From: John Stanley (stanley@peak.org)
Date: Thu Feb 05 2004 - 13:14:33 CST
Seth Breidbart (sethb@panix.com):
>Yes, the threading is almost correct. That's the best that the
>newsreader can do with the limited information it has.
You missed the point, which was continued in that paragraph. There is
no "almost correct" threading for something that is not an article.
We do not deal with non-articles in this standard, other than to define
what IS a valid article. We do not define how to process non-articles,
we say they are to be discarded.
>That's much better than threading them completely
>incorrectly, which is what using only the References: header would do.
Using the References header is the ONLY way to thread articles, by
definition. An article that does not contain a References header is, by
definition, not part of a thread (other than as the parent of a new
thread). It is not a followup to any other article, by definition, and
thus is it incorrect (not "almost correct") to thread it with another
article.
I understand your concern about missing References headers. However, at
the point you start treating something as an article (by displaying it in
your news reader, e.g.) you have to deal with it as an article. That
means that the threading is given by the References header, if present,
and that the article is NOT a followup if it does not have one. That's
the natural result of saying that References is MANDATORY in followups.
That's the defining statement for a followup.
To take non-articles and pretend that some of them are followups is
outside the scope of this draft, since we are dealing with articles. To
claim that we must have kludges in the standard to deal with non-articles
is silly. How many other non-articles do we need to make allowances for?
Do we want to have a kludge that tells people that an "article" lacking
a newsgroups header should appear in any newsgroups that are listed in
the Followup-To header? That seems like the right place to put such
non-articles, if we are going to tell people how to deal with
non-articles. How about articles without Date headers are displayed
with the NNTP-Posting-Date instead? Articles without a From should be
scanned for a signature that contains some identifying data, so that
can be displayed to the reader as the From header?
>We don't _want_ to play guessing games, sometimes we _have_ to play
>guessing games.
And when is one of these times we have to play guessing games? No, not
just "have" to play them, "_have_" to play them? Certainly not just
because your client won't show you an article sans References header as
part of another thread when its Subject header is similar to another
article's, I would hope. We've covered the "Re: help" subject, and I'm
sure you've seen discussions where two widely different threads have
similar subjects. "Similar subject" is not how this standard defines
threads.
>If the display of articles isn't part of the news system then people
>will care even less about the news system than now (if that's
>possible).
If they don't care about the news system, then they are free not to use
it. This fear that they will care less is not a reason to lump display
issues in with true protocol issues.
>>> So what? It's useful for clarifying the _meaning_.
>>
>> Except it does not do so.
>
>Yes it does. It just doesn't do it _perfectly_.
>> According to this clarified meaning, tell me: does the
>> subject-content "Re: foo bar nockers" have a back-reference or not?
>It probably does.
There has been no clarification of meaning here. "It probably does" is
not any clearer than the "it probably does" you would have if we say
nothing about "Re: " in the standard. Guesses is guesses, no matter how
much you wrap them up in fancy paper and put gaudy ribbons on them.
The correct answer is: if the article does not contain a References
header, then by definition the Subject does not contain a back-reference,
since there is nothing to back-reference to, even if the Subject starts
with the characters "Re: " which are defined to be back-reference. If it
does contain a References header, then it might contain a back-reference
or it might not, but we don't have to guess based on that, we know the
article is a followup and to what article it refers. For the former case,
we know it isn't a back reference, and for the latter the information is
redundant. We gain no information. Nothing is clearer.
>> How do you know?
> I'm just guessing.
You are free to guess. This standard isn't being written to define how
you guess at things, it is written to define how you know things.
> So magically all postings will conform once we publish.
So magically all clients will stop inserting "Re: "? Of course, we
cannot "magically" do anything, so there will still be non-articles that
are being treated as if they were. We can do no better than say "discard
them", else we get into a deep maze of twisty passages all alike trying
to define how undefined and unknown headers are supposed to be treated.
> Yet most _people_ who read netnews want to see them.
Somehow I doubt that a majority of news users want to see any particular
article that lacks a References header when it should have one. And, of
course, since they are being treated as articles anyway, they will be seen
by everyone who wants to see them, so it doesn't matter if most or none
want to see any particular one. Nothing stops anyone from seeing articles
without References headers if they want to see them.