[Ian Jackson]

1. Are the mailman developers aware that mailman can lose mail ? The message which mailman tried to send (via SMTP) was rejected by my local Exim because the sender was wrong, and the only trace I can now find of it is in the various error logs. The message body is nowhere to be found.

Yes, we are aware of that. I believe this is a design issue: Mailman should rather dequeue than defer messages that for non-obvious reasons aren't accepted by the SMTP server. This means, among other things, that Mailman won't try resending a message that tickles some bug in your MTA over and over...

However, "non-obvious reasons" should not include error responses on MAIL FROM. In current CVS, such errors cause deferral and writes a message stating the problem to Mailman's logs.

[ I considered having Mailman try to contact the site Mailman admin, as well -- but as the likely cause of refused senders is relaying restrictions being set too tight, doing automatic mailing could rather easily start an avalanche of messages. ]

2. What is the published, documented way of finding out or resetting a list owner password ? I can probably dig into python and get it out that way, but I would like a proper published interface ?

The list admin passwords are stored as one-way crypted strings (either with MD5 or with standard UNIX crypt). So, to find out what the password was, you'll have to do some brute-forcing...

Change of list admin passwords can be done at the bottom of the DEFAULT_URL/admin/LISTNAME web page, by using the site admin password as the "old" list password.

3. What is the site admin password for ? I deduce its existence from the presence of a program to set it ...

See answer to previous question. The site admin password works everywhere either a list admin password or some member's password would work.

In general I have a feeling that I'm missing some of the documentation.

So do I -- Mailman documentation is currently all too sparse. Any effort to change this would be greatly appreciated :)

Much of the code have good docstrings and is quite well commented, though.

Harald