On Wednesday 23 April 2003 7:11 pm, James Mohr wrote:

On Wednesday 23 April 2003 01:51, Vince Littler wrote: <snip>

...

No, the point about a standard is not moot. If Mutt implements a certain something and calls it threading and Outlook implements something else and calls that threading, which is right? If I reply to a post and quote [and I make a _subjective_ judgement that the topic is changed] and I only change the subject line, then who is anybody to tell me that I have not changed the thread. And who is anybody to argue if I say that Mutt is broken because it does not recognise my new thread. Without a standard, there is nothing to argue about one way or the other, there is no such thing as a thread and this whole thread is a void.

<snip and change of context>

Take your original message as an example. It has the message ID: <200304230051.45646.suse@archipelago.eclipse.co.uk> In Anders' reply to you, there is this header: In-Reply-To: <200304230051.45646.suse@archipelago.eclipse.co.uk>

Now we get to the core of it. Firstly, there is some sort of data structure within email which supports threading. But secondly, this data structure is _only_ [partially] visible at the receiving end, but it is controlled at the sending end, where it is not actually normally 'visible' to the sender as part of his Application Interface [ie email client]. My own take on this is that although the infrastructure is in place, the threading feature is not sufficiently mature for anything other than voluntary use because: 1] there is not sufficient agreement on the application of the feature [eg the issues raised by Anders] 2] hence there is no actual standard in place 3] hence there is no client support for a threading standard, particularly at the sending end [eg 'reply to current thread' and 'reply to new thread']. Now, if _all_ these issues were sorted out, I would feel that barking at newbies for violation of thread integrity would be justified. But let's face it, on the majority of clients there is no button to press or drop down dialog or anything to see what thread you are on or change it in any way - the user only has the reply button and the 'subject' fields to play with - I know, you told me how to do this - it is easy, but it is far from obvious. In other words, the problem is _not_ a User problem, it is a User Interface problem. As such, I think the constructive place for addressing the problem is with 1] 2] and 3] above. The hardest part is 1], but with this, 2] could be easy and a selection of clients supporting 3] would be totally possible. Until then, perfect threading will surely be a dream. And if Linux flourishes the way we would all wish, we are going to see some newbie top posters who are into some very Microsoftish ways... regards Vince Littler

On Wednesday 23 April 2003 8:59 pm, James Mohr wrote:

I disagree there. First, I would consider a decade or so of use to be "sufficiently mature". There *are* applications (i.e. mail readers the implement this feature and the fact that some applications do not take advantage of this feature does not mean the feature is not standardized. Should we all ingore HTML standards on our web pages simply because they are not implemented by Microsoft?

Well, if it ain't written down, it ain't a standard, regardless of the length of 'use', its age or even its shoe size. And what Christopher Mahmood says about ezmlm-idx indicates that there are multiple implementations trying to achieve the same end. As for HTML standards, I think that just makes my point. What we should do is ignore Microsoft practice [what they do is practice _not_ standards] and use w3c standards. Only if we are feeling kind, we should then detect MSIE and code specially for its limitations, but otherwise we should expect Microsoft to mend their browser. Similarly with threading. If there was a _published_ standard [with broad agreement and enough wrinkles ironed out to be broadly beneficial], I would want a client to support that standard. I would want SuSE to provide clients which complied [presuming on the broadness of SuSE's shoulders!]

Also there **is** support for threading on the client end. My email reader (Kmail) **does** support it. I can group my email messages by thread, as well as collapse or expand those threads. When I click "reply" or some variant, my email client rightly assumes that this is a continuation of the same conversation. If not, I click "Post to mailing-list" and it creats a brand new message which is independant of the orignal thread. I am certain that many of the email clients other people use have similar features. Just because they are not labled 'reply to current thread' and 'reply to new thread' does not negate them. In fact, in my mind having different labels confirms the "rightness" of them. You *reply* to the *same* thread, you *post* a *new* thread.

Your explanation makes total sense - once you know the bit about how messages are associated with threads. However, the problem remains that to most people 'thread' and 'subject' are the same thing and even in Kmail, 'thread' is not obvious as a message attribute in any sense other than the subject [particularly when you initiate the process of composing a message]. Solve that problem and I think that the instantiation of hijacked threads would halve and when the issue is pointed out in the remaining cases, people would be far more inclined to comply, purely through comprehension. This would only leave thread drift, which would be subjective, although godwinating a thread would always cross the line...

Vince, if it were obvious then I wouldn't have needed to point it out. Since many people do not use threaded readers they obviously do not *see* the benefits of threaded messages. That's why I pointed it out and explained what it means (and I really don't see how my post was "barking").

That's [guard dog] barking not barking [mad]. No, that's my perception of reactions to thread hijacking in general [through the ages], rather than of you. I think this has taken us somewhere constructive. regards Vince Littler

On Wednesday 23 April 2003 23:06, Vince Littler wrote:

Well, if it ain't written down, it ain't a standard, regardless of the length of 'use', its age or even its shoe size.

Oh, so it's acceptable to fart in an elevator? So, you have no problem with someone saying your wife/girlfriend/mother is fat? In any "society" there are certain rules and "standards". Those that are written down are "de jure" and those these are not are "de facto". A de jure standard ist you don't punch a loadmouth in the face (whether he deserves it or not). A de facto standard says you don't fart in an elevator. If you continue, people simple won't get in the elevator with you. It is a standard to say "please" and "thank you" in polite society. These are not written down, but still *de facto* standards. <snip>

Your explanation makes total sense - once you know the bit about how messages are associated with threads. However, the problem remains that to most people 'thread' and 'subject' are the same thing and even in Kmail, 'thread' is not obvious as a message attribute in any sense other than the subject [particularly when you initiate the process of composing a message]. Solve that problem and I think that the instantiation of hijacked threads would halve and when the issue is pointed out in the remaining cases, people would be far more inclined to comply, purely through comprehension. This would only leave thread drift, which would be subjective, although godwinating a thread would always cross the line...

Other than satisifying someone's need for laziness, what does real benefit does allowing people to hijack threads serve? There are many, many benefits for following netiquette and starting a completely new thread. Not the least of which is that more people will see your question and you will have a greater chance of getting an answer. Where is the benefit for the group (i.e. the mailing list) to accept that thread hijacking is okay? As far as I can see none. Do you have one? If you do not use a threaded reader, then you will not notice the threads. So, either way you see no benefit from threads. On the other hand, those of us that use threaded readers encounter problems when people hijack threads. Just like avoiding people who fart in elevators, continued behaviour that is unacceptable by a group as a whole or by individuals within that group will only alienate them and they may simply ignore you. There are many, very competent and profession people on this list who consider thread hijacking a breach of netiquette. They are within their rights to put a condition on the great support they provide for **free**, are are within their rights to simple ignore people who do not follow certain rules of behaviour. It's the same anywhere. Since we are not discussing any real technical aspects of this, I personally find the discussion pointless. I am simply pointing out that there has been a certain expected behaviour on the Internet for 15-20 years and has thus become a de facto standard. Many people consider it impolite. I simply pointed this out. If you personally choose to ingore this and behave in a manner that many consider impolite simply because it is not written down, that is your choice. <snip> Regards, jimmo -- --------------------------------------- "Be more concerned with your character than with your reputation. Your character is what you really are while your reputation is merely what others think you are." -- John Wooden --------------------------------------- Be sure to visit the Linux Tutorial: http://www.linux-tutorial.info --------------------------------------- NOTE: All messages sent to me in response to my posts to newsgroups, mailing lists or forums are subject to reposting.

On Saturday 26 April 2003 10:57 am, Graham Murray wrote:

Vince Littler writes:

...

Similarly with threading. If there was a _published_ standard [with broad agreement and enough wrinkles ironed out to be broadly beneficial], I would want a client to support that standard. I would want SuSE to provide clients which complied [presuming on the broadness of SuSE's shoulders!]

There *is* a published standard - RFC2822. The following extract expands

Found it at http://rfc.net/rfc2822.html and your extract is 3.6.4. - Identification fields. Excellent, this is what I think the whole threading issue needs [and looking back, I see that rex also pointed out this one]. And [at the risk of repeating myself] a client is required which supports this _fully_. As Carlos pointed out, both James Mohr and myself use K Mail. James thinks it is threading compliant, I don't - or rather I think the support is partial. All the complaints over broken threading come from the receiving end where the client visibly supports threading. There are NO complaints about "I cannot control the thread I am sending my message on" from anyone at the sending end, but this is where all of the complaints originate. Andre Truter said:

So, if we see 'hijacking' a conversation as rude and improper, then how can we justify 'hijacking' a thread? It comes down to the same thing.

Jim Norton said:

Some people will simply reply to a mail so they don't have to worry about knowing the email address to post a message to a list. They change the subject and throw in the text according to what they are writing about. <snip> I don't think people are neccessarily being rude, more like uninformed.

I think the reason there are no complaints is firstly that there are 3 conceptual candidates for threading [at least]: 1] Subject line 2] rfc2822 support 3] "normalizing the subject" as used for archiving this list And secondly when someone composes a message, they are only confronted with any choice over 1], which looks very much the same as the threading which they see at the receiving end. Suggestion: =*=*=*=*=*= So what is required is [talking KMail here] in the window for composing a reply, besides the fields for [sending] Identity:, From:, Reply To:, To:, CC:, and Subject:, another field Thread References:. This field would be non editable and would have a tick box beside it saying "Start New Thread", which would blank and restore the Thread References field as it was ticked and unticked. And when the message is sent, the Thread References would be incorporated or not as appropriate. At a stroke this should resolve the thread hijacking issue, because at the sending end you will actually have a visible handle on the real data structure which controls the thread. So are there any takers for this suggestion? Let me know by private email and I will enter it on the KDE wish list. regards Vince Littler

The 03.04.27 at 23:17, Anders Johansson wrote:

...

Found it at http://rfc.net/rfc2822.html and your extract is 3.6.4. - Identification fields.

Please note that that is not a standard, it is a proposed standard. It hasn't been accepted yet.

You should note that most are in the request for comments status, and have been so for many years. They are considered in fact standards. -- Cheers, Carlos Robinson

The 03.04.23 at 21:59, James Mohr wrote:

Since many people do not use threaded readers they obviously do not *see* the benefits of threaded messages.

They may not even know they can thread messages. Pine can, but it is somewhat hidden (K). So can balsa, mozilla... -- Cheers, Carlos Robinson

My take on 'thread hijacking' is that the technical reasons, like list managers, email clients, technical standards, etc is a secondary reason for it being a bad idea. I think it is just plain bad manners. (violating social standards) A thread is like a conversation and if you reply to a message with content that is not relevant to that message, you are chipping in.

For example, say I was talking to a person(s) about email clients. One person would say: "I like to use Evolution, because it can show threaded messages" Then I reply to him: "The weather here in SA is getting colder now"

Some people may say that I am rude, others may think that I am slightly crazy, because I replied with a comment that is totally irrelevant to the discussed topic.

So, if we see 'hijacking' a conversation as rude and improper, then how can we justify 'hijacking' a thread? It comes down to the same thing.

When you want to talk about something else, start a new conversation. Same with email.

I can tell you why some people "hijack" threads. And it boils down to not being aware of what they are doing to people who are using threaded email clients. Some people will simply reply to a mail so they don't have to worry about knowing the email address to post a message to a list. They change the subject and throw in the text according to what they are writing about. They don't KNOW what this does to a "thread". Now the only purpose this thread has served was to educate people on the ramifications of replying to a thread instead of creating a new one. I don't think people are neccessarily being rude, more like uninformed. Having said all that, this thread has now outlived its usefullness and I would suggest those of you who wish to argue amongst yourselves, maybe private email would be the way to go?

* James Mohr (suse_mailing_list@jimmo.com) [030423 10:51]:

I was referring to the mailing-list/server software. It has to be done at that level because there is no other way to keep track of of individual messages. Each message has an identifier which is at least relative to the mailing-list/server software (if not "world unique"). So replies are in references to this ID. The client has be able to recognize that fact in order to sort/organized the messages.

Actually ezmlm-idx, the list manager we use, doesn't rely on message IDs or reference headers for exactly this reason. All of the threading in the archives is done by normalizing the subject. For example, if I wanted to get this thread I could email suse-linux-e-thread.141961@suse.com where 141961 is the ezmlm message number of one of the postings in the thread. This pretty much always works correctly since the "thread hijacker" would have completely changed the subject. In contrast, hypermail (which creates the html archives), uses the message IDs for threading so the "hijacking" is always seen there. -- -ckm

* James Mohr (suse_mailing_list@jimmo.com) [030423 12:45]:

...

Actually ezmlm-idx, the list manager we use, doesn't rely on message IDs or reference headers for exactly this reason. All of the threading in the archives is done by normalizing the subject. [snip]

Aha. Thanks for the, Christopher. It's always a good day when I learn something new. While this does help with the archives, I see the normalizing the subject would be difficult in the "real time" of a mailing list. Or am I missing something?

Not really. I have mutt sort by by threads and unset strict_threads so that both the references and subject information is used. Of course, strict_threads would be ideal (and maybe faster) but as long as people use broken mail clients it makes mailing lists hard to read. Even this compromise doesn't work on mailing lists run by very broken stuff like the Lyris list manager on Windows. No references, the message-ids get munged, and the software is so stupid that it will repeatedly insert the list into the subject (e.g., 'Subject: [SLE] Re: [SLE] blah').

Were you talking about this header?

X-Message-Number-for-archive: 141961

Yes, it's also in the sender from.

So, if I wanted all the messages to a thread, I simply look at the header any message and pull out this number. Is that right?

Yep. There are a few other archive command as well, email suse-linux-e-help to see them. -- -ckm

The 03.04.23 at 20:11, James Mohr wrote:

I was referring to the mailing-list/server software. It has to be done at that level because there is no other way to keep track of of individual messages. Each message has an identifier which is at least relative to the mailing-list/server software (if not "world unique"). So replies are in references to this ID. The client has be able to

Not really. You are replying to an email from: From: Vince Littler Date: Wed, 23 Apr 2003 00:51:45 +0100 User-Agent: KMail/1.5.1 Message-Id: <200304230051.45646.suse@archipelago.eclipse.co.uk> with: From: James Mohr Date: Wed, 23 Apr 2003 20:11:40 +0200 User-Agent: KMail/1.4.3 References: <200304230051.45646.suse@archipelago.eclipse.co.uk> In-Reply-To: <200304230051.45646.suse@archipelago.eclipse.co.uk> Both "Message-Id" were added by the client software. For example, if you look at my messages on the list, you will see something like: Message-ID: So, they are added by the client software, when sending, and has to guaranteed unique. You can see your own answer makes reference to the "Message-Id" of Vince. There is indeed an ID for the list archive, but that is not used for threading at all: it was added a month or so ago. X-Message-Number-for-archive: 141961

recognize that fact in order to sort/organized the messages. If your client does not support this mechanism then the pointis moot. You won't

You both are using KMail. :-) -- Cheers, Carlos Robinson