How to make contiguous data in class C++

Question

I am using Visual C++ on Windows 10.
I want to make class Vector4 such that it has members x, y, z, t and they stored contiguously to provide operator[] and other functions:

class Vector4
{
public:
    float operator[](size_t idx)
    {
       return &x + idx;
    }
public:
   float x, y, z, t;
}

But I found out that such implementation works not on every compiler(On Visual C++ it works well). And I dont know how to fix it, maybe write something like this: class alignas(sizeof(float)) Vector4;

This cannot be done in a standard-compliant manner. You either have to concede the `.x` and `.y` access in favor of `.x()` and `.y()` or you have to use a `union` extension that allows you to access inactive members. As far as I am aware, all of the big 3 compilers guarantee it to be safe. — Fureeish, Jun 11 '22 at 08:57
Use an array and have reference members x, y, z, t refering to elements in that array — André, Jun 11 '22 at 09:19
You mean like `union {float xs[4]; struct {float x, y, z, w;};};`? — Dmytro, Jun 11 '22 at 09:23
@André ... and have your objects take 3x more space. You should also a least explain how to write an assignment operator. — n. m. could be an AI, Jun 11 '22 at 11:03
@n.1.8e9-where's-my-sharem. Indeed, but that is not one of the constraints in the question. And, as Fureeish nicely mentioned, there is not really a standard compliant alternative. — André, Jun 11 '22 at 16:59
@André "There is not really a standard compliant alternative" is the right answer. "Add reference members" is not one. It is not something a real world programmer would do in a real world program. — n. m. could be an AI, Jun 11 '22 at 17:13

ttemple · Answer 1 · 2022-06-11T12:14:04.333

I have the same need, and I did something like this:

class Vector4
{

public:
    // This is our data (4 doubles)
    union {
        double data[4];    // access as an array
        struct {
            double x;       // access as individual values
            double y;
            double z;
            double t;
        };
    };

public:
    // default construct
    Vector4() : data{ 0, 0, 0, 1 }{}
    // construct from 4 doubles
    Vector4(const double& x, const double &y, const double& z, const double & t) :
        x(x), y(y), z(z), t(t){}

    // const and non-const [] operators
    double& operator[](const size_t Index){ return data[Index]; }
    double operator[](const size_t Index) const { return data[Index]; }
};

This will allow accessing by [] or by .x, .y, etc.

Vector4 my_vec(1.1, 2.2, 3.3, 1.0);
auto x = my_vec.x;
auto y = my_vec[1];

I also implemented many other useful functions, such as dot product, cross product, etc.

Thank you, it must work, but I find other solution to this question easier — Mr_Ins1der, Jun 12 '22 at 09:30

score 0 · Accepted Answer · answered Jun 11 '22 at 12:09

From: C++ - Class contiguous data

You can use #pragma pack(1) with e.g. MSVC and gcc, or #pragma pack 1 with aCC.

As said in the answer #pragma pack(1) will disables padding and guarantees that the members are contiguous:

That basically disables padding and guarantees that the floats are contiguous. However, to ensure that the size of your class is actually 4 * sizeof(float), it must not have a vtbl, which means virtual members are off-limits.

See also the official C++ documentation: Implementation defined behavior control

The answer gives two ways:

#pragma pack(1)
class FourFloats
{
    float f1, f2, f3, f4;
};

and

#pragma pack(push, 1)
class FourFloats
{
    float f1, f2, f3, f4;
};
#pragma pack(pop)

How to make contiguous data in class C++

2 Answers2