From: Russ Allbery (rra@stanford.edu)
Date: Thu Dec 28 2000 - 16:59:29 CST
J B Moreno <planb@newsreaders.com> writes:
> I disagree -- I think of the message as consisting of 3 parts: header,
> body, and an optional footer aka the sig. This isn't how it is
> transmitted or received over NNTP of course, but this is how I, as a
> reader, deal with the articles and I think it's a useful way of doing
> so. If you look at it this way, then it is obvious that the sig
> shouldn't be quoted.
I will repeat: I don't want my news reader to distinguish between the sig
and the body when quoting. So it obviously *isn't* obvious, wouldn't you
agree? If it were really obvious, then people who are familiar with
Usenet and have used it for many years wouldn't disagree with you.
> Section 4.3.2 is (mainly, with the possible exception of the description
> of the sigdash itself, which I'd like to see moved to section 4.3.1) an
> "informational" on how to prepare a followup, and I see not including
> the sig as part of the necessary info (not as important as including an
> attribution, or correctly quoting using ">", but necessary).
I think this is a quality of implementation issue because it's a user
interface issue. I can possibly see a note, but I really would rather we
not get into the business of giving advice to news reader authors anyway;
there are better places than a standard to do that.
I don't think we should be telling people how to construct the attribution
either; that's also note material as far as I'm concerned *unless* we're
going to make it part of the protocol. Which I think we previously
discussed and decided against.
Saying that lines beginning with ">" are considered to be quoted material
is part of the protocol; it's saying something about the meaning of the
format of the body of a Usenet article. I agree with it because it can be
phrased as a *protocol* statement. Saying that the signature is not part
of the body and is separated from the body by "-- " would be a protocol
statement. Saying that the signature should not be quoted by followup
agents is not a protocol statement; there is no interoperability issue
here, other readers aren't going to misunderstand the message if the
signature is quoted, and given that there *are* reasons for quoting the
signature, it's clearly a user interface issue. Quality of implementation
issues don't belong in the protocol description.
> The final destination of all non-control messages is a person: and
> unless the current poster is commenting upon it, the sig of the previous
> poster is superfluous.
Superfluous is a value judgement. It's not the purpose of a protocol
standard to make value judgements, in my opinion. It's the purpose of a
protocol standard to define the meanings and purposes of things and to
establish their syntax so that multiple implementations can interact
without misunderstanding each other. Statements about whether news
readers should quote signatures do not fall into either of those
categories.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>