I've been busy for weeks, hence the lack of post on this weblog. Here's a little one on strange things I noticed about RangeValidator and Double.
A few days ago I was trying to find the cause of an exception in an ASP.NET web application. It happened when I set a RangeValidator's Type to ValidationDataType.Double, then set the MaximumValue to Double.MaxValue.ToString(). I found two things :
-
Apparently, the real maximum value of a Double is 1.7976931348623157E+308. However, Double.MaxValue.ToString() produces 1.79769313486232E+308 which is not parsable with Double.Parse() because it is too big. The cause is that the default formatting for the Double type is 15 digits + exponent whereas the value itself has 17 fractional digits + exponent. When ToString() is called, it rounds the value and the result is a bigger number than the real value. Same goes for Double.Minimum value.
Therefore, when you want to transform a Double to a string so that String.Parse works, use ToString("r"). This formatting ensures that a roundtrip Double->String->Double always works.
-
The RangeValidator uses a regular expression to validate the user input. You can see this with Reflector in the BaseCompareValidator.Convert method. The expression is :
@"^\s*([-\+])?(\d+)?(\" + decimalSeparator + @"(\d+))?\s*$";
As you can see, it does not handle numbers formatted in scientific notation. Therefore, Double.MaxValue produces the exception I was getting : "The value '1.7976931348623157E+308' of the MaximumValue property of 'RangeValidator1' cannot be converted to type 'Double'."
So the only Double values that can be used in a RangeValidator are the ones that can be written in string format without the scientific notation. I didn't check but I doubt you can write all Double numbers that way, there must be a limit to the number of digits in the string representation of a Double.