All this fits well with the design model that says that such
conversions are only done (explicitly or implicity) with UA level
authorization.
Oops. Well I thought we were discussing the behaviour of an MTA
that can't reach a 7-bit destination. I am pretty sure that the
decision at Atlanta was "convert" not "bounce" in that circumstance.
Greg was trying to keep the nested encoding in there because it is
a simpler encoding for a poor dumb MTA to do.
If conversion is not authorized, then the message should be bounced.
Well if there is to be a Conversion header then the default if there
is no header should be "Conversion: allowed", because the number of people
who will want their mail bounced rather than converted will be very close
to zero. We don't need a "Conversion: allowed" in every mail message.
With this observation, I belive wea re almost all the way back to
where we were in April, May, and June...\Stef
I think the position has moved to one with which the implementors are
more happy and I think that is very important.
Bob Smart