Floating point error: Overflow !

[gvn]

Hello,

I use Turbo C for compiling firstly

No problem in compiling. The program is about bank interests and have many long calculations.

Can u give some informations about this alert?

/*
Floating point error: Overflow
Abnormal program termination
*/

[quzah]

You either have too big a number, or two many digits of precision. If you're using a 'float', try replacing all instances with 'double' instead.

While you're at it, you probably want to get a real compiler (ie: Not 16bit) just for the hell of it.

Quzah.

[gvn]

with double, all calculations are wrong ! any idea about it?

any documentation or something about overflow?

[quzah]

If you require precision, never use floating point numbers. Floating point numbers by nature are unprecise and have nasty little precision problems due to the way they're stored.

I believe in limits.h or ... can't quite plaec it ... there are limits which define the upper and lower maximums for your floating point numbers. An overflow simply means you're running beyond that number.

Quzah.

[confuted]

yes, the limits are in limits.h

unsigned char is UCHAR_MAX
unsigned short is USHRT_MAX

so float is probably something like FLOAT_MAX

[ygfperson]

Are you dividing by zero anywhere?

[MrWizard]

Originally posted by blackrat364
yes, the limits are in limits.h

unsigned char is UCHAR_MAX
unsigned short is USHRT_MAX

so float is probably something like FLOAT_MAX

Limits.h only contains limits for integral types. You need to look in float.h for floating-point types limits.