.NET invalid cast exception on double division

Question

I am working on a .NET application to reduce image file sizes using ImageMagick. My application stores some "settings" in a Hashtable for use in various areas of the application. I am now adding a feature to reduce image size based on desired DPI in the transformed image. Basically, I am looking up the DPI in the unaltered image, calculating the percentage of the desired DPI in relation to the current DPI, and then I will resize the image dimensions by this proportion.

ImageMagick reports image DPI as floating point values. So 200x200 DPI is really 199.975x199.975. Thus I use Math.Ceiling() to get the values into my application. When I try to use the desired DPI from my settings Hashtable to do the percentage calculation I get an invalid cast exception. I don't know why this is happening.

Here is a test case that fails in the same manner as my actual code:

using System;
using System.Collections;

namespace typetest
{
    class Program
    {
        private struct DPI {
            public double x;
            public double y;
        }

        public static void Main(string[] args)
        {
            Hashtable vars = new Hashtable();
            DPI dpi;
            string dpiString = "199.547:199.547";
            string[] ret;

            vars["newDpiX"] = 150;
            vars["newDpiY"] = 150;
            ret = dpiString.Split(':');

            dpi.x = ( (double)vars["newDpiX"] / Math.Ceiling(double.Parse(ret[0])) ) * 100;
            dpi.y = ( (double)vars["newDpiY"] / Math.Ceiling(double.Parse(ret[1])) ) * 100;

            Console.WriteLine("New DPI percentage = " + dpi.x + "%x" + dpi.y +"%");
        }
    }
}

If I change the division to

double newDpiX = 150.0;
dpi.x = ( newDpiX / Math.Ceiling(double.Parse(ret[0])) ) * 100;

it will work as expected. What is going on here?

Answer 1

Try

vars["newDpiX"] = 150d;

Answer 2

Use:

dpi.y = ( Convert.ToDouble(vars["newDpiY"]) / Math.Ceiling(double.Parse(ret[1])) ) * 100;

I think your problem is that vars["newDpiY"] = 150; stores an int, not a double.

Answer 3

Hashtable hold objects, you assign 150 to it, so it's stored as an int. Try assigning 150.0 or (double)150, or anything to that effect.

Answer 4

The Hashtable stores Objects. When you put an int in an object, you "box" it. Unboxing requires the same type, you can't convert directly from a boxed int to a double.

This will fail:

object obj = 10;
double dbl = (double)obj;

This will succeed:

object obj = 10d; // a double value, could also have been 10.0
double dbl = (double)obj;

And this will also work:

object obj = 10; // an int
double dbl = (double)(int)obj;

by first unboxing to an int and only then casting to double.

EDIT:

See also http://blogs.msdn.com/ericlippert/archive/2009/03/19/representation-and-identity.aspx