Converting X,Y coords from a float into a DWORD?

Hey guys, I have the following packet containing the X and Y coordinates of an entity:
[66 96 5B 45] [9A D9 03 44]
X ___________Y

* they are in 32 bit float format

I did the following (read the comments):

Code:

---

int main()
{
        const BYTE pkt[]={0x66, 0x96, 0x5B, 0x45, 0x9A, 0xD9, 0x03, 0x44};

        ////// *** not works: return wrong values
        DWORD px = *(DWORD*)&pkt[0];
        DWORD py = *(DWORD*)&pkt[4];
        float fx = (float)px;
        float fy = (float)py;
        //////

        ////// *** works: it returns the correct values
        float fx = *(float*)&pkt[0];
        float fy = *(float*)&pkt[4];
        //////
       
        DWORD ix = (DWORD)fx;
        DWORD iy = (DWORD)fy;

return 0;       
}

---

Why one works and the other doesn't?

In first case, you just cast integer value 1163630182 (px) to float... so

Code:

---

float fx = (float)px;

---

will return exactly 1163630182 but as float type.

this will return correct float values from your array

Code:

---

        DWORD px = *(DWORD*)&pkt[0];
        DWORD py = *(DWORD*)&pkt[4];
        float fx = *((float*)&px);
        float fy = *((float*)&py);

---

Ah ok, so I need to treat the DWORD variable as a pointer to a float number to be able to cast it to a float var.
Thanks for clarifying!

Another variant of conversion. We can use Unions to map the same memory (values) as multiple types.

Code:

---

        union F2D {
            DWORD dwVal;
            FLOAT fVal;
        };
       
        F2D fThis;

        const BYTE pkt[]={0x66, 0x96, 0x5B, 0x45, 0x9A, 0xD9, 0x03, 0x44};

        ////// *** #3 variant
        fThis.dwVal = *(DWORD*)&pkt[0];
        float fx = fThis.fVal;

        fThis.dwVal = *(DWORD*)&pkt[4];
        float fy = fThis.fVal;

---

Excelent, using unions is a nicer way to play with the coordinates. Thanks very much for the idea!

---------- Post added 2011-12-08 at 05:35 AM ---------- Previous post was 2011-12-01 at 08:59 PM ----------

Grooguz, I'm posting again here because I got other question about this subject. Please, check my simple example:

Code:

---

#include<stdio.h>
typedef unsigned int DWORD;
 
void dump(unsigned char* buf, int size)
{
        for(int i=0;i<size;i++)
                printf("%02X ", buf[i]);
        printf("\n");
}
 
int main()
{
        unsigned char pkt[]={0x66, 0x96, 0x5B, 0x45};  //=3513
 
        DWORD dx = *(DWORD*)&pkt[0];
        float fx = *(float*)&dx;
        dump((unsigned char*)&fx,sizeof(fx));
 

        DWORD dx2 = 3513;
        float fx2 = *(float*)&dx2;
        dump((unsigned char*)&fx2,sizeof(fx2));
 
        return 0;
}

//    output:
//      fx  = 66 96 5B 45
//      fx2 = A8 C3 FD BF

---

As you can see I got another sequence of bytes different from the original input {0x66, 0x96, 0x5B, 0x45}. How to make it work properly?