From: Bruce Lilly (blilly@erols.com)
Date: Thu Aug 12 2004 - 16:32:31 CDT
Bill Davidsen wrote:
> I tried that, it seems only Shmuel liked it.An unambiguous definition
> followed by a policy of what should be a followup.
I'm not convinced that we need to define "a followup". We do need
to define the syntax and semantics of the References header field
(and should probably do likewise for In-Reply-To). Leaving aside
the syntax ("[RFC 2822]"), the semantics are those of indicating
relationships between messages (probably also "[RFC 2822]"). So
there are messages with neither References nor In-Reply-To fields;
those messages do not have any defined relationship to other
messages (though perhaps one could stretch a bit and say that
Supersedes and Control: cancel fields define a type of
relationship...). And on the other hand, there are messages with
either an In-Reply-To field or a References field; those messages
do have a relationship to other messages. There is no difference
in the way either set of messages is transported, stored, etc.;
the only difference is a convention for display in some user
agents. That's not an issue for either USEFOR or USEPRO, and
is debatable for USEAGE (is it a "high-level" display issue or
not?).