One addition to Ned's comments on this...
Eric Thomas (the LISTSERV developer) is aware with what is going on with
this. He has followed this list, at least intermittently. Ned is
quite correct that he can, if he sees it as useful, modify LISTSERV
far quicker than we can specify things. It is also important to understand
that Eric is convinced (probably correctly) that LISTSERV is *the*
dominant automated list explosion and redistribution tool in the world,
measured in terms of total message traffic processed per day, and that
he is somewhat irritated with us for having put MIME together without more
explicit and obvious attention to the problems it would cause him. The
major problems there are not the header-stripping--he deployed a
solution to the header-stripping problem (and some other behavior that
was undesirable from an Internet standpoint)-- but issues about
specification of character sets (and other content-type information)
when MIME messages pass through gateways that may perform character
translation.
As we have discussed before on this list, these problems can be solved,
but picking the combination of a good solution and one that will work
well in a network environment in which there are hundreds of gateways
between the NJE/BSMTP and the TCP/SMTP environments -- gateways on
different platforms and with different degrees of maintenance -- is not
easy.
In any event, LISTSERV's defaults were chosen to work well in the BITNET
environment, where fairly weak UAs predominate (at least at the time the
choices were made), transport bandwidths were historically low, and the
accumulation of junk headers is really not appreciated by users and
sites. And it is very much constructed on the model of acting as an
agent for the list owner: the agent receives the message, sets up header
information for its convenience, and then sends the results back out.
IETFHDR is the alternative to this if one wants behavior that we see as
more normal on this side of the boundary.
If a receiver wants to receive MIME messages, then he/she
must set IETFHDR for him/herself, for the particular mailing list on
that server.
This is not quite true. As a LISTSERV list owner, if I expect MIME
traffic on my list, I can set IETFHDR for everyone receiving it. Some
of them will be very surprised, and will not consider MIME to be worth
the trouble, but that is another issue.
I think the appointment of a small, informal, and patient ad hoc
committee to have a serious discussion with Eric about what he needs
from "us"--what he might do and what we might do, even at this late
date-- would probably be productive. I think that changing MIME to
accomodate what we might think would be good for LISTSERV without asking
him would be guaranteed to raise the temperature of the discussions and
not guaranteed to actually solve the problem.
To take one specific example, I suspect that it would not be terribly
difficult technically for Eric to provide a "SET MIME" capability that
would deliver somewhat less header information (and might involve fewer
other departures from the LISTSERV norm). But I'd be a little surprised
to see him doing it unless we can reach fairly clear agreement with him
about what is needed (i.e., what headers, specifically, need to be
retained and how they should be handled).
john