Question

I’m programming in C# using .Net 3.5, trying to control the format in which floats are output as strings.

However, I’m seeing some unexpected behaviour. If my code contains (for example):

float value = 50.8836975;

Edit Sorry, that (deleted) code “sample” was unhelpful. Basically, my question was seeking to explain the results of my debugging statements below when I set a breakpoint after “value” – a C# float – had been assigned the result of a calculation. Jon Skeet‘s answer is exactly what I needed (his first line takes me to task for the unhelpful code).

Then I see the following results when I try various options in my Immediate window:

?value
50.8836975
?value.ToString("G9")
"50.8836975"
?value.ToString("F9")
"50.883700000"

Can anyone explain why my F9-formatted value seems to have lost 3 digits of precision?

Answer 1

Your question is unclear because you’ve given a real value without an f suffix, and tried to assign it to a float variable. If you’re actually using a float variable, then the exact value is

50.883697509765625

If you’re actually using a double variable, then the exact value is:

50.8836974999999966939867590554058551788330078125

I get the same results as you for F9 if you use a float, but not if you use a double.

The reason for the reduced precision is revealed by the documentation for System.Single (float):

By default, a Single value contains only 7 decimal digits of precision, although a maximum of 9 digits is maintained internally.

I believe F is correctly displaying all of the real digits of precision, in an attempt to prevent you from believing that you’ve actually got more information than you have.