text formatting truncates on macos using dotnet binding

[emperor-limitless]

Just compiled about 3 hours ago after cloning at the same time, using the C# binding, text somehow either just breaks or truncates.
as an example, login becomes lgn, exit ei.
just to confirm I haven't modified it, I replaced the binding file again, and the problem still exists.
I think it's a macos only issue, windows is fine last I checked, though the windows one was compiled about months ago so perhaps.
macos 15.3.2, arm64.

[emperor-limitless]

also, I forgot to mention, I am not certain if it's a binding issue or a library on macos in general issue.
additionally, I have compiled without java support, after testing the release package, it seems to be both different in size and behavior, because it seems to require some other library because if I try to load it, it crashs with a library not found message, so I'm assuming it tries to load java for a reason or another.

[emperor-limitless]

ok, after asking claud and showing the header and binding, the problem seems to be because we expect wchar_t for windows, which is valid, but on macos, wchar_t is 4bytes unlike windows's 2, so problems happened, the solution looks like this

#if OSX
    [DllImport(DllName, CallingConvention = CallingConvention.Cdecl)]
    [return: MarshalAs(UnmanagedType.Bool)]
    private static extern bool Speech_Output(IntPtr textPtr, [MarshalAs(UnmanagedType.Bool)] bool interrupt);
#else
// Windows thing here.
#endif

after doing a platform check, and adding it for mac only, then, do the same and change the mac speak method like this.

#if OSX
    // macOS-specific implementation for Speak
    public bool Speak(string text, bool interrupt = false)
    {
        if (string.IsNullOrEmpty(text))
            return false;

        // On macOS, convert to UTF-32 for wchar_t (4 bytes per character)
        byte[] utf32Bytes = Encoding.UTF32.GetBytes(text + "\0");
        
        // Allocate unmanaged memory
        IntPtr textPtr = Marshal.AllocHGlobal(utf32Bytes.Length);
        
        try
        {
            // Copy the bytes to unmanaged memory
            Marshal.Copy(utf32Bytes, 0, textPtr, utf32Bytes.Length);
            
            // Call the function
            return Speech_Output(textPtr, interrupt);
        }
        finally
        {
            // Free the unmanaged memory
            Marshal.FreeHGlobal(textPtr);
        }
    }
#endif

and do that for anything that expects wchar_t, like braille too.
however one issue with this is, from my understanding, C# doesn't by default have preprocesser directives for platform checking, and you have to manually add it in the csproj, like this

<Project Sdk="Microsoft.NET.Sdk">
  <PropertyGroup>
    <IsWindows Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::Windows)))' == 'true'">true</IsWindows>
    <IsOSX Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::OSX)))' == 'true'">true</IsOSX>
    <IsLinux Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::Linux)))' == 'true'">true</IsLinux>
  </PropertyGroup>
  <PropertyGroup Condition="'$(IsWindows)'=='true'">
  <DefineConstants>Windows</DefineConstants>
  </PropertyGroup>
  <PropertyGroup Condition="'$(IsOSX)'=='true'">
  <DefineConstants>OSX</DefineConstants>
  </PropertyGroup>
  <PropertyGroup Condition="'$(IsLinux)'=='true'">
  <DefineConstants>Linux</DefineConstants>
  </PropertyGroup>

but that obviously requires a form of project. My solution is perhaps having this on nuget which will solve that problem entirely.