Stefano Bagnara wrote:
Namely there is no call-back when the MAIL FROM or RCPT TO is invoked
by a remote SMTP server.
I agree this is a big missing piece.
I think we should provide a more fine API for this. Sometimes mailets
only care of connection data, sometimes of SMTP envelope data (sender,
recipients...), sometimes only of mime message header, sometimes of the
full body, sometimes of the mime structure of the body, sometimes
combinations of the previous.
I think we should provide a mean to make the container aware of this.
Now that even james started supporting fastfail operations (in protocol
handlers) it would be cool if we all could agree on something like
"mailets" for the in-protocol operations.
*nods* it would be wonderful if we could agree on this. At the moment
we've implemented it as a separate Interface. You register the listener
and when the emails come in, then you have a hook as to what you want to
do with it. As you said to "fastfail operations" (i love that phrase).
If the JAMES implementation is already doing this functionality; then it
sounds as if it should be relatively trivial to move this handler
functional out as an interface and include it as part of the official
Mailet API.
btw - i wasn't expecting a reply at the weekend, so thank you.
