If this header field is not meant to be used internally between
Internet mail UA-s, does this mean that a sender should send
two copies of his message, one to the X.400 gateway which has
the "Obsoletes:" header field, and one to Internet recipients
which does not use this header field?
As Keith points out, it is simply impossible to implement such a rule since in
general there's no way to predict the characteristics of the final destination.
However, none of this should be necessary, as I explained in a previous
response. While the use of Obsoletes: in this way on the Internet isn't
recommended or required or anything like that, it doesn't change the fact
that both the syntax and semantics of this header are clearly defined in a
standards-track RFC. This means that nobody is allowed to come along and
use it for a conflicting purpose.
This doesn't mean that this function will be exclusively served by this header
on the Internet, however. Someone could come along and define a different
header with similar or even identical semantics for use in Internet mail and
even recommend or require its use. This is why I'd like to see the
Obsoletes/Supersedes issue dealt with. I suspect the right answer is to cut
over to using Supersedes instead of Obsoletes -- there is a heck of a lot more
use of news systems and gateways than there are X.400<->Internet gateways.
(Note that I speak as someone who will then be forced to change a bunch of
code -- this is not to my advantage, but I think it is what's best for the