A colleague and I were testing how well Flash MX 2004 and .NET work together through web services. There are a few issues with Int64 and Decimal, which are quite worrying, but what really surprised us was an issue with Double. Sending Double.MaxValue was resulting in Infinity being received by Flash.

Now Flash has only one numeric data type called Number and according to the documentation its precision is 1.79769313486232E+308. This translates to 1.79769313486232 * (10 ^ 308), ‘^’ meaning ‘to the power of’. The reason we found this surprising is because according to the documentation Double.MaxValue is exactly the same number, 1.79769313486232E+308, so everything should have worked correctly.

What we forgot was that a Double (as well as Single) have inexact precision. This means that for really large numbers their decimal representation becomes approximate. You can quickly verify this by trying the following code:

      public double GetDouble() {

            return 1.79769313486232e308;

      }

This won’t build and will return the following error message, “Floating-point constant is outside the range of type 'double'”. This is happening because what gets printed out when you use Double.MaxValue and what is printed in the documentation is the approximate maximum value of double. You can even try

      public double GetDouble() {

            return Double.Parse(Double.MaxValue.ToString());

      }

and it won’t work. The real maximum value of Double is 1.7976931348623157E+308. Just try the following

      Console.WriteLine(1.7976931348623157e308 == Double.MaxValue);