From: Bruce Lilly (blilly@erols.com)
Date: Sat Mar 27 2004 - 16:04:21 CST
Charles Lindsey wrote:
> In <4060FB56.5090603@erols.com> Bruce Lilly <blilly@erols.com> writes:
>
>
>>Charles Lindsey wrote:
>>
>>>In <405E8B63.6030900@erols.com> Bruce Lilly <blilly@erols.com> writes:
>>>
>>>
>>>
>>>>For more insight into what In-Reply-To adds, consider
>>>>http://users.erols.com/blilly/mparse/usefor/ref+irt.png
>>>>and
>>>>http://users.erols.com/blilly/mparse/usefor/ref-only.png
>>>
>>>
>>>Your examples are interesting, but your interpretation of them is
>>>incorrect.
>
>
>>You're being a bit presumptuous, since I came up with the example.
>
>
> But you failed to state how the References header was to be constructed in
> your presumed extension to the existing protocols.
I have; moreover it is largely irrelevant.
>>RFC 2822 does explicitly address the case, but stops short of giving
>>rules for constructing References. In the example, each message's
>>References field has the following properties, which are consistent
>>with the rules for a response to a single message:
>>1. Each message which is an ancestor is mentioned
>>2. The last message-id is one of the immediate predecessors
>
>
> Yes, that is one set of possible properties.
#1 is necessary, #2 is not, but conforms to assumptions implicit
in the current Usefor draft.
>>Message-ID: <z>
>>References: <a> <q> <c> <r> <d> <s> <h> <t> <i> <u> <k> <j> <w> <v> <l> <x>
>>In-Reply-To: <l> <x>
>>Edges: 1 3 7 11 15; 1 5 9 11; 5 12 15; 2 4 8 14 16; 2 6 10 14; 6 13 16
>
>
>>where each semicolon-separated sequence in Edges lists a path
>>through directed edges of the corresponding message-ids in the
>>References field. So a is an immediate predecessor of c, c is
>>an immediate predecessor of h, etc. That contains all of the
>>information necessary to construct the entire dependency graph
>>without the need to examine any other messages.
>
>
> Indeed so. But the whole point is that it is not necessary to construct
> the entire dependency graph in order to provide an acceptable presentation
> of articles within a newsreader.
It is if the definition of "acceptable" includes conformance to
reality rather than some arbitrary presentation.
> The Zawinski algorithm already does it
> pretty well assuming a fairly crude extension to the rule for contructing
> the References header (and would do it even better with a small tweak to
> fix that bug).
The IMAP sorting draft does not address construction of References fields
at all AFAICS.
