9
Replies
The email system is, by its fundamental nature, an asynchronous delivery system. "Bounced" emails might not "return to sender" for many days, if at all. They will be accepted for delivery as-tendered, and if subsequent attempts at delivery (and re-delivery) are unsuccessful, they might be returned to you "somewhere." You must write a separate process, in some language-of-choice, that will look through the "in-basket" of the dead-letter user and attempt to harvest any failed-delivery notifications that might be there.
As an aside, be wary when you design this process! In the real world of the sometimes-hostile Internet, "failed delivery notifications" can be spoofed, e.g. in an attempt to shut-down a legitimate mailbox. (It's also sometimes used as a way to get your name taken off a junk-mail list.) You need to include some recognizable-to-you metadata in the outgoing message that will allow you to authenticate it upon return. You need to be able to recognize that "this is a piece of mail that I actually sent," and that "the send-to address has not been tampered with."
Most systems are designed so that mailings are sent only to e-mail addresses that have been tested in some way. For instance, by sending out an e-mail to that address with a "click me" link. Only if the user does actually click on that link will we assume that the e-mail address is probably good. Nevertheless we must be prepared that it may one day cease to be "good." ("Especially in today's economy" 
... but that's another story, eh?)
AdChoices