I was just reading s3cr3t0's messages over in Citadel Support and that raised a question and possible suggestions. What does Citadel do now when an e-mail delivery attempt fails? In the past, I've seen some mail servers (can't remember which ones) will send a notification e-mail back to the sender right away to let them know the delivery failed, but that the server will keep trying.
When the server finally gives up, it will send a notification failed e-mail: giving up.
whether or not the server _has_ to give up is dependend on the reply of the remote side.
temporary errors could be told to the user on a first temporarily failed attempt.
i.e. DNS or Connection failures are a little more complicated, since you can't differentiate between whether there is a facility error (i.e. DNS unavailable or a route broken) or the user just did a typo...
Which leads to a suggestion. Say that a user does send an e-mail to the wrong address. They get a notification right away that the delivery failed, but Citadel will keep trying. The user fires off another e-mail with the corrected address. This now leaves Citadel with a bogus e-mail that it will keep trying to deliver and eventually fail on. How hard would it be to provide a means for the user to cancel any e-mails that are sitting in the outbound queue?
some lines .js in webcit ;) which will make the better ui anyways.
I'm thinking that the first notification failed message could include a special field in the header with the original ID for the failed e-mail. The user would simply reply to the notification failed message; perhaps changing the subject to "cancel" or somesuch, and the original e-mail gets purged from the queue. Theoretically, there shouldn't be any security issues because the user first has to authenticate to access their inbox, and you'd need the notification failed message with the proper ID.
Of course, how likely is it that a user will care enough to cancel out bogus e-mails in an e-mail server's outbound queue? :P
Mail Binder
self maintainance would probably also be a good feature, but not realy easy to implement...
we have several places where the not knowing the URL of webcit is a massive drawback. some users don't even run webcit...
currently its rather a concern howto do rate limmiting so citserver doesn't overrun peers or even the DNS server when doing a massive delivery..
