float to string using roundtrip

Question

I'm trying to convert a float to it's string representation without the result appearing in scientific notation.

I first tried :

ToString("0." + new string('#', 7))

but this doesn't seem to work for large values. For example :

float largeNumber = 12345678f;
string str = largeNumber.ToString("0." + new string('#', 7));

results in "12345680"

I then tried ToString("R")

this works for the large number above, but if the numbers get too large, it displays them in scientific notation. For example 5000000000f results in "5E+09". And small numbers like 0.0005 result in 0.0004999999966

I've also tried mixing the 2, but I still get scientific notation in some cases.

My test program is pasted below. I appreciate that there will be precision issues, but I'm wondering if I can do any better than what I have?

class Program
{
    static void Main(string[] args)
    {
        Write(0.123456789f);
        Write(0.12345678f);
        Write(0.1234567f);
        Write(0.123456f);
        Write(0.12345f);
        Write(0.1234f);
        Write(0.123f);
        Write(0.12f);
        Write(0.1f);
        Write(1);
        Write(12);
        Write(123);
        Write(1234);
        Write(12345);
        Write(123456);
        Write(1234567);
        Write(12345678);
        Write(123456789);

        Console.WriteLine();

        float f = 5000000000f;
        for (int i = 0; i < 17; ++i)
        {
            Write(f);
            f /= 10;
        }

        Console.WriteLine();

        f = 5000000000f;
        for (int i = 0; i < 17; ++i)
        {
            Write(f < 1 ? f + 1 : f);
            f /= 10;
        }

        Console.Read();
    }

    static void Write(float f)
    {
        //string str = f.ToString("0." + new string('#', 7));
        //string str = f.ToString("R");
        string str = Math.Abs(f) < 1 ? f.ToString("0." + new string('#', 7)) : f.ToString("R");

        Console.WriteLine(str);
    }
}

Answer 1

The 0 in the format string is not for zeroes, it represents "digit if non-zero and zero if zero", while '#' represents "digit if non-zero or nothing if zero".

You can use the format f.ToString("0." + new string('#', 7)) for numbers over zero

I just tested it in PowerShell and it works fine for me, although it's probably using doubles or decimals:

PS C:\> $test = "{0:0.0######}"
PS C:\> $test -f 0.5
0,5
PS C:\> $test -f 0.4443423
0,4443423
PS C:\> $test -f 123.4443423
123,4443423
PS C:\> $test -f 45425123.4443423
45425123,4443423

Definitely, the problem seems to be float precision:

PS C:\> $test -f [float]45425123.4443423
45425120,0

Answer 2

try this:

largeNumber.ToString("r")

You can find the list of available formats here: http://msdn.microsoft.com/en-us/library/dwhawy9k%28v=vs.110%29.aspx

Answer 3

The problem is that float only supports 7 digits of precision. There's no way to represent 12345678f precisely in a float, so it gets converted to the nearest representable float value, which is 12345680f. It's not the size of the number, but the number of digits of precision that is the key issue.

Also, 0.0005 cannot be represented exactly in a binary floating-point number; the closest 8-bit binary floating point number to 0.0005 is 0.0004999999966

decimal supports much greater precision, and can represented base-10 numbers precisely using the standard N format specifier.