On November 22nd, 2011, 9:24 a.m., Thomas Lübking wrote:

plasma/private/containment_p.h (Diff revision 1)
public:
		71	while (!applets.isEmpty()) {

ibeg your pardon?
to me this looks like you delete the same entry forever (what *will* crash at a point) since you don't remove it - should be "takeFirst", yesno?

On November 22nd, 2011, 9:27 a.m., Ben Cooksley wrote:

Other code handles this for us - which is why qDeleteAll() crashes.

On November 22nd, 2011, 9:57 a.m., Thomas Lübking wrote:

sorry, but the only way i see this could be correct is that qlist (fun with templates) subclasses pointer Types and reimplements delete to automatically remove them from the list, what is, nicely spoken, sick.
is there any detailed (official) explanation on this?

On November 22nd, 2011, 10:11 a.m., Aaron J. Seigo wrote:

see ContainmentPrivate::appletDestroyed(Plasma::Applet *applet) in kdelibs/plasma/containment.cpp

Only information I have is https://git.reviewboard.kde.org/r/102981/
I haven't checked the code, but i'm guessing that Plasma::Applet must do something similar to what Akregator does.
In any case, I tested and Plasma does not infinitely loop - it does close properly.

- Ben

On November 22nd, 2011, 7:39 a.m., Ben Cooksley wrote:

Qt 4.8 contains changes which break qDeleteAll() in certain scenarios. This patch fixes one of those scenarios to ensure that Plasma Desktop does not crash on quit or logout.

Compiled, Plasma Desktop no longer crashes on exit.