Broken list threads

[David Woodhouse]

On Sat, 2011-04-02 at 17:05 +0300, Vangelis forthnet wrote:
> On Thu Mar 31 2011 21:53:51 BST , Shevek shevek at shevek.co.uk wrote:
> 
> >Anyone have any idea why emails from a couple of people [Richard
> >(richard at richsim900.plus.com) and Vangelis (northmedia1 at
> >the.forthnet.gr)] break the threading?
> 
> Hi, as far as I'm concerned, I must be doing something wrong (not 
> intentionally), because I'm not at all familiar with mailing lists...:-(

It's not really special to mailing lists; it's normal email behaviour.

Each message has a unique Message-Id: header; the one for your message
to which I'm replying is:
Message-ID: <E52A4F5065004D9D86CA03B225A94057 at vasonote>

Any reply to your message should contain an 'In-Reply-To:' header and/or
a 'References:' header which bears the Message-Id: from the previous
message.

If you look at the full headers of this message, for example, you should
see it says In-Reply-To: <E52A4F5065004D9D86CA03B225A94057 at vasonote> 

That's how a mail client (or archive) can show you the mail threading,
and know that my message is a reply to yours.

When you post a message which *doesn't* have an In-Reply-To: or
References: header, that starts a new thread instead of continuing the
old one. And the list software is actually set up to *reject* any
attempt to start a new thread if it has 'Re:' in the subject, because
it's almost certainly a user's mail client misbehaving.

> I am basically reading the mailing list from the archive (it comes in two 
> incarnations, i.e.
> http://lists.infradead.org/pipermail/get_iplayer/2011-April/date.html &
> http://www.mail-archive.com/get_iplayer@lists.infradead.org/maillist.html), 

> When I changed the list format to "arranged by thread" I became aware of 
> what Shevek is talking about...
> I am using Windows Mail, which is the native e-mail client of my OS 
> (WinVista SP2 x86), when I send an e-mail to get_iplayer at lists.infradead.org
> my mail starts a new thread...

If you just compose a new email to the list, that's expected.

That's the correct thing to do if you intent to start a new thread, but
if you mean to continue an existing thread, then it's the wrong thing to
do.

(Note that a lot of people seem to be lazy, and try to start a new
thread by hitting 'reply' on an existing one and changing the subject.
That is called 'thread hijacking', since the In-Reply-To: and
References: headers are still identifying the "parent" message in the
thread, even though the subject got changed.)

> What am I exactly to do to reply within an existing thread? I have not 
> properly subscribed to the list, would this mean that my inbox be filling 
> everyday with whatever the members of the list send? For various
> practical reasons (that I don't want to go into for now...), this is
> something I could do without...

It's easy enough to filter it into a separate folder rather than letting
it pollute your inbox, surely? I have *dozens* of mailing lists that I
almost never read, which are all filtered into separate folders so that
it's easy to reply to messages properly if I need to.

> For this specific mail, I selected the e-mail & thread Shevek started from 
> within the archive (version 1) & clicked on the e-mail address on top, 
> underneath the subject...
> the outcome of which I shall see after I hit "Send"...

This time it was correct; I suspect you used *my* archive. If you go to
http://lists.infradead.org/pipermail/get_iplayer/2011-March/001193.html
and let your mouse hover over the address at the top, you'll see that
the mailto: url contains '...&In-Reply-To=<AANLkTi...' which means that
your mailer *does* get the threading information right in your reply.

But the mailto: URL you get when you use the button on mail-archive.com
is broken; it doesn't include the In-Reply-To: header. And it even
bypasses my 'Re: in Subject but no References' filter, because it
doesn't include 'Re:' in the subject either!

As long as you use *my* archive to reply, and not the broken one on
mail-archive.com, you should be fine.

-- 
dwmw2