From: Eivind Tagseth (eivindt@multinet.no)
Date: Tue May 04 2004 - 03:30:50 CDT
Here's yet another variant, just to see if this makes it easier to reach
a compromise. See my annotations in brackets:
The semantics of the Subject-header for a followup-article are the same as for
a normal article.
[
I'm worried that Charles' current wording may be misinterpreted to mean
that copying and prepending with "Re: " is the only valid subjects, and
thus not let the user manually change it to whatever is desired.
I can't see how someone will actually think this, but there seems to
be a lot of thoughtlessness in todays newsreaders. If noone agrees,
remove it, or move it to USEAGE.
]
However, since the topic discussed in a followup-article
is usually the same as in the precursor, the followup-agent SHOULD by default
automatically suggest to the user a Subject content that is copied from
that of the precursor's Subject, possibly preceeded by a short prefix-string,
referred to as a "back-reference". The back-reference is used to indicate
to the end-user that this article's topic is the same as the predecessor's.
[
I'm trying to say what a "back-reference" means here, so that the rules
for the back-reference only applies to the back-reference. This means
that other initializations of the Subject, such as done by AI-technology
are still legal, and so are tools for posting large multiarticle documents
as parts of the same thread. Did I catch the "true" meaning of the
back-reference btw?
]
The back-reference MUST be the exact case-sensitive 4-character string "Re: ",
and MUST NOT be prepended if such a string is already present. The choice
of adding a back-reference or not is up to the implementor of the
followup-agent.
[
I'd like to make it clear to the reader that adding back-references is
_optional_. I think Charles' text is a bit vague here, using the word
"possibly" without further explanation.
]
NOTE: The use of a back-reference is controversial, mostly because it is
forcing structured content into a non-structured header meant for
human interpretation only. There is no way to perfectly remove/ignore/
localize the back-reference, since there is nothing keeping the user from
starting the Subject with the same string as the back-reference, even if
it is not intended to have the same meaning as a back-reference (described
above).
[ Since I'm saying that the choice of adding the back-reference is up to
to the implementor, I think a few notes explaining the pros and cons
is necessary for the implementor to make the decision. This is my
attempt of summarizing the cons. ]
On the other hand, it is current practice to include a back-reference,
and some reading agents and users choose to interpret the existance or
non-existance of a back-reference as an indication of subject-change.
Not adding a back-reference may cause problems for such, arguably broken,
agents (broken, because in most situations the information they read
out of the back-reference is better and more precisely read of of other
headers, such as the References header and the Subject headers of the
predecessors).
[ And this is my attempt of summarizing the pros. ]
I hope this input makes it easier to reach a compromise. At least it is
(obviously) a text that I could "live with".
Eivind