1

I have a constructor that creates a matrix from size_t dimensions, I want to add support for int dimensions. However, before passing the ints to the size_t constructor I would like to make sure that they are positive.

 Matrix(vector<double> vals, int rows, int cols ) 
 \\throw something if rows<= 0 || cols<= 0
:Matrix(vals,static_cast<size_t>(rows), static_cast<size_t>(cols))

Is something like this possible?

NoamV
  • 65
  • 5

1 Answers1

2

Pass them through a function,

int enforce_positive(int x) 
{ 
    if (x <= 0) { 
        throw something; 
    } 
    return x; 
}

Matrix(vector<double> vals, int rows, int cols ) 
  : Matrix(vals,
           static_cast<size_t>(enforce_positive(rows)), 
           static_cast<size_t>(enforce_positive(cols)))

You can make this a template, of course.

You could also use a "proof type" and make the condition explicit in the prototype. Something like

temolate<typename T>
class Positive
{
public:
    explicit Positive(T x): val(x)
    {
        if (val <= 0) {
            throw something;
        }
    }
    operator T() const { return val; }
private:
    T val;
};

Matrix(const vector<double>& vals,
       Positive<int> rows,
       Positive<int> cols ) 
  : Matrix(vals,
           static_cast<size_t>(rows), 
           static_cast<size_t>(cols))

//...
int x = ...;
int y = ...;
Matrix m(stuff, Positive(x), Positive(y));
molbdnilo
  • 64,751
  • 3
  • 43
  • 82
  • Why not do the conversion as part of the function? Just change the return type to `size_t` and you'll save one `static_cast`... – fabian Apr 03 '22 at 08:06
  • Because the purpose is to make sure that an `int` is positive, not to convert the type. – molbdnilo Apr 03 '22 at 08:09