
bruce@hams.com wrote:
However, the main overhead of mailing lists is their overhead on _your_time_ to manage them, not the computer. Mailman is very good about reducing _that_ overhead, much better than ezmlm.
I'm a long-time majordomo user who's evaluating alternatives including ezmlm, mailman, and listar. I'd appreciate an explanation of how mailman is better at reducing the list manager's overhead than ezmlm.
Ezmlm's main advantage is that it uses the qmail "percent hack" to get
Actually, it's called "VERP" (Variable Envelope Return Path), not "percent hack". The "percenthack" qmail configuration variable turns on user%host@relay style relaying. But your explanation is, otherwise, accurate.
If my address bounces, the bounce mail goes to that address, a qmail default rule catches it, and ezmlm decodes bruce%hams.com and realizes that I've had a bounce. It doesn't have to look at the bounce message at all, just the address it goes to.
Right, and the neat thing about this from the user's point of view is that ezmlm keeps track of which messages each recipient bounces so if the problem is temporary, it tells them and they can retrieve the messages from the archive.
When I get sick of watching undecoded bounces go by, I'll add that "percent hack" feature to Mailman.
You really need VERP support in the MTA, otherwise you'll have to inject a copy of each message to each recipient. Right now, qmail is the only MTA I've aware of that does VERP. Venema wants to add it as an option to Postfix, though.
-Dave