Ian R. Justman wrote:
Hi, all.
Thanks for your thoughts on my previous questions.
Now, I have a new dilemma on my hands.
I'm trying to adapt the script "postfix-to-mailman.py" to operate with
Ecartis rather than Mailman. What it boils down to is that I'm not
having a lot of luck with it right now.
First, some knowns:
1. I have the original script, postfix-to-mailman.py, working with an
existing Mailman installation. I really like how it works which is why
I'm using this as a base (why reinvent the wheel?). Some ancillary
info: this is for a list which a friend of mine runs, notably the
IRCServices-related lists.
2. I am installing Ecartis for another entity--one I helped
found--because I'm far more comfortable with it and I used it before.
3. I don't like dealing with aliases for mailing lists. They cause
unnecessary clutter.
4. I like Ecartis. (Important)
5. I hate Mailman. (Really important)
6. I don't know jack about Python. (Extremely important)
The postfix-to-python.py (and by extension, my adaptation,
postfix-to-ecartis.py) are set up as pipe transports, the entry for
which in /etc/master.cf being something like this:
mailman unix - n n - - pipe
flags=FR user=mailman
argv=/[...]/mailman/mailman-base/postfix-to-mailman.py
${nexthop} ${user}
And in this case, it's called with the following line in a transport
table (this is the actual line):
ircservices.za.net mailman:
The Ecartis incarnation is named "ecartis" and is formed the same in
master.cf as indicated above, but the file is located in a different
spot on my filesystem, and a different domain points to that entry.
I've made some basic changes, the obvious ones being from the mailman
executables to ecartis, as well as to how Ecartis handles list names and
addresses as well as command flags. They don't seem to be enough at
this point. Furthermore, my lack of Python-fu helps matters even less.
Now, with all that said, what's a good way to debug a persnickety pipe
transport when all that shows up in the log is a "temporary error" once
all the basic stuff the script CAN report as written has been taken care
of? What kind of facilities can Postfix offer in terms of aiding in
debugging?
you should run the script manually (as the user that is supposed to run
it, not as root or as yourself) instead of via postfix.
Postfix-specific stuff aside, what's a good basic reference for a Python
n00b? I am even tendering the thought of just ditching this thing in
favor of writing a script in a language I do know, Perl.
if you are more confortable with perl, use perl. here is an example
using a shell script instead. This example assumes addresses of the form
[EMAIL PROTECTED] (i.e. the domain part is not
used) and it calls the list manager as '/path/to/mlm listname listcommand'.
== transport
lists.example.com listmanager:
== master.cf
listmanager unix - n n - - pipe
flags=FR user=${listmanager_user}
argv=${listmanager_command} ${user} ${extension}
== main.cf
listmanager_user = mailman:mailman
lismanager_command = /usr/local/bin/listmanager.sh
owner_request_special = no
== listmanager.sh
#!/bin/sh
list="$1"
command="$2"
mlm=/path/to/mlm
if [ -z ${command} ]; then
command=post
fi
# we pass unknown commands to -owner. it is too late to bounce
# use access maps to reject unknown commands.
case ${command} in
admin|bounces|confirm|join|leave|owner|request|subscribe|unsubscribe|post)
;;
*)
command="owner";
esac
#echo "cmd=$command, list=$list" >> /tmp/listmgr.log
cat | $mlm "$command" "$list"
