Question

Suppose for instance I’m defining a Complex class for representing complex numbers. I would like to define two constructors, so that I can write for example:

Complex z1 = new Complex(x: 4, y: 3);
Complex z2 = new Complex(r: 2, theta: Math.PI / 4);

However, I cannot define the constructors like this:

public Complex(double x, double y) { ... }
public Complex(double r, double theta) { ... }

because both constructors would have the same signature, which is not allowed. But in C# 4 I can write this, using an optional argument:

public Complex(double x, double y) { ... }
public Complex(double r, double theta, bool unused=true) { ... }

It works, I can then use the above constructor calls as intended. The sole purpose of the unused argument is to make the signatures different; it’s totally unused, both when defining and when calling the constructor.

To me this seems to be a an ugly trick: is there any better option?

Answer 1

Make the constructor private and have a static factory style function.

public static Complex CreateComplexPolar(double r, double theta);
public static Complex CreateComplex(double x, double y);

You can do validation on the inputs based on what they should be.

Another possibility would be to create a type that encapsulates the inputs and use constructors as you previously mentioned.

public struct PolarCoordinates
{
  public double Rho;
  public double Theta;
}

public struct CartesianCoordinates
{
  public double X;
  public double Y;
}

public Complex(PolarCoordinates pc);
public Complex(CartesianCoordinates cc);