UnhandledException when using ResourceDictionary with CustomResource and WindowsAppSDK 1.7.25

[PetarMladenov]

Describe the bug

Custom class defined in xaml constructs new CustomXamlResourceLoader in its static constructor. In xaml the setup is the following:

<Application.Resources>
     <ResourceDictionary>
         <ResourceDictionary.MergedDictionaries>
             <XamlControlsResources xmlns="using:Microsoft.UI.Xaml.Controls" />

	     <ResourceDictionary>

		<local:UserThemeResources x:Key="ResourceLoaderInitializer" />
		
		<ResourceDictionary.MergedDictionaries>
			<ResourceDictionary Source="{CustomResource DarkResourcesPath}"/>
		</ResourceDictionary.MergedDictionaries>
	     </ResourceDictionary>

	<!-- Other merged dictionaries here -->
         </ResourceDictionary.MergedDictionaries>
         <!-- Other app resources here -->
     </ResourceDictionary>
</Application.Resources>

In the constructor of UserThemeResources we set:

CustomXamlResourceLoader.Current = new UserThemeResourceLoader();

This builds and runs fine with WindowsAppSDK 1.6.25.

In 1.7.256 version of SDK we get:

{Microsoft.UI.Xaml.UnhandledExceptionEventArgs}
    Exception: {"The text associated with this error code could not be found.\r\n\r\nNo custom resource loader set. [Line: 17 Position: 27]"}

and also the static constructor of UserThemeResources is not fired at all.

Steps to reproduce the bug

UserThemeResources_Test.zip

Run the application with WindowsAppsSDK 1.7.25 restored. Notice the exception

Expected behavior

No exception and the constructor to fire.

Screenshots

NuGet package version

None

Windows version

Windows 11 (24H2): Build 26100

Additional context

No response

[michael-hawker]

I know for 1.7 the following issue was fixed (and indeed improved loading of resources in dictionaries): #8832 - but I wonder if this is an unintended regression with that change? @evelynwu-msft thoughts?

[evelynwu-msft]

I know for 1.7 the following issue was fixed (and indeed improved loading of resources in dictionaries): #8832 - but I wonder if this is an unintended regression with that change? @evelynwu-msft thoughts?

You're almost certainly correct.