Re: [Nmh-workers] Format function to create wrapped header lines?

[Ralph Corderoy]

Hi Ken,

Complainer here.  :-)

> I don't think anyone uses [References] as the complete, ultimate list
> of messages in that thread.  I suspect most programs that utilitze it
> only care about the immediate parent message.

No, you make use of all the message IDs in References because you may
not have received the immediate parent yet.  UUCP, 300 baud, overnight
low-cost calls, etc.  Or just that you were dropped from the CC for a
while.  trn's tree display that you referred to, and was excellent for
navigation, shows this IIRC;  the current message has a parent that's
unavailable, and a grandparent that is, for example.

> _If_ MUAs are following the standards, then a References header should
> contain a complete list of parent messages back to the origin message.

The Usenet RFCs allow for it to be shortened IIRC.  They were well used
to wars and "is this the longest thread ever" September posts.

> I suspect that all threading programs care about is the immediate
> parent, though.

https://www.jwz.org/doc/threading.html is one famous threading
algorithm, and explains history regarding the RFCs.  It certainly uses
more than just the immediate parent.

DJB also touches on shortening References.

    However, if there are more than about ten identifiers listed, the
    writer should eliminate the second one.
        — https://cr.yp.to/immhf/thread.html

> I recall Ralph had some reason he used References instead of
> In-Reply-To, but I don't remember what it was.

In-Reply-To's content is more ill-defined, especially historically, and
awkward to parse as more than one thing can look like a message ID.
Both JWZ and DJB touch on it.

-- 
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy