Re: possible new sort option

On 08/06/2018 01:29 AM, andré via balsa-list wrote:
Although having a reply-to address in the header is definitely preferable for mailing lists, I don't see how that has much affect on threads, except that from time to time a reply doesn't go directly to the list, but appears in a subsequent message with a reply to the reply.
It's not Reply-To: header in question but the two headers about the thread (I forget the exact names) In-reply-to: which is the message-id of the message being responded to, and References: which can be a list of previous messages in the thread.  As I understand it, when JWZ came up with his algorithm, the latter two headers were not in general use, so that's why he had the fallback of using Subject:.

If someone replies to the sender, and that person replies back to the list, all the reference headers should still be in place, so threading should be OK, with the apparent omission of one message, which wasn't actually sent to the list.  I am concerned about people whose responses to mailing lists seem to consistently break threading (in my case, when viewed in Pan, an NNTP reader).  My suspicion is that their mailer is not using the reference headers, and Pan is not doing JWZ.  I suppose I will just have to check some of those instances in more detail to be sure.  (I do have one list I follow in both Balsa and Pan, so I think I'll be able to compare.)

Note we have two separate issues being discussed:
1) Ordering entire threads by either the earliest (current practice) or most recent message in the thread. 2) Avoid an earlier message being indented under a later message (which happens, for example) with Subject's like "New bug" and most likely when the most recent message does NOT have Reference or In-Reply-To header (which, of course, it wouldn't if it's actually a new thread using the same Subject as an old one.)

I think Peter's proposal to allow the first also accomplishes the second, by no longer including Subject in the threading algorithm. While that might miss some threading connections currently made, it's probably the path of least resistance and easiest implementation.

Also - I don't think this has anything to do with the format of the folder - mbox, maildir, .... should all be handled the same way.


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]