least likely to know how to fix it, the admins at the forwarding SMTP
server.
This analogy requires some massaging because SMTP and snailmail cannot
be mapped one on one. Nevertheless, I can comment on this situation:
I send mail to "you(_at_)yourhouse".
Someone at yourhouse decides to resubmit the message to "you(_at_)otherplace"
The postoffice responsible for otherplace cannot deliver the message.
The postoffice at otherplace will initiate a return to sender.
So far, so good.
The returned mail will most likely take a similar route (if not the
same route) back to me. That's exactly what I'm proposing all the
time.
The returned mail will be delivered to me by my local postal service,
by someone familiair to me, by an entity I recognize. It won't be
handed over to me by whatever service it was that tried delivery to
otherplace.
Until now, your analogy supports my proposal very well. Thank you for
that.
Now we're going to mix in a bit of fraud (compare this to viruses and
spam). I've sent a couple of letters to you. You want to reply to my
letters but you do not want to pay for it. So, what do you do? You
tell the postoffice this letter should be forwarded to your new address.
The postoffice accepts it, tries to deliver it, fails, "returns" the
letter to me, I accept the "bounce" and start reading your reply.
Nice, we can commit fraud because of a friendly postal system.
Repeat. Repeat again. At some moment, more and more people start
this kind of fraud. You can bet your **** that the postal system
will not accept messages for forwarding anymore, despite several
people complaining about the change. It simply would not be allowed
anymore and both black hats and white hats suffer from this.
There is another method to forward the mail:
1) I send to you(_at_)yourhouse
2) someone(_at_)yourhouse sticks my letter in another envelope
3) someone(_at_)yourhouse sends the letter to you(_at_)otherplace
4) otherplace cannot be reached
5) postal service returns letter to someone(_at_)yourhouse
6) envelope is opened, my letter is found
7) someone(_at_)yourhouse returns letter to me
Of course, "someone(_at_)yourhouse" equals
"postmaster(_at_)yourdomain(_dot_)tld"
Again, your analogy supports my proposal.
cheers,
Alex