Source routing -- why not?
John Levine <[email protected]>
| Newsgroups | gmane.ietf.mxcomp |
|---|---|
| Organization | I.E.C.C., Trumansburg NY USA |
| Message-ID | <[email protected]> |
After further deliberation about source routing and SPF, I have come around to the conclusion that Frank is to some degree right, and if you want to use SPF/Sender-ID, you should use source routes. SPF, after all, simply reintroduces source routes into Internet mail, a decade or so after they disappeared. SPF and Sender-ID assert that some paths for a particular message are valid and others aren't. Well, OK, that's just what source routes do. In particular, SES and its ilk are just clumsy recreations of source routes. If a@a sends mail to b@b which then forwards it to c@c, the return path on the second hop should be <@b:a@a> which says exactly what SES would, this message was from a@a but if you want to write back, you need to send your response via b. Host b needs somehow to remember that this particular relay is OK, but that's not new. RFC 2821 says that source routes are deprecated except in unusual circumstances, but MTAs should handle them if they see them. Relative to the past 20 years of e-mail, I'd have no trouble calling SPF unusual. Regards, John Levine, [email protected], Taughannock Networks, Trumansburg NY http://www.taugh.com