IMAP definitions on threading

[John Stanley <[e-mail]>]

Harald Tveit Alvestrand <harald@xxxxxxxxxxxxx>:

>John, I believe you are using language stronger than is justified.

I see nothing improper in my language. 

>If you check out draft-ietf-imapext-thread-13 (the draft is expired, 
>unfortunately), you will find the method called ORDEREDSUBJECT referred to 
>as "threading" -

And obviously since it is in an IETF draft it is gospel. I shudder to
think how many patently wrong things have appeared in USEFOR IETF draft
products in the past, but apparently I should assume that every other WG
never says anything daffy in their drafts.

>... the fact that you do not like this particular name for this way of 
>ordering messages does not mean that you have the right to claim that it 
>"does not exist".

First of all, I did not claim that "that way of ordering articles" does 
not exist. I've been pretty clear that I understand that it is done, and 
that we have no control over it being done that way. So, implying that 
I've said "it does not exist" is nonsense. 

But then, what part of "the Subject header contains NO article
relationship information" do you see as saying "I don't like this and thus
it does not exist?" If you can read paragraphs that point out obvious
technical limitations in calling something "theading" and pretend that the
only argument being presented is "he doesn't like it", then there is a
problem.

Perhaps you'd like to disprove my statement? Ok, here are four subject
headers. Please put them in the correct threading order, using the
"threading information" you find in those headers. Be explicit, and 
describe exactly what threading information you base your decision on.
You may open your test booklet and begin now.

1: Subject: A question
2: Subject: Re: A question
3: Subject: Re: A question
4: Subject: Stop asking stupid questions

Warning: it's a trick question. You cannot answer correctly. But then,
that's the point. There is no data to generate the right answer. The 
Subject header simply does not allow threading, especially not the 
identification of specific threads that someone here claimed was possible. 
Sorting, yes. Threading, no. And not just "because I don't like it", 
because the RFC doesn't support it. 

Oh, wait, that was synthetic, so it obviously isn't like real-world data.
Ok. Here's four real subjects from a real newsgroup. Put them in the
proper threaded relationship. I'll make it easier, I'll make it multiple
choice, but you still have to show your work. If you simply answer "B"  
without supporting work, you get no credit. "Process of elimination" is 
acceptable.

1:        how does ByteLoader work? 
2:    Re: how does ByteLoader work? 
3:    Re: how does ByteLoader work? 
4:    Re: how does ByteLoader work?  

A: 1->2->3->4
B: 1->2, 1->3, 2->4
C: 1->2->3, 1->4
D: 1->4, 4->2, 2->3
E: 1->4, ?->2, ?->3  (? is "article not listed above")

And the correct answer is: