C ++/C#浮点问题 [英] C++/C# Float issues

问题描述

尝试从C#中的C ++读取结构时遇到问题. 我做了多次，并且始终有效，但是我从未使用过浮点数.

I have a problem when trying to read a struct from C++ in C#. I did this more times and it worked always, but i never used floating point numbers.

这些是我在C#端使用的结构:

These are the structs i use on the C# side:

[StructLayout(LayoutKind.Sequential)]
public struct ImageData
{
    public UInt32 Width;
    public UInt32 Height;
    public UInt32 Length;
    public IntPtr PixelPointer;
}
[StructLayout(LayoutKind.Sequential)]
public struct UserData
{
    public UInt32 misc;
    public Boolean IsValid;
    public VirtualPosition RightHand;
    public VirtualPosition LeftHand;
    public VirtualPosition Head;
}
[StructLayout(LayoutKind.Sequential)]
public struct VirtualPosition
{
    public Position3D RealPosition;
    public Position3D ProjectedPosition;
}
[StructLayout(LayoutKind.Sequential)]
public struct Position3D
{
    public float X;
    public float Y;
    public float Z;
}

这是我在C ++方面使用的结构:

This are the structs i use on the C++ side:

    typedef struct XnVector3D
    {
       XnFloat X;
       XnFloat Y;
       XnFloat Z;
    } XnVector3D;
    class VirtualPosition {
    public:
        XnVector3D RealPosition;
        XnVector3D ProjectedPosition;

        VirtualPosition(void) { }
        VirtualPosition(XnVector3D real, XnVector3D projected) {
            RealPosition = real;
            ProjectedPosition = projected;
        }
    };

    class UserData
    {
        public:
        int misc; // Misc number to check if i have the correct pointer ( debug only )
        bool Valid;
        VirtualPosition RightHand;
        VirtualPosition LeftHand;
        VirtualPosition Head;
    };

现在使用C#部分获取数据:

Now the C# part which has been used to get the data:

public unsafe UserData GetUserData(UInt32 user)
{
    IntPtr pos = GetData(_Index, user);
    UserData* ptr = ((UserData*)pos.ToPointer());
    UserData result = *ptr;
    return result;
}
[DllImport("KinectNative.dll", CallingConvention = CallingConvention.Cdecl)]
private static extern IntPtr GetData(Int32 index, UInt32 user);

最后是用于检索数据的部分(C ++):

And finally the part which is used to retrieve data( C++ ):

EXPORT UserData* GetData(int index, XnUserID user) 
{
    if (_LastData != NULL) _LastData->~UserData();
    system("cls");
    _LastData = new UserData(_Context->GetDevice(index)->GetData(user));
    std::cout << "X: " << (int)_LastData->Head.ProjectedPosition.X << " Y:" << (int)_LastData->Head.ProjectedPosition.Y << std::endl;
    return _LastData;
}

在C#中读取时，杂项编号是正确的，并且控制台会告诉我正确的编号，但是所有浮点数都已损坏，并且在C ++和C#中获得不同的编号时.无论如何，差异是可以接受的. (在C ++中为340，在C#中为0.xxxxxxx左右).

The misc number is correct when reading it in C# and the Console tells me the correct numbers, but all floats are corrupted and when i get different numbers in C++ and C#. Anyway the difference is to large to accept it. ( Got 340 in C++ and around 0.xxxxxxx in C# ).

任何帮助都将受到欢迎.谢谢.

Any help would be welcome. Thanks.

推荐答案

是因为.Net Boolean的大小与C ++实现bool的大小不同吗?我认为VC ++对bool使用1个字节，而.Net对System.Boolean使用2个或4个字节.

Could it be because the .Net Boolean is not the same size as your C++ implementation bool? I think VC++ uses 1 byte for bool while .Net uses 2 or 4 bytes for System.Boolean.

尝试将IsValid定义为Byte而不是Boolean，然后查看它是否有效.

Try defining IsValid as Byte instead of Boolean and see if it works.

我认为您还可以在Boolean之前添加MarshalAs属性，以强制将其封送为1字节:

I think you can also add the MarshalAs attribute before Boolean to force it to be marshaled as 1-byte:

[StructLayout(LayoutKind.Sequential)]
public struct UserData
{
    public UInt32 misc;
    [MarshalAs(UnmanagedType.U1)]
    public Boolean IsValid;
    public VirtualPosition RightHand;
    public VirtualPosition LeftHand;
    public VirtualPosition Head;
}

但是我不确定它是否适合您的情况，因为您直接使用指针而不是封送处理.

But I'm not sure it's going to work in your case, since you directly use pointers, not marshaling.

这篇关于C ++/C#浮点问题的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！