This isn't intended as a flame of Dan Bernstein, but I fail to understand how
the behavior he describes for ezmlm is much of an improvement:
> Results in sample situations:
>
> (1) Your account is removed. Messages bounce. Ten days later, ezmlm
> sends a warning, which bounces. Ten days later, ezmlm sends a
> probe, which bounces. ezmlm removes you from the list.
So it takes 20 days for a bad address to get kicked off? I sometimes have
100 messages in a day, so a bad address doesn't get kicked off the list
until I've sent out as many as 2000 messages that won't be delivered??
At the moment, I have my limit at 40 messages. This means that an outage
of a few hours generally doesn't result in kicking someone off the list,
but one of a half day or longer (depending on traffic, recently I'm running
about 50/day) probably does.
> (2) You briefly run over quota. Your MTA foolishly bounces a few
> messages. Ten days later, ezmlm sends its warning, including the
> list of missed message numbers. You retrieve the missing messages
> from the archive.
An interesting feature, but see below. It also assumes that users are
smart enough to know how to use a list archive. From my experience, maybe
1/3 of my subscribers can figure that out or bother to, and I think I have
a lower percentage of clueless users than some other people on this list do.
And I think a lot of system administrators would take umbrage at your
statement that bouncing messages because of an exceeded quota is foolish.
> (3) You forward copies of incoming mail to a summer account. At the
> end of the summer, the account is removed. Messages bounce,
> though you don't realize it. Ten days later, ezmlm sends a
> warning, which bounces, but you've been alerted to the problem.
> You fix the problem. Ten days later, ezmlm sends a probe, which
> doesn't bounce.
I'm lost here. If my mail is being forwarded to a bad address, how am I
alerted to the problem by ezmlm's warning? Or are you assuming that I
independently find out about the problem during that 10 day window and
correct it?
> Warnings and probes are sent from cryptographically secure addresses, so
> an attacker can't force you off the list unless he can watch your mail.
I'm no expert, but I don't think there's such a thing as a cryptographically
secure address for a bulk mailer. Please elucidate.
> All of this depends on qmail's VERPs, which reliably identify the
> subscription address and message number for every bounce message.
Which depends on all MTA's and MUA's that handle the message sending back
decipherable bounce messages that uniquely identify both the message and
the intended receiver. I haven't written a MLM, but I've worked with both
smartlist and majordomo, and there are a lot of situations that neither of
them can track down. What's your secret?
--
Mike Nolan
nolan@tssi.com
Follow-Ups:
References:
|
|
