Re: I-D ACTION:draft-ietf-imapext-thread-00.txt

[Paul Overell <listsub@xxxxxxxxxxxxxxxxxxx>]

In article <MailManager.965753074.4273.mrc@xxxxxxxxxxxxxxxxxxxx>, Mark
Crispin <MRC@xxxxxxxxxxxxxxxxxx> writes
>On Tue, 8 Aug 2000 10:26:23 -0500, Jon Miner wrote:
>> Just wanted to note (as has probably already been noted) that message
>> threading is better based on the "In-Reply-To:" header and the
>> "References:" header.  Using these fields makes subthreading much
>> easier.  This didn't seem to be mentioned in the document.
>
>We are working on the specification (and reference implementation) for the
>REFERENCES algorithm and have been for the past month.  Please be patient.
>
>What you should look at in the current document is the framework provided
>(e.g. new algorithms can be added at any time, representation of trees, etc.)
>

Will the REFERENCES specification deal with referenced, but unavailable,
messages?

For example, message C references B references A and, for whatever
reason, messages A and C are available but B is not available.

Then I would still want B to be represented in the thread structure -
otherwise it would give the false impression that C replied to A.

The syntax in the draft currently can only return sequence numbers or
UIDs whereas for message B all that is known is its message-id.

Suggestion:

If A is message 1, and B has message-id <x@y>, and C is message 3 then
the server response would be

   S: * THREAD (1 "<x@y>" 3)

and the corresponding syntax

   thread-members    = (nz-number / msg-id) *(SP (nz-number / msg-id))
                       [SP thread-nested]

   msg-id            = string


Regards
-- 
Paul Overell                                             T U R N P I K E