> A configuration error that lasts 20 days is more than just a configuration
> error problem.
Probably. And the address is kicked off the list after that time.
It's quite common, however, for someone to make a configuration error on
Thursday that doesn't get fixed until Monday. If you send a warning
message on Saturday or Sunday, it'll bounce too. Not helpful.
> But I thought you were only saving the first bounce, so you must be saving the
> missing messages somewhere else, and that has got to take up SOME disk space.
Of course---they're in the archive. Recording the _numbers_ missed by a
particular user is no big deal.
> > Bouncing messages because of a _temporarily_ exceeded quota is a bug.
> Um, define temporary here. 5 minutes? 30 minutes? 20 days?
Anything under the local timeout. 65 hours is about the minimum possible
timeout before users start screaming; 120 hours is more common.
> OK, so if I'm keeping a copy at my main account, what do I need the list
> of missing messages for?
You don't. There were three situations, see, numbered 1 and 2 and 3.
They were _different_ situations, illustrating _different_ features.
Situation #1 showed why the probe was useful. Situation #2 showed why
the warning was useful. Situation #3 showed why the warning/probe
separation was useful.
> Can't you be informative or instructive without insulting me, too?
Can't you avoid spouting nonsense about things you don't understand?
You were faced with something surprising, something beyond your
experience---namely, that VERPs handle every bounce message reliably.
You had no clue what VERPs were.
You should have expressed your surprise and asked how VERPs worked,
But you didn't. Instead, you baldly claimed that the surprising fact
wasn't true. You claimed that VERPs depend on support from other MTAs.
Don't blame me for your mistake.
Put an end to unauthorized mail relaying. http://pobox.com/~djb/qmail.html