From: John Stanley (stanley@peak.org)
Date: Thu Jul 01 2004 - 00:57:15 CDT
Since Charles has muddied the waters with his modified, mangled, and
abusively named versions of the References text, I decided to look up the
actual proposals and their source.
From draft-ietf-usefor-article-13.unpaged, which has been through two
last calls in this group already:
A followup MUST have a References-header, and an article that is not
a followup MUST NOT have a References-header.
(This is what Charles erroniously and misleadingly calls
"References-John". It appears with essentially the same meaning in
RFC1036: "It is required for all follow-up messages, and forbidden when a
new subject is raised.")
On Wed May 26 2004 - 05:49:37 CDT, Charles suggested:
A followup MUST have a References-header. An article that is not a
followup MAY have References-header if it has some other form of
dependency on an earlier article(s) (see, for example, section 6.21.2
for its use in conjuntion with "message/partial").
This makes no mention of "funny-followup", it clearly uses the terms
"followup" and "not a followup".
Under the current text, a "not a followup" is prohibited from having a
References header. This allows a reading agent to idenfity unambiguously
what is and is not a followup, for whatever reason it desires to determine
that information. Usually that is for displaying threading information; a
use that this group has already determined to be sufficient for the use of
RFC2119 language where it otherwise should not be.
Under the proposed text, a "not a followup" MAY have a References header.
Since a "not a followup" may have the same References header that a
followup has, the ability to determine "followup" and "not a followup"
is lost. Gone. Eliminated.
Notice that the preceeding two paragraphs are true no matter WHAT the
definition of "followup" is. They deal with two cases: followup and
not-a-followup. The change to the definition of followup has nothing to do
with this, nor does the "Re: " hack.
This change was suggested as a means of solving the problem of people who
uses References in articles which are not clearly followups as currently
defined by the standards. A much simpler solution (namely, changing the
definition of "followup" to fit the current practice) has been suggested,
but allegedly this causes problems elsewhere in the draft. I've seen none,
and the only one anyone other than Charles has come up with (that I
remember, anyway) has been dealt with.
The problem with References-Charles is serious, and occurs in an area
where we previously agreed to bend the rules and use language stronger
than would be permitted by a strict reading of the other standards. If the
the informatiomn contained in the References header is not needed, then
the excuse for the RFC2119 language goes away, as does the need for the
header in the first place. We would have no reason to make any comment
regarding References other than "See RFC2822".
You can't have one without the other. If we don't need it, get rid of it.
If we need it, it needs to contain the information that it is designed to
contain. I'm in the latter camp, which some people are trying to paint
as suggesting a change, but is really just continuing the status quo.
If someone tells you that this is a change I have suggested, you should
doubt the veracity of the rest of what they tell you.