Hello Russell, On Sat, 2015-10-17 at 20:46 +1100, Russell Coker wrote:
I have found the cause of the DKIM problems. When Mailman unfolds and refolds the DKIM-Signature header it replaces tabs with spaces. This makes the difference between a message that passes according to /usr/bin/dkimproxy-verify (part of the libmail-dkim-perl package) and one that fails.
Is anyone here any good at Python who can have a look at this? Some special case code to use tabs for the DKIM-Signature header should solve these problems without causing any other problems.
Note that a problem that was reported in the past was Mailman replacing spaces with tabs when folding the Subject header. Obviously we don't want to regenerate that problem so changing the behavior of all headers would be a bad thing.
Sounds like the issue is to configure Mailman to leave spaces as spaces, and tabs as tabs, if possible. There may be a standard as to whether the headers should have spaces, or tabs, but I would expect that conformance by the mail readers would be more in the breach than being correct, especially in the light of Outhouse (outlook). Regards, Mark Trickett