54593 - Thread sorting is funny, causing odd 'next' behaviour

Status: VERIFIED FIXED

(SeaMonkey :: MailNews: Message Display, defect, P3)

Description

[Peter Janes]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.2.16-22 i686; en-US; m18) Gecko/20000928
BuildID:    2000092808

I like to read news threaded and sorted by descending date (so new
thread/messages are at the top of the pane).  When I click the 'next' button on
an unopened thread, it often takes me to a message far down the thread instead
of the message immediately following the current one.

Reproducible: Sometimes
Steps to Reproduce:
Here's an example (see the thread 'querky behavior with paragraph indent and
numbered lists' on netscape.public.mozilla.mail-news, and please excuse the
ASCII-gram):

> Message 1______09/28/2000_01:14
> Message 5______09/28/2000_08:07

Clicking 'next' gives:

v Message 1______09/28/2000 01:14
__v Message 2____09/28/2000 01:24
______Message 3__09/28/2000 03:33
____Message 4____09/28/2000 01:21
> Message 5______09/28/2000 08:07

with 'Message 4' highlighted.

Actual Results:  Message 4, the next message by date, is highlighted when 'next'
is pressed.  'Next' again goes to message 5, completely skipping messages 2 and 3.

Expected Results:  'Next' should go to the next message in the displayed list,
not the next message in the sort order.

Comment 1

[Peter Janes]

When I hit the "bottom" of a thread, I often can't advance to the next thread,
either.  If message 4 were missing from the example, I'd go through 1, 2 and 3,
then be unable to advance to 5 with 'next'.

Comment 2

[Asa Dotzler [:asa]]

Much of this was recent;y fixed or improved by checkins for bug 57139.  Please
test a new build and report back. thanks.

Comment 3

[Peter Janes]

I see a bit of improvement in 2000110108, but it's still quite odd.  I don't
think I've been "stuck" at the end of a thread recently, but the behaviour as
described originally still exists.

I still jump all over the thread, generally to the end, and then continue on out
of it to the next thread.  If I keep going to the 'next' message, I eventually
return to the next unread message in the original thread, but I can't discern
any pattern to when or why I return, or why I can read some threads in their
entirety in the right order and some jump out to different messages or to
different positions in the thread.

Here's an example that I've just performed, all messages being originally
unread.  This is from my IMAP mailbox, but similar behaviour on an NNTP server.

The 'next' sequence, starting at message 1, is:
1, 6, 7, 9, 10, 13, 14, 2, 3, 4, 5, 8, 11, 12

1 ApacheCon Jetspeed talks
2 __Re: ApacheCon Jetspeed talks
3 ____Re: ApacheCon Jetspeed talks
4 __Re: ApacheCon Jetspeed talks
5 __Re: ApacheCon Jetspeed talks
6 __Re: ApacheCon Jetspeed talks
7 Can't build Jetspeed source
8 __Re: Can't build Jetspeed source
9 __Re: Can't build Jetspeed source
10portlet - security proxy?
11__Re: portlet - security proxy?
12__Re: portlet - security proxy?
13__Re: portlet - security proxy?
14____Re: portlet - security proxy?

Comment 4

[Keyser Sose]

Marking as NEW because the reporter is still seeing it.

Comment 5

[Aaron Mathews]

I still observe the behaviour in 2000111508. In my case, while going through 
the thread it will randomly jump to almost the top of the message list window.

Comment 6

[Peter Janes]

This behaviour really messes up deleting messages, too, since you never get back
to the previous messages unless you hit 'next'.  Similar behaviour also when
navigating the folder/newsgroup list.

Comment 7

[[e-mail] (gone - use [e-mail] instead)]

I believe Seth is the right engineer for this now.  Assigned to : seth, QA :
stephend

Comment 8

[Eric Vaandering (no email)]

*** Bug 65511 has been marked as a duplicate of this bug. ***

Comment 9

[(not reading, please use [e-mail] instead)]

the view code needs to be fixed to fix this.

we've got a plan ready for fixing the view code, but performance comes first.

Comment 10

[Brendan Eich [:brendan]]

How about a target-milestone <= 1.0?

/be

Comment 11

[Keyser Sose]

*** Bug 65962 has been marked as a duplicate of this bug. ***

Comment 12

[Eric Vaandering (no email)]

Are people still having this problem? My quick check of a recent build worked fine.

Comment 13

[(not reading, please use [e-mail] instead)]

yes, this is fixed.

we brought over the 4.x view code when we landed mailnews performance branch.

marking fixed.

Comment 14

[[e-mail] (gone - use [e-mail] instead)]

Verified FIXED:
2001-04-16-04 Windows 2000
2001-04-16-04 Mac OS 9.1
2001-04-16-10 RedHat 7.0

Comment 15

[Pratik]

Bug 65962 was marked as a duplicate of this bug and I am still seeing the
behaviour that I reported in 65962. Maybe 65962 should be reopened?

Comment 16

[[e-mail] (gone - use [e-mail] instead)]

Yes, I'll re-open that bug.  When you filed that bug, I wasn't sure how this 
behavior would be implemented, so we need to re-evaluate the bug that you filed. 
 Sorry about that.