Skip to main content

Enabling Globalization Invariant Mode for .NET Core App on Raspberry Pi Running LibreElec

I had an app I wanted to run on my Raspberry Pi 3 running LibreElec. In LibreElec you can install the dotnet core 2.2 runtime as an addon, and in Visual Studio you can compile for ARM processors with ‘Target Runtime’ set to ‘linux-arm’ in the publish profile. So, I published to a folder from VS using that profile, and I copied the output over to my RPi which had the dotnet runtime installed. I did a simple dotnet Whatever.dll to run the app (actually in this case, it was /storage/.kodi/addons/tools.dotnet-runtime/bin/dotnet Whatever.dll because of the way the addon is installed) and was met with this error:

FailFast:
Couldn't find a valid ICU package installed on the system. Set the configuration flag System.Globalization.Invariant to true if you want to run with no globalization support.

   at System.Environment.FailFast(System.String)
   at System.Globalization.GlobalizationMode.GetGlobalizationInvariantMode()
   at System.Globalization.GlobalizationMode..cctor()
   at System.Globalization.CultureData.CreateCultureWithInvariantData()
   at System.Globalization.CultureData.get_Invariant()
   at System.Globalization.CultureInfo..cctor()
   at System.StringComparer..cctor()
   at System.AppDomain.InitializeCompatibilityFlags()
   at System.AppDomain.CreateAppDomainManager()
   at System.AppDomain.Setup(System.Object)
Aborted

What’s happening here is that .NET Core relies on ICU for globalization. On Linux, if the ICU binary is not present (which is the case with LibreElec), the application will fail to start. To resolve the issue you either need to install ICU on the system or turn off globalization with the runtime configuration options for globalization. Since I wasn’t interested in trying to compile ICU for LibreEelec (there are no binaries ready-to-go), I chose the latter.

Rather than set the DOTNET_SYSTEM_GLOBALIZATION_INVARIANT environment variable, I wanted to modify the Whatever.runtimeconfig.json file. There are two ways to do this automatically at publish-time:

  • Set an option in the project file, like this

    <ItemGroup>
     <RuntimeHostConfigurationOption Include="System.Globalization.Invariant" Value="true" />
    </ItemGroup>
    
  • Add a runtimeconfig.template.json file with the properties you want included by default, like this

    {
     "configProperties": {
      "System.Globalization.Invariant": true
     }
    }
    

I tested both, and ultimately selected the first method. Before enabling globalization invariant mode, my runtimeconfig.json looked like this:

{
  "runtimeOptions": {
    "configProperties": {
      "System.GC.Server": true
    }
  }
}

Afterward, it looked like this:

{
  "runtimeOptions": {
    "configProperties": {
      "System.Globalization.Invariant": true,
      "System.GC.Server": true
    }
  }
}

Hat tip to this issue which is one of the top results when you Google the specific error message.

Comments

Popular posts from this blog

Migrating Hg Repos with hg-fast-export and Windows Subsystem for Linux

Introduction I prefer Mercurial (hg) to git . I don’t really have any reason for this preference - they both do the same thing, and the user experience for 90% of the use cases is the same. It probably comes from the conditions of the DVCS landscape when I started using these systems. Some of this may have been perception only, but it looked like this: GitHub didn’t have free private repos BitBucket did have free private repos BitBucket was very hg-friendly Joel Spolsky had an amazing tutorial that served as both a how-to for hg as well as a general intro to DVCS hg was much more Windows-friendly than git Since hg was written in python, I felt like extending it would be easier than doing so for git if I ever needed to (admittedly, this is a pretty ridiculous reason) hg felt like a more unified, “coherent” system than the very linux-y feeling git and its extensions (also pretty ridiculous) Where they differed, I liked the verbs hg used better than git’s counterparts

Stubbing Static Methods with PostSharp

TypeMock uses the Profiler API to allow mocking, stubbing, etc. of classes used by code under test. It has the ability to handle sealed classes, static classes, non-virtual methods, and other troublesome-yet-oft-encountered scenarios in the world of unit testing. Other frameworks rely on proxies to intercept method calls, limiting them to be able to only fake virtual, abstract, and interface members. They also rely on dependecy injection to place the proxies as the concrete implementation of calls to the abstracted interface members. Anyone working with a legacy codebase is bound to run into static method calls (especially in the data access layer), dependencies on concrete types with non-virtual methods, and sealed class dependencies (HttpContext anyone?). The only way to unit test this without refactoring is with TypeMock. I've never used TypeMock, and I'm sure it's a great product, but it's not free. I decided to spike some code to see if I could solve the prob