Against Extensibility in MARID Records
OK, for expedience I will concede that XML is better suited to extending
the data in various directions, like a tree.
Margaret Olson gave some good examples of info that one might want to
put in an XML document describing a mail sender. I can see why it could
be useful to publish them, but I think it'd be a disaster to allow MARID
extensions like that.
One of the main points of MARID, as I understand it, is to develop
something that can be implemented relatively quickly and will
interoperate among senders and recipients all over the net. I don't
know what I'd do with descriptions of how much mail a domain sent and
for what purposes (starting with whether I'd believe those assertions,
which I probably wouldn't unless it was from a sender whose mail I was
going to accept anyway) but I'm sure that if I asked six people the same
question, I'd get six different answers. That's not interoperability,
I don't see any problem if the MARID record includes a pointer to a
place where all of the extended XML goop can be found, probably as a
URL. It's going to be big enough to need a TCP session anyway, and http
is the only TCP service I know that is well standardized and is known to
scale reasonably well via caches and the like. RSS shows us that it's
practical to fetch XML info records via http.
The MARID record needs to contain info that's universally understood.
I understand what "these IPs can send our mail" means. I don't
necessarily understand all the implications, but I understand that
assertion. Much beyond that, it's still all in the fog, and I don't
think any of us are well served by a spec that requires mail recipients
to download fog they can't use and don't understand. If people want
private extensions for, say, a domain to tell its own MUAs how to parse
the received headers, that's fine, but it's not needed to interoperate
among domains so I don't see that it belongs in MARID.
If we take all the extra semantics off the table, the simple SPF format
is quite adequate, and we can reasonably defer possible standardization
of the XML over http, probably starting with conventions for the URL to
ask about both the domain and the mailbox, until later.
John Levine, johnl(_at_)iecc(_dot_)com, Primary Perpetrator of "The Internet for
Information Superhighwayman wanna-be, http://www.johnlevine.com, Mayor
"I dropped the toothpaste", said Tom, crestfallenly.