Truncating a number to specific number of decimals

 

Today I encountered a strange problem with a code snippet I wrote and it took a while to figure it out what the issue was. The issue was with using the type "Double" in doing fractional calculations. Our application has a need to truncate a number to a specified number of decimal positions. For this requirement, the existing Math.Truncate(...) function has to be extended to control the number of decimal positions. 

The initial function for the custom Truncate function is as below: 

public static double Truncate(double value, int decimals)

{

    double factor = Math.Pow(10, decimals);

    double result = Math.Truncate(factor * value) / factor;

    return result;

}

When this is unit tested through a Console application, LINQPad and MS Tests, all the attempted tests ran successfully. When this is published for production usage, seldom we noticed that the decimal values were being rounded down slightly deviating from the expected results.

After troubleshooting all other application functions related to the calculations, I found out that the extended Truncate function was the culprit.

In the past, 4-5 years ago I wrote a similar function in VB.Net, for a different project and for a different comany, using the Decimal type. It worked fine without issues. I used Double this time as all other variables in the application were of type Double that uses this function, to avoid casting.

So, I changed the type to Decimal and tested the application again. Now, all the previous results that were showing wrong values showed the correct values as expected.

The final version of the function is as below:

public static decimal Truncate(decimal value, int decimals)

{

    decimal factor = (decimal)Math.Pow(10, decimals);

    decimal result = Math.Truncate(factor * value) / factor;

    return result;

}

… and I changed the overloaded function to deal with the existing code to avoid casting all over the places.

public static double Truncate(double value, int decimals){

    return (double)Truncate((decimal)value, decimals);

}

 

So, whenever you are dealing with calculations where precision is important, the better type to use is Decimal.