> So we have two sets of recipients: one set in the MimeMessage, and another > carried within the MailImpl object. Without knowing the convention used > within JAMES, which we should document, there is no way to know > which set of > recipients will be used without tracing the code all the way to > Transport.send[Message]().
Recipients as an argument to sendMail() is used, or the recipients from James' MailImpl. recipients in MimeMessage will be used After the message leaves James (eg downstream MTA's) One way of using the transport overrides the > recipients in MimeMessage, the other doesn't. As best I can tell from > RemoteDelivery, the convention in JAMES is to put the real recipients in > MailImpl's collection, allowing the MimeMessage to have "display names" > (e.g., a list name). > > If this understanding is correct, Sort of, the "recipients" are used in the SMTP RCPT TO: command, the MimeMessage recipients are placed in the message header. Therfore there are reasonable grounds for having two sets (Bcc'ed mail for example) >my earlier interpretation is wrong. > Either way, we really should document the convention and clean up > the code. > Furthermore, if I am correct as to how we are using the recipient list(s), > I'd like to suggest that the MimeMessage's recipient list be used as the > real list unless MailImpl's collection is not null. That would eliminate > the potential for error and cut down on redundant data. And it is a > backwards compatible change. > > FWIW, the sample configuration in the Javadoc still refers to > NotifyPostmaster. Why should it not? d. -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
