[Top] [All Lists]

Re: Implications of MIME for Transport

1992-04-15 11:10:18
Steve, Marshall, et al --

It should be clear to us all that choosing something this is arbitrarily
incompatible is stupid!  This goes for INTERNET vs {X.400, MSMAIL,
cc:MAIL, Novell MHS, DaVinci, etc, ad nauseum}.

Our analysis has to look at what all installed base systems do in order
to see what we can do to match up with them to some desired degree of

In doing this, we need to also decide some things about exactly what
semantics we assign to things like "delivery" and "receipt" so that our
notifications will be consistently meaningful in all cases.

I suggest that we start with the discussion of what we want these things
to mean, regardless of how some existing implementation might happen to
define them now.

In this regard, it is my opinion that X.400 has correctly defined the
semantics of these terms.  This opinion has nothing to do with what
X.400 has done with its notification PDU definitions, or what RFC1148ter
has done with the mapping from X.400 Notification specifications to
RFC822 body text.  Whatever RFC1148ter did, it was dealing with the
total lack of any defined notification format or semantics on the RFC822
side.  We are now approaching an effort to define the RFC822 side, so we
need to rethink the whole RFC1148ter specification of this stuff.

But first, we have the task of discovering what the existing installed
base does and then deciding what semantics we wish to standardize, and
then decide how to represent the chosen semantics in RFC821/822 syntax.

In doing this, we must develop an INTERNET abstract model of the MUA/MTA
relationship and functionality, or we will find ourselves engaged in
endless arguments about this whole thing.  Perhaps this is a first
prerequisite while we are doing the investigation of how the installed
base does what it does.

I am very much concerned here that we (in the INTERNET) not adopt the
OSI attitude toward the installed base.  It is in our own best interests
to find ways to maximize interworking with the installed base (including
that which is not inside our current IP INTERNET).

As I have said with some frequency, every interesting connection has two
ends, and it does not matter where they are.  What we are trying to do
is to maximize the number of interesting potential connections.