earle@jplopto.uucp (Greg Earle) (01/06/88)
(Pardon me if this is explicit in RFC-822, but my copy of that is in a storage locker right now awaiting my impending move - sorry) Does a bang path vs. use of `%' imply any connectivity information? In other words, I have more-or-less been assuming that a `%' in a path implied IP-level connectivity between the primary host and the secondary; e.g. `foo%bar@blat.Some.DO.MAIN' implied an IP-level link between `blat' and `bar'. Maybe even implying SMTP service between them. I also always assumed that a bang path implied UUCP level connectivity (at least), and possibly IP-level as well (e.g., those sites running 4.3BSD UUCP over TCP/IP links), but that part not implied. But recently I've been seeing things like roy%phri@uunet.UU.NET which threw me for a loop, as I assumed that phri had a UUCP link (perhaps over Telenet) to uunet, but not an IP-level link. I assumed that this setup was definitively described by `phri!roy@uunet.UU.NET' and not with `%'. Something like `roy%phri.UUCP@uunet.UU.NET' would make more sense, I can parse this into someone (perhaps on the Internet) wanting to get mail routed to phri and deciding to let uunet handle the `smart' routing on the UUCP side, and thus merely specifying the final destination in Internet form. But `roy%phri' alone is quite a different matter; thus my confusion. Can someone speak authoritatively to this issue? Greg Earle earle@jplopto.JPL.NASA.GOV Indep. Sun consultant earle%jplopto@jpl-elroy.ARPA [aka:] Rockwell Science Center earle%jplopto@elroy.JPL.NASA.GOV Thousand Oaks, CA ...!cit-vax!elroy!smeagol!jplopto!earle
blarson@skat.usc.edu (Bob Larson) (01/06/88)
In article <5165@elroy.Jpl.Nasa.Gov> earle@mahendo.JPL.NASA.GOV (Greg Earle) writes: >(Pardon me if this is explicit in RFC-822, but my copy of that is in a > storage locker right now awaiting my impending move - sorry) > >Does a bang path vs. use of `%' imply any connectivity information? No, since neither mean anything to an rfc822 mailer in the local part of an address that contains a '@'. In local addresses, your mailer may do anything it wants. (Including interpreting port of the local address as another system to send the mail to.) The "<@foo,@bar,@belch:user@system>" type of addressing might be usable if rfc822 didn't specificly allow validating of all system names mentioned. (The <> are not optional, this is frequently gotten wrong.) -- Bob Larson Arpa: Blarson@Ecla.Usc.Edu blarson@skat.usc.edu Uucp: {sdcrdcf,cit-vax}!oberon!skat!blarson Prime mailing list: info-prime-request%fns1@ecla.usc.edu oberon!fns1!info-prime-request
roy@phri.UUCP (Roy Smith) (01/07/88)
In <5165@elroy.Jpl.Nasa.Gov> earle@mahendo.JPL.NASA.GOV (Greg Earle) writes: > recently I've been seeing things like > roy%phri@uunet.UU.NET > which threw me for a loop, as I assumed that phri had a UUCP link (perhaps > over Telenet) to uunet, but not an IP-level link. It's worse than that. Phri has no connection to uunet at all (I should know; I'm phri!postmaster). I assume you're talking about "Reply-To:" addresses on our articles; I just followed-up to Greg's article using the rn "F" command, and see "Reply-To: roy@phri.UUCP (Roy Smith)" in the outgoing headers. If somebody is turning that into "roy%phri@uunet.UU.NET", something is broken. It's not clear what path outgoing articles take from here. Our main news feed is cmcl2, but we broadcast local articles to philabs, allegra, and rutgers as well. -- Roy Smith, {allegra,cmcl2,philabs}!phri!roy System Administrator, Public Health Research Institute 455 First Avenue, New York, NY 10016
owens@vtopus.CS.VT.EDU (John Owens) (01/07/88)
The % sign in the local-part of an RFC-822 address can mean anything that the domain wants it to mean. (It could even be part of a username.) On most systems, something%somethingelse@this-domain means that the mailer for this-domain should send the mail to whatever it would take something@somethingelse to mean. If uunet wants user@host to imply user@host.uucp, that's quite reasonable. On one system I configured, user@host (or user%host@that-domain) meant that host in the local domain. In sendmail's delivery ruleset (S0), I would route the hosts via the appropriate protocol for each. Some would use UUCP, some SMTP, some DECNET, etc. In other words, it's all up to the destination. (and user%host is much nicer than host!user because there's no ambiguity of left-vs-right precedence). -John Owens Virginia Tech Communications Network Services OWENSJ@VTVM1.BITNET owens@vtopus.cs.vt.edu +1 703 961 7827 vtopus!owens
jbuck@epimass.EPI.COM (Joe Buck) (01/07/88)
In article <5165@elroy.Jpl.Nasa.Gov> Greg Earle asks about such
things as roy%phri@uunet.UU.NET and what connectivity they imply.
Greg, mail has evolved quite a bit since the days of RFC-822, though
there are many old mailers still around. "%" never implied any
form of connectivity -- it was often used for networks that speak
completely different protocols from SMTP and such.
For example, my address is jbuck@epimass.epi.com. If you run a
current Internet mailer, it will try to find the host, fail, and then
obtain an MX record telling it to pass the message to uunet.uu.net
(our official Internet forwarder). If your mailer can't handle MX
records, the equivalent "right thing" will happen if you mail to
jbuck%epimass.epi.com@uunet.uu.net. The % means "let the host after
the @ deal with the routing". By the way, you can usually tell if
you have an obsolete Internet mailer if it ever fails with the
message "no such host" for a host outside your own domain (I'm sure
that the mail gurus can come up with exceptions to this statement).
You don't need to know every host, only every top-level domain.
The UUCP world (at least that subset registered with the UUCP
project) now does routing. I generally use @, not !, in my mail.
So, neither @ nor % implies any form of link-level connectivity.
Because many UUCP sites attempt to reroute ! paths as well, even
! does not necessarily imply connectivity, though I wish it did.
--
- Joe Buck {uunet,ucbvax,sun,decwrl,<smart-site>}!epimass.epi.com!jbuck
Old internet mailers: jbuck%epimass.epi.com@uunet.uu.net
Argue for your limitations and you get to keep them. -- Richard Bach
rick@seismo.CSS.GOV (Rick Adams) (01/07/88)
Somehow the From: line arrived at uunet as roy@phri. uunet mapped that into roy%phri@uunet.uu.net which is a valid address. UUNET is smart enough to figure out if the site is a uucp site or an arpa site, so in fact the mail will get delivered. There is nothing wrong with that address. However, it would be nice to determine where the domain got stripped. --rick
honey@umix.cc.umich.edu (Peter Honeyman) (01/07/88)
In article <1804@epimass.EPI.COM> jbuck@epimass.EPI.COM (Joe Buck) writes: >If you run a current Internet mailer, it will try to find the host, >fail, and then obtain an MX record telling it to pass the message to >uunet.uu.net (our official Internet forwarder). backwards -- look for MX before A. see rfc974. incidentally, the orignal message asked if anyone could speak authoritatively on the subject. the answer is "no, but watch 'em try." sorry, couldn't resist. peter