Simulation support: fixed a problem with SimulationSupportPythonWrapp…

…er::run().

Indeed, it may happen, when trying to call disconnect() on pSimulation, that pSimulation has been deleted just before. So, to call disconnect() hangs up IPython and renders our Python Console window unusable.

Unfortunately, there is no (easy?) way to test whether pSimulation is still valid. So, we don't call disconnect() anymore and, instead, pass Qt::UniqueConnection in our calls to connect(). This leaves the connection with waitLoop, but this isn't (shouldn't be?) a problem since waitLoop gets out of scope when leaving SimulationSupportPythonWrapper::run().

src/plugins/support/SimulationSupport/src/simulationsupportpythonwrapper.cpp

@@ -276,9 +276,11 @@ bool SimulationSupportPythonWrapper::run(Simulation *pSimulation)
         // Keep track of any simulation error and of when the simulation is done
 
         connect(pSimulation, &Simulation::error,
-                this, &SimulationSupportPythonWrapper::simulationError);
+                this, &SimulationSupportPythonWrapper::simulationError,
+                Qt::UniqueConnection);
         connect(pSimulation, &Simulation::done,
-                this, &SimulationSupportPythonWrapper::simulationDone);
+                this, &SimulationSupportPythonWrapper::simulationDone,
+                Qt::UniqueConnection);
 
         // Run our simulation and wait for it to complete
         // Note: we use a queued connection because the event is in our
@@ -294,8 +296,6 @@ bool SimulationSupportPythonWrapper::run(Simulation *pSimulation)
 
         waitLoop.exec();
 
-        disconnect(pSimulation, nullptr, this, nullptr);
-
         // Throw any error message that has been generated
 
         if (!mErrorMessage.isEmpty()) {