Return Array from Class

Question

I need to return 3 values. X, Y, Z. I've tried something like this, but it does not work, can anyone help me a bit? I've looked here: Return a float array in C++ and I tried to do same thing, except with 1 dimensional array to return.

class Calculate
{
 float myArray[3][4], originalArray[3][4], tempNumbers[4];
 float result[3]; // Only works when result is 2 dimensional array, but I need 1 dimension.

public:
 Calculate(float x1, float y1, float z1, float r1,
  float x2, float y2, float z2, float r2,
  float x3, float y3, float z3, float r3)
 {
  myArray[0][0] = x1;
  myArray[0][1] = y1;
  myArray[0][2] = z1;
  myArray[0][3] = r1;

  myArray[1][0] = x2;
  myArray[1][1] = y2;
  myArray[1][2] = z2;
  myArray[1][3] = r2;

  myArray[2][0] = x3;
  myArray[2][1] = y3;
  myArray[2][2] = z3;
  myArray[2][3] = r3;

  result[0] = 1;
  result[1] = 2;
  result[2] = 3;
 }

 float* operator[](int i)
 {
  return result[i]; //Value type does not match the function type
 }

 const float* operator[](int i) const
 {
  return result[i]; //Value type does not match the function type
 }
};

Should probably be &result[i].

Arc
– Arc

2010-09-24 15:12:58 +00:00
Commented Sep 24, 2010 at 15:12 — Arc
– Arc, Commented Sep 24, 2010 at 15:12

Reinderien · Accepted Answer · 2010-09-24 15:26:16Z

5

Instead of returning a pointer, it's usually better practice to accept a pointer and write out the results there. That way someone can allocate a regular array on the stack and have it initialized by your Calculate.

Something like:

class Calculate
{
 float myArray[3][4], originalArray[3][4], tempNumbers[4];

public:
 Calculate(float x1, float y1, float z1, float r1,
  float x2, float y2, float z2, float r2,
  float x3, float y3, float z3, float r3, float *result)
 {
  myArray[0][0] = x1;
  myArray[0][1] = y1;
  myArray[0][2] = z1;
  myArray[0][3] = r1;

  myArray[1][0] = x2;
  myArray[1][1] = y2;
  myArray[1][2] = z2;
  myArray[1][3] = r2;

  myArray[2][0] = x3;
  myArray[2][1] = y3;
  myArray[2][2] = z3;
  myArray[2][3] = r3;

  result[0] = 1;
  result[1] = 2;
  result[2] = 3;
 }
};

Some other tweaks you can do - separate the constructor from the calculation, since constructors are more for initialization; and pass arrays for safer memory control:

class Calculate
{
    float myArray[3][4], originalArray[3][4], tempNumbers[4];

public:
    Calculate(const float initArray[3][4])
    {
        for (int i = 0; i < 3; i++)
            for (int j = 0; j < 4; j++)
                myArray[i][j] = initArray[i][j];
    }

    void DoCalculation(float result[3]) const
    {
        result[0] = 1;
        result[1] = 2;
        result[2] = 3;
    }
};

int main()
{
    float myArray[3][4] =
    {
        { 0, 1, 2, 3 },
        { 4, 5, 6, 7 },
        { 8, 9, 0, 1 }
    };
    float result[3];
    Calculate calc(myArray);
    calc.DoCalculation(result);
    return 0;
}

edited Sep 24, 2010 at 15:26

answered Sep 24, 2010 at 15:11

Reinderien

16.7k9 gold badges56 silver badges92 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Stan Over a year ago

Thanks, worked great, even better because it takes less code.

jkerian Over a year ago

Note that passing pointers is not generally considered a better practice in C++, but then, neither is returning pointers to internal members of a class.

strager Over a year ago

@jkerian, Yup; this is why I prefer the approach by @Thomas Matthews.

Thomas Matthews · Accepted Answer · 2010-09-24 16:49:58Z

2

Another alternative is to make the result as a separate structure and either return it as a value or pass it by reference:

struct Result_Type
{
  double values[3];
// Alternatively:
// double x;
// double y;
// double z;
};

// Returning a result
const Result_Type calculate_result_1(/* yada yada yada */)
{
  Result_type new_result;
  new_result.value[0] = 0;
  new_result.value[1] = 0;
  new_result.value[2] = 0;
  return result;  // Return the result as a object
}

// Or passing a result to be modified
void  clear_result(Result_Type & any_result) // <-- Note pass by reference
{
  any_result.value[0] = 0;
  any_result.value[1] = 0;
  any_result.value[2] = 0;
  return;
}

You may find that this is a preferred design since you can pass around results, modify the Result_Type structure to perform operations with other result vectors (math term). The matrix can also be considered as a composition of result vectors.

This may make the code easier to read also.

answered Sep 24, 2010 at 16:49

Thomas Matthews

58.1k18 gold badges105 silver badges165 bronze badges

1 Comment

rubenvb Over a year ago

+1 except for one thing: I would not put an array in the struct, but three seperate values (of which each has its own meaning, like x, y, and z for example).

Andrew Jaffe · Accepted Answer · 2010-09-24 17:18:09Z

2

result[i] is a float, not a float*, so you can do

const float operator[](int i) const
{
  return result[i]; 
}

But I think you do want to return a reference to get the correct semantics, so you want

const float& operator[](int i) const
{
  return result[i]; 
}

float& operator[](int i)
{
   return result[i];
}

Right? (I think this is OK -- it compiles but it's been a while since I've done this...)

edited Sep 24, 2010 at 17:18

answered Sep 24, 2010 at 15:16

Andrew Jaffe

27.2k4 gold badges54 silver badges59 bronze badges

Comments

UncleBens · Accepted Answer · 2010-09-24 15:19:12Z

0

In the code, where you are getting an error, you are not trying to return a pointer. You are trying to return a single float at the given index.

Calculate c;
float first = c[0];
float second = c[1];
float third = c[2];

If you meant to return a pointer to the results array, then you would have to return the array, e.g

float* GetResult() { return result; }

It probably doesn't matter much which you'll keep because the end effect is pretty much the same. If you overload operator[], you'll have more control, though, as you can check out-of-bound accesses.

answered Sep 24, 2010 at 15:19

UncleBens

41.4k6 gold badges62 silver badges90 bronze badges

Comments

Chubsdad · Accepted Answer · 2010-09-25 04:49:20Z

0

Some responses have indicated the use of pointers. The problem is who allocates that pointer and who frees it. Also one needs to check if the incoming pointer is NULL and so on. Instead I would suggest the following declaration of the constructor.

 Calculate(float x1, float y1, float z1, float r1, 
  float x2, float y2, float z2, float r2, 
  float x3, float y3, float z3, float r3, float (&r)[3])

This is much more safer as references can not be NULL.

edited Sep 25, 2010 at 4:49

answered Sep 25, 2010 at 1:31

Chubsdad

25.7k6 gold badges82 silver badges144 bronze badges

Collectives™ on Stack Overflow

Return Array from Class

5 Answers 5

3 Comments

1 Comment

Comments

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

5 Answers 5

3 Comments

1 Comment

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related