"Dennis Oelkers" <oelkers / zrz.TU-Berlin.DE> schrieb im Newsbeitrag
news:c9ndoc$qa5$1 / mamenchi.zrz.TU-Berlin.DE...
> Hello Robert,
>
> Robert Klemme wrote:
> >>- Add an "In-Reply-To"/"References" flag to the header by using the
> >>   message-id of THAT mail => the message would appear on c.l.r., but
> >>   threading would be broken
> >
> >
> > That's my favorite.  It would be even better if the GW could figure
the
> > correct message id and insert that.  But with these headers it seems
quite
> > impractical.
>
> This is not possible because in cases like this there is no real clue to
> which thread/posting this is a reply to (without utilising human
> intelligence or complex algorithms of course).

That's what I figured.

> >>- my solution: as mua's are more or less allowed to do whatever they
> >>   want my fear is that we get more and more of such borderline cases
> >>   so the only solution is that we establish identical posting
policies
> >>   on both the mailing-list host and the gateway/nntp host we're
posting
> >>   too. (ML-Maintainers? Any Comments on this one?)
> >
> >
> > I guess this is impractical since it sounds like this would rule out
some
> > mail clients.  People will not be happy about that.
>
> The point is that the policy of ruby-talk is quite sleazy whilc
> they're very strict for Usenet postings. People can send almost any
> garbage to the mailing list if they're subscribed to it and allowed
> to post. In my opinion it is the right behaviour to drop any mail
> which is not well-formed as soon as possible.

Well, but look at it from a user's perspective who is ignorant of the news
group: he has a mail client and obviously that mail client sends valid
mails (otherwise a whole lot other instances will reject his email).  So
he's likely reluctant to change his MUA just because of ruby-talk.  I know
people are peculiar when it comes to their favourite mail reader, news
reader, editor or whatever.  Just look at the tons of
my-operating-system-is-better-than-yours flame wars...

> This leads to the conclusion that we'll either have some sort of
> inconsistency between those two medias, or we would have to synchronize
> the policies up to a certain point where those inconsistencies converge
> against 0.

I could live with thread inconsistencies.  At least we have all messages
of a thread available - even if not properly sorted.  As said, that's my
favorite solution.

Kind regards

    robert