On Tue, 7 Dec 1993, Dave Crocker wrote:
Trying to rationalize inter-personal and bboard message headers
sounds laudable. The difficulty is finding a way to do it that
is reasonable and doesn't disrupt the installed base.
The Message-ID proposal, in effect, relies on the development of
a heuristic. This is probably not a good idea. Heuristics as
add-on processing rules are fine. As standards they are terrible.
The other possible alternative would be to make Message-ID-s
mandatory, and enforce that requirement. Or have you a third
solution.
A possible third solution would be that intelligent scanning
of the parts of a message whose lines begin ""> " and comparing the
rest of the lines with previous messages could help a message
system to recognize reply links. But that is an even worse
heuristic.
For Followup-to/Reply-To, you are trying to distinguish "type"
of address from the header name. That is, you are trying to alter
an attribute of the message, to define an attribute of an address.
This isn't clean design. It's made worse by the possibility,
and probability that senders will need to send the same message
to individuals AND to bboards.
I don't have a counter-proposal and wish I did.
"Followup-To" and "Reply-To" are operators that modifies the
behaviour of two different common UA commands, the command
"reply to the author" and the command "reply to the whole
group". See in that way, there is no confusion.