Aaron, please note that read_whole_message_network has been deprecated in
cvs-head.
[EMAIL PROTECTED] wrote:
A BUGNOTE has been added to this bug.
======================================================================
http://www.dbmail.org/mantis/bug_view_advanced_page.php?bug_id=0000145
======================================================================
Reported By: bjohnson
Assigned To:
======================================================================
Project: DBMail
Bug ID: 145
Category: LMTP daemon
Reproducibility: always
Severity: minor
Priority: normal
Status: new
======================================================================
Date Submitted: 04-Jan-05 01:32 CET
Last Modified: 18-Jan-05 13:15 CET
======================================================================
Summary: LMTP loses return-path
Description:
This is actually reported against 2.0.1 but that was not a choice in the
product version.
LMTP receives a received a "MAIL FROM:" but does not store the return-path
in the database.
The logs attached below show:
1) MAIL FROM command was received with the return-path recipient
2) dbmail-lmtp does not log the fact that it inserts the return-path
3) A database query in fact verifies that the return-path was not inserted
into the database.
======================================================================
----------------------------------------------------------------------
aaron - 12-Jan-05 19:41 CET
----------------------------------------------------------------------
What the issue here? I don't believe that the envelope sender is kept
neither with lmtp nor with pipe/dbmail-smtp deliveries. Do you want to
start recording the envelope sender in the database?
----------------------------------------------------------------------
paul - 12-Jan-05 20:15 CET
----------------------------------------------------------------------
Required reading:
Common Internet Message Headers
http://www.faqs.org/rfcs/rfc2076.html
Return-Path: should be there in a message that leaves the smtp
environment.
Apparently the mta doesn't think the message is leaving smtp when using
the lmtp transport.
----------------------------------------------------------------------
aaron - 12-Jan-05 20:43 CET
----------------------------------------------------------------------
Ok, looks like lmtp's job to prepend this header. I'm reading through the
code right now to see where a good place to insert the extra header might
be...
----------------------------------------------------------------------
aaron - 18-Jan-05 13:15 CET
----------------------------------------------------------------------
Ok, here's an idea: read_whole_message_network is static to lmtp.c, and it
malloc's the whole message buffer. So we can add an argument,
"header_prepend", that gets copied into the new buffer first thing before
any network reading happens.
Bug History
Date Modified Username Field Change
======================================================================
04-Jan-05 01:32bjohnson New Bug
12-Jan-05 19:41aaron Bugnote Added: 0000524
12-Jan-05 20:15paul Bugnote Added: 0000525
12-Jan-05 20:43aaron Bugnote Added: 0000526
18-Jan-05 13:15aaron Bugnote Added: 0000542
======================================================================
--
________________________________________________________________
Paul Stevens [EMAIL PROTECTED]
NET FACILITIES GROUP GPG/PGP: 1024D/11F8CD31
The Netherlands_______________________________________www.nfg.nl
