Quoting from Eric S. Raymond's mail on Mon, Sep 24, 2001 at 06:20:46PM -0400:
Sunil Shetye <shetye(_at_)bombay(_dot_)retortsoft(_dot_)com>:
Quoting from Sunil Shetye's mail on Wed, Sep 05, 2001 at 12:06:33PM +0530:
fetchmail does not send any bounce message now on a bad address.
I think the patch in 5.8.17 has removed the call the send_bouncemail()
causing no bounce messages to be sent in any case.
This patch should undo it...
OK, now I'm officially confused.
I remember taking this code out. It was after 5.8.16 in response to a
bug report by Quoc Luu. He reported:
Well, I don't think that bug report was sent to fetchmail-friends list
then. I have read one and only one bug report from "Matthias Andree"
<ma(_at_)dt(_dot_)e-technik(_dot_)uni-dortmund(_dot_)de> which stated:
==============================================================================
FF>> Problem #4:
FF>>
FF>> fetchmail sends two bounces per failed mail, the first one is ok, the
FF>> second one looks somewhat broken (mind the Diagnostic-Code: and Status
fields):
FF>>
FF>> First bounce: (note 550 is not in antispam)
FF>>
FF>> | General SMTP/ESMTP error.
FF>> | ---
FF>> | Reporting-MTA: dns; ludwig.ping.de
FF>> |
FF>> | Final-Recipient: rfc822; testuser
FF>> | Last-Attempt-Date: Sun, 05 Aug 2001 19:41:48 +0200 (CEST)
FF>> | Action: failed
FF>> | Status: 5.0.0
FF>> | Diagnostic-Code: 550 <testuser(_at_)ludwig(_dot_)ping(_dot_)de>: User
unknown
FF>>
FF>> Note that the Final-Recipient is unqualified, which is not
FF>> allowed. 5.0.0 is not very specific, but given that neither Postfix nor
FF>> fetchmail support RFC-2034, that may be the best we can get.
FF>>
FF>> Second bounce:
FF>>
FF>> | Some addresses were rejected by the MDA fetchmail forwards to.
FF>> | -----
FF>> | Reporting-MTA: dns; ludwig.ping.de
FF>> |
FF>> | Final-Recipient: rfc822; testuser
FF>> | Last-Attempt-Date: Sun, 05 Aug 2001 19:41:48 +0200 (CEST)
FF>> | Action: failed
FF>> | Status: t.0.0
FF>> | Diagnostic-Code: testuser: 221 Bye
FF>>
FF>> The Status and Diagnostic Code looks bogus and don't conform to
FF>> RFC1891. Plus, "Action: failed" does not match a "221" reply code.
==============================================================================
...and here was your response to it...
==============================================================================
FF>> > Problem #4: bogus bounces
FF>>
FF>> I fixed this, too.
==============================================================================
v5.8.15 is still handling bad addresses weird: it sends 2 bounce messages
to the sender and forwards the original message to postmaster without any
error message.
So I removed the source of one (1) bounce message in this case. I believe
my reasoning was that the SMTP listener fetchmail passes the message to would
itself ship a bounce message in this case. Do you have any idea why this is
not happening?
My comments are only based on the above mentioned bug report...
I think it would be better to verify if that bugreport was true in the
first place...
I had looked through the code in version 5.8.16 and it did not seem
possible for two bounce messages to be sent even in the case of weird
errors...
That bug report had specified two messages with the following errors:
"General SMTP/ESMTP error" which is generated by handle_smtp_report()
which is called by open_sink() for errors in from address. open_sink()
returns if there are any errors in from address.
"Some addresses were rejected by the MDA fetchmail forwards to" which
is generated by open_sink() for errors in to address. This part of the
code is not reached when any errors in from address are found.
The bug report unfortunately had not printed the fetchmail
transaction, it had only printed the SMTP transaction. So, the bug
report was incomplete. It could not be deduced from that bug report if
the error messages were related to the same mail in the first place!
Sunil Shetye.