[Testing] Testing Error with IDialogAware implemented ViewModel The DialogCloseCallback has not been properly initialized

[james1301]

Description

Trying to test a ViewModel that implements IDialogAware in Prism 9.0 produces an error that crashes TestExplorer as it fires from an async void on invocation of the DialogCloseListener.

Repo: https://github.com/james1301/PrismDialogTestingError

Steps to Reproduce

1. Create a class that implements IDialogAware, with a method that invokes to close the ViewModel
   e.g. this.RequestClose.Invoke(dialogResult);
2. Create a unit test to call that method.
3. Error in Test Output window

Platform with bug

WPF

Affected platforms

Windows

Did you find any workaround?

To set use DialogUtitlities to initialise the listener.

e.g. DialogUtilities.InitializeListener(viewModel, (r) => { });

Relevant log output

Unhandled exception. System.InvalidOperationException: The DialogCloseCallback has not been properly initialized. This must be initialized by the DialogService, and should not be set by user code.
   at Prism.Dialogs.DialogCloseListener.Invoke(IDialogResult result)
   at System.Threading.Tasks.Task.<>c.<ThrowAsync>b__128_1(Object state)
   at System.Threading.QueueUserWorkItemCallback.Execute()
   at System.Threading.ThreadPoolWorkQueue.Dispatch()
   at System.Threading.PortableThreadPool.WorkerThread.WorkerThreadStart()

[dansiegel]

thanks @james1301 I can certainly see how this would have been an issue. As you've already discovered the correct way to initialize it for Unit Testing, you'll notice that I've moved this to the Documentation repo as an issue for us to better address in the docs.

This isn't something that we're planning on changing as this is a bit by design and the helper is there for you. Note that you can can also use that callback to help you in your unit tests for instance to confirm that a parameter is properly passed back.

[james1301]

Yeah it wasn't so bad, for the tests I knew I had to update as RequestClose was throwing up a compiler error. It was the ones that weren't were throwing up the test runner error which was a bit harder to find and track down as there is no link to the tests that cause it.