"Julian" == Julian Mehnle
"RE: Architectural issues with the SPF specification"
Mon, 10 Jan 2005 13:58:34 +0100
Julian> I know what "overloading" means (or I would have asked
Julian> "what _is_ overloading?").
Sorry, I misread your question. I did not mean to be gratuitous but
I'm afraid that in retrospect it looks as if I were. My apologies.
Julian> I just wasn't sure what aspect of my setup you considered
Julian> "overloading". That I use "mehnle.net" as a domain name
Julian> as well as a host name?
Yes.
Julian> If that's what you meant, then it is because the DNS RFCs
Julian> don't make any distinction between the two that would
Julian> contradict my view. To also be a host name, a domain name
Julian> just has to have an A record (and satisfy some syntax
Julian> requirements). There is no deeper concept of "host name
Julian> vs. domain name". The host name "io.mehnle.net" is still
Julian> a domain name, too, even if you don't want it to, and even
Julian> if there are no further hosts in the "io" sub-domain.
I dunno whether it is deep or shallow but in ordinary conversation it
seems to be useful to let the name distinguish between to domain and
the hosts therein. Does the utterance: "mehnle.net is broken" refer
to one computer or to the container? Before the rise of "the web"
folks would have almost without exception taken that to mean the
container. As I have mentioned, if a domain preserves the naming
distinction between the container and the contents, or as I have been
putting it, the naming distinction between the collective and the
individual, it is easy and natural for that domain to recognize as
fabrications those Received header fields that claim to have received
mail from a host using the domain name alone as the argument to the
HELO command. When such a domain MXes its domain name to specific
hosts, and even when envelope sender addresses are coerced not to
reveal host names, these fabrications are sufficiently numerous to
make this a worthwhile check, at least in my experience.
For these reasons, to preserve sane human conversation and to preserve
a useful criteria for recognizing forgeries, I question the
advisability of explicitly encouraging naming practices that do not
preserve the naming distinction between hosts and domains.
If questions remain please mail me directly. This is only a minor
point and I do not think it warrants further discussion on
spf-discuss.
jam