DN>> You have just discovered why responsible programmers do not use
 DN>> floating point arithmetic for financial applications. It is also one
 DN>> reason why C has not replaced COBOL in the financial applications
 DN>> area.

 CM> Thanks for your response.  I actually discovered this when I dug into
 CM> those old, dusty programming texts, I've got on the shelf.  I've
 CM> worked arounf the problem for now by using doubles instead.  This
 CM> offers enough precision to carry through correctly for my
 CM> application's needs.

It would be far better to use a structure for monetary amounts. (Or a money
class in C++)

 typedef struct
 {
     ULONG dollars;
     ULONG cents;
 } MONEY;
 typedef MONEY *pMONEY;

 Then write you own math and display preperation functions such as:

 AddMoney ( pMONEY, pMONEY, pMONEY);
 SubtractMoney ( pMONEY, pMONEY, pMONEY);
 MultiplyMoney ( pMONEY, pMONEY, pMONEY);
 DivideMoney ( pMONEY, pMONEY, pMONEY);
 FormatMoney ( char*, pMONEY);

It's not hard. Write it one time and use the code in every program you
write that deals with money. You'll never have to worry about floating
point errors creaping into your programs unnoticed. Use MONEY structures to
multiply and divide so you can hande fractions.
 

Carl

--- FleetStreet 0.90wb