wierd output of fprintf for double variable

[dave539]

I don't do too much programming any more. However, I just got bit by something 'interesting'. I am caculating some percentages for names in an array. It works great IF what I am calculating the percentage for has some value other than 0. On the other hand, if the percentage should be zero what I get is -nan % instead of 0.0 %. I have tried this a couple of different outpout specs including 5.1F, D, and others.

I need a clue (yes that does mean I'm clueless). Does -nan stand for anything?

Thanks.

[jim1944]

"nan" stands for "not a number". It means that the value being converted is not a valid IEEE floating point number.

Jim

[Gareth Jones]

Basically that. "-nan" means "not a number", with the negative sign bit set.

NaN is actually a valid IEEE-fp value, but it has no mathematical value. It's produced when a mathematically undefined operation is performed, such as 0 ÷ 0 (or perhaps -0 ÷ 0 in this case).

Gareth

[jim1944]

Good point. I never saw a NaN that resulted from something like a divide by 0. I've always avoided undefined operations. So, when I got NaN, it was always the result of a memory clobber or failure to initialize. Thus I assumed that was how you got such a thing. Your post prompted me to look it up for fuller definition (http://en.wikipedia.org/wiki/NaN). Interesting.

Thanks
Jim

[dave539]

I first thought of hte memory clobber thing too. A tried adding parrentheses around all sort fo things where the math was done....that didn't help.

BTW, the actual math in my case should be 0.0 / 0.0 I finally left everything the way I started and set the divisor to 0.00000001d and that worked like a champ. So, 0.0 / 0.0000001d equals 0.0 (for me at least).