Version: 1.8.2 (using KDE KDE 3.4.2) Installed from: Debian testing/unstable Packages OS: Linux It appears that when threading mode is enabled and there are unread messages in collapsed thread(s) (either because said thread(s) have been manually collapsed or because one prefers not to automatically expand threads with unread messages), "Go to next unread message" and "Go to previous unread message" simply refuses to expand the collapsed thread(s), instead skipping ahead to the next unread visible message. It becomes even sillier when there are no other unread messages except those in the collapsed thread(s), and Kmail is configured to start looking again from the beginning of the current folder. Kmail will then repeatedly ask if I want to go to the next unread message in <current folder>, but get nowhere.
(The version I'm looking at now is 1.9.1 (KDE 3.5.1).) This is no wonder as KMHeaders::findUnread() and KMHeaders::findUnreadAux() uses QListViewItem::itemAbove() and QListViewItem::itemBelow() to step through the message list. I have written a new implementation using an iterator. I hope QListViewItemIterator::operator--() isn't as slow as QListViewItem::itemAbove(). I also found a couple of peculiarities (they look like bugs, but it also appears they will never be executed): In KMHeaders::findUnreadAux(): > while (item) > { > msgBase = mFolder->getMsgBase(item->msgId()); > if (!msgBase) continue; Looks like an infinite loop should msgBase ever be null. The code in KMHeaders::findUnread() that forces the thread where the first new or unread message was found to be sorted appears to terminate too early in some cases. It stops when next == item, but item is the item found that is "not necessary the first unread item". When looking for the previous unread message there could be more unread messages in the unsorted branches below. I don't know if there are any threading issues, but if no unread item is found the first time there doesn't seem to be much point in doing the second pass. The sorting-enforcing bit would be more elegant using an iterator.
Created attachment 15348 [details] New findUnread (sketchy) There is duplicated code because I didn't want to create a new method or change the signature of an existing one myself. The two last remarks in my last comment haven't been implemented.
Created attachment 15442 [details] Make readSortOrder() (re)use the very useful findUnread() method Appears to be working in all three cases of GlobalSettings::self()->actionEnterFolder(). Maybe you already had this in mind when findUnread() got its onlyNew parameter?
*** This bug has been marked as a duplicate of 50806 ***
No, I read bug 50806 before filing this bug and found that it was not the same thing. At least this bug wasn't fixed. And it still isn't.
*** Bug 151509 has been marked as a duplicate of this bug. ***
Working with kde 4.2. version.