Hi All,

In addition to this: from the Codeplex Practices&Patterns group I did receive confirmation that this behviour of the IModuleManager is a reproducable error: see http://compositewpf.codeplex.com/discussions/286333#post727221

It looks like I've found a workaround for this problem, and I'm not quite sure that it would be THE solution. So any reactions would still be very much welcome!

I've given up the principle that the ViewModel would be instantiated from the View.xaml. That brings me to the next "least objectable alternative" of initializing it from the View's code behind. This is the new command for that:

    <Import(AllowRecomposition:=False)> Public Property ViewModel As StartViewModel

And of course the related xaml code is skipped:

    <UserControl.DataContext> <vm:StartViewModel/> </UserControl.DataContext>

The ViewModel has to be reworked, so that it can be discovered by MEF. It needs an <Export> attribute and it can no longer inherit from mt Base_ViewModel. Instead, I use an ImportingConstructor to inject the necessary objects:

<Export()> Public Class StartViewModel
   
    <ImportingConstructor()> Sub New(EA As IEventAggregator, MM As IModuleManager)

        If Not IsNothing(EA) Then
            Locale_EventAggregator = EA
        End If

        If Not IsNothing(MM) Then
            Locale_ModuleManager = MM
        End If

        ...
        ...

    End Sub

But that lead to a new problem: the IModuleManager and the IEventAggregator appeared to be NOT the same instances as the ones created for the other modules. So in my bootstrapper I force the container to be the same container for both Prism and MEF with this command:

    Protected Overrides Function CreateContainer() As CompositionContainer
        Dim Local_Container = MyBase.CreateContainer()
        CompositionHost.Initialize(Local_Container)
        Return Local_Container
    End Function

As I said: I'm not completely certain of what I have done here, and comments or improvements to this solution would be highly appreciated!