On 10 November 2015 at 17:45, James Cloos <[email protected]> wrote: > >>>>> "DC" == Dave Cridland <[email protected]> writes: > > DC> No. The reference identifier is *always* the service domain name and > only > DC> includes the hostname if there is secure delegation. See RFC 7673 ยง4.1: > > That does not match what we agreed on in the dane wg. > > Well, you're conflating the MX case with the SRV case and assuming they're the same, I think.
The SRV case requires the service domain always, whereas the MX case appears considerably more complex, but seems to want the service domain name, the hostname, and any incorrect aliases, with the hostname as the primary identifier and the others purely legacy: https://tools.ietf.org/html/rfc7672#section-3.2.2 This is desirable, because the essential service offered by an MTA is equal irrespective of the domain of the email, and it's possible (and sensible) to use multiple, concurrently pipelined transactions for multiple recipient domains. (And yes, this requires an interesting combination of ESMTP service extensions, but it is possible). In XMPP, and other SRV-based protocols, the choice of initial contact domain may radically alter the behaviour of the service from the outset. This is particularly the case with C2S, but in principle can be the same with S2S. Dave.
