Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!INFOODS.MIT.EDU!KLENSIN
From: KLENSIN@INFOODS.MIT.EDU (John C Klensin)
Newsgroups: comp.protocols.ibm
Subject: Re:      Re:      re: e-mail and System/370 Systems Person
Message-ID: <9004122105.AA19599@lilac.berkeley.edu>
Date: 12 Apr 90 21:07:14 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Reply-To: John C Klensin <KLENSIN@INFOODS.MIT.EDU>
Organization: The Internet
Lines: 54

Stephen C. Woods writes:
>    The problem is that the people who are most likley to have the skills
>wanted here (System/370) are the most likley to be running mailers that
>wouldn't know an MX record from a RC11 (a 256K word (16 bit) disk drive).
>
>Please remember that not everyone in the world runs up to date software.
>(mumble cusss...) Even if they'd like to budget constraints may prohibit, and
>manpower constraints (or lack of source) may make rewriting unfeasable.

Stephen,
  I'm a lot more sympathetic to this problem than I often sound.  But the
bottom line is...
 -> Internet hosts (other than MILNET) have been *required* to run with
full DNS support for *years*.  "Required" doesn't mean all do, but it does
imply some obligations to do so or incur the "costs" of not doing so.
 -> Similarly, gateways are supposed to handle things so that the
conventions of the networks which they support can be adhered to.
user@domain is the only reasonable and appropriate address, on the
Internet, for a host for which 'domain' is valid.  If, in the transition to
mumblenet, a gateway has to transform that to something else, nothing that
I said should be construed to prohibit that.
  -> There are positive technical advantages -- again, within the Internet
-- to using MXs rather than explicit routes.

At some level, the "real" question is whether
 (i) The people who are running current and rule-conforming software should
be hampered, penalized, or inconvenienced because there are sites that don't.
   or
  (ii) The sites that run obsolete or non-conforming software (for whatever
reason) should be inconvenienced while the sites that conform get optimal
behavior.
  While I would not suggest that the level of inconvenience in the second
case should be made any worse than the situation requires,   I think it is
very difficult to make a convincing argument for the first.

   --john
    Klensin@MIT.EDU

p.s.: Please note that the following, which arrived in your message, is
usually cited as the classic example of an address form that no one knows
how to parse.  It will get you into a lot of trouble whose symptoms will be
people being unable to reply to your mail, and is worth fixing, whatever
that takes:
  >  From:         ucla-seas!scw%CS.UCLA.EDU@mitvma.mit.edu
It *is* one of the natural consequences of trying to source-route; what
probably happened here was that the bang-route "ucla-seas!scw@CS.UCLA.EDU"
arrived at the BITNET->Internet gateway "mitvma.mit.edu", which "fixed" it
into the form above, following the exact rules we have been discussing.
The problem is that there is no established precedence rule about ! and %,
so the above can be construed as (at least):
 (i) ucla-seas!scw at CS.UCLA.EDU via mitvma.mit.edu (probably what was
intended)    or as
 (ii) scw%CS.UCLA.EDU at ucla-seas via mitvma.mit.edu
-------