How to generalize a struct to an int with the assignment operator?

Question

I'm trying to use a struct to pack multiple smaller integers into a uint32_t.

struct PackedData {
    PackedData & operator=(uint32_t x) {
        a = (x >> 24) & 0xFF;
        b = (x >> 8) & 0xFFFF;
        c = x & 0xFF;
        return *this;
    }
    uint32_t operator=(PackedData x) {
        uint32_t temp;
        temp = (x.a << 24) & 0xFF000000;
        temp |= (x.b << 8) & 0xFFFF00;
        temp |= x.c & 0xFF;
        return temp;
    }

    uint32_t a : 8;
    uint32_t b : 16;
    uint32_t c : 8;
};

Originally, I had hoped to inherit from a uint32_t, similar to what you can do with an enum class.

struct PackedData : uint32_t {

Unfortunately that does not compile.

The weird thing is that the struct I've pasted does compile, and doesn't even raise warnings with -Wall -Wextra --pedantic.

On the bright-side, I can see that my struct is 4 bytes, and I am able to assign a 32-bit integer to the struct. However, I cannot assign the struct to a 32-bit integer.

I'm sure the syntax for the following method is completely wrong, but I don't know how and the compiler isn't helping me.

uint32_t operator=(PackedData x) {
    uint32_t temp;
    temp = (x.a << 24) & 0xFF000000;
    temp |= (x.b << 8) & 0xFFFF00;
    temp |= x.c & 0xFF;
    return temp;
}

Can this idea even work, or is there a better way?

Answer 1

You can define an implicit conversion operator as follows:

struct PackedData {

    // ...

    operator uint32_t () const {
        return (a << 24) | (b << 8) | c;
    }
};

Now you can do:

PackedData p { 0xef, 0x1234, 0xab };
uint32_t x = p;

The value of x is 0xef1234ab.

Note that the masking when converting from the bit-fields is not necessary. The compiler already takes care of that internally.

If it's super important that your struct is always the correct size due to assumptions elsewhere in a larger project, it's generally a good idea to add a static assertion. The most appropriate place to do that is typically following the class definition:

struct PackedData {
    // ...
};

static_assert(sizeof(PackedData) == sizeof(uint32_t));