OK, I know that it's impossible, but it was the best way to formulate the title of the question. The problem is, I'm trying to use my own custom class instead of float (for deterministic simulation) and I want to the syntax to be as close as possible. So, I certainly want to be able to write something like
FixedPoint myNumber = 0.5f;
Is it possible?
Yes, by creating an implicit type cast operator for FixedPoint if this class was written by you.
class FixedPoint
{
public static implicit operator FixedPoint(double d)
{
return new FixedPoint(d);
}
}
If it's not obvious to the reader/coder that a double can be converted to FixedPoint, you may also use an explicit type cast instead. You then have to write:
FixedPoint fp = (FixedPoint) 3.5;
= operator, it'll just introduce a cast. If you'd define it as an explicit cast, you'd have: FixedPoint myNumber = (FixedPoint) 0.5f;. You'll still have a float at first, but it'll be automatically cast to a FixedPoint. Note that this means you'll still have rounding errors from the floating point representation: you can't actually read the literal 0.1f, you'll still receive a floating point number which best represents the binary repeating number 0.1 as argument to your overloaded method.
= operator in a case like this. What Fero is suggesting is they handle the casting of a FixedPoint to another type.Overload implicit cast operator:
class FixedPoint
{
private readonly float _floatField;
public FixedPoint(float field)
{
_floatField = field;
}
public static implicit operator FixedPoint(float f)
{
return new FixedPoint(f);
}
public static implicit operator float(FixedPoint fp)
{
return fp._floatField;
}
}
So you can use:
FixedPoint fp = 1;
float f = fp;
Create an implicit type cast.
This is an example:
<class> instance = new <class>();
float f = instance; // We want to cast instance to float.
public static implicit operator <Predefined Data type> (<Class> instance)
{
//implicit cast logic
return <Predefined Data type>;
}
If the Implicit is not what you want in = overloading, the other option is to use the explicit operator on your class such as below, which one will cast to it, making it understood by the user:
public static explicit operator FixedPoint(float oc)
{
FixedPoint etc = new FixedPoint();
etc._myValue = oc;
return etc;
}
... usage
FixedPoint myNumber = (FixedPoint)0.5f;
decimalis still a floating-point type. It just uses 128 bits instead of 32/64 and a decimal exponent instead of a binary one.