oh wow, this is actually pretty bad, not your fix/workaround, but the bug itself, i know i didn't think it would be a Qt bug, but from seeing your code and playing with it a bit it may actually be.

Can you confirm this actually fixes the problem for you?

I like it a bit better because:

• doesn't create the Splash when it's not needed
• removes the processEvents call which in general is a bit nasty and "shouldn't be needed"
• Uses the pointer to function method for close instead of the version with quotes

diff --git a/src/k3bapplication.cpp b/src/k3bapplication.cpp
index 76e764f77..379c53091 100644
--- a/src/k3bapplication.cpp
+++ b/src/k3bapplication.cpp
@@ -49,6 +49,7 @@
 
 #include <QCommandLineParser>
 #include <QDebug>
+#include <QTimer>
 
 
 K3b::Application::Core* K3b::Application::Core::s_k3bAppCore = 0;
@@ -91,12 +92,9 @@ void K3b::Application::init( QCommandLineParser* commandLineParser )
     else {
         m_mainWindow->show();
     }
-    
-    processEvents();
-    
+
     if (splash) {
-        splash->hide();
-        QMetaObject::invokeMethod( splash, "close", Qt::QueuedConnection );
+        QTimer::singleShot(0, splash, &QWidget::close);
     }
 
     qRegisterMetaType<KSharedConfig::Ptr>( "KSharedConfig::Ptr" );

It does work for me. However, can you explain the changes a bit further? Why did you comment out processEvents()? What does the functional call do in this context?

In D28469#642536, @ictwod wrote:

It does work for me. However, can you explain the changes a bit further? Why did you comment out processEvents()? What does the functional call do in this context?

Oh sorry, i totally missed your answer.

processEvents in general is bad design or a workaround for bad design.

The thing about events is they are supposed to be that, things that someone sends and you process when you get them. Manually calling processEvents says "oh i know someone sent an events and i need to process it now because otherwise bad things will happen", which is just bad.

Since you mention that my patch also works for you, i'll commit that one.

Your patch was not bad, but the one i did is a bit better for the reasons mentioned in the previous commit.

I hope that doesn't discourage you to keep working on more patches!