Talk:Floating-Point Exception: Divide by 0, Domain or Overflow

From RAD Studio
Jump to: navigation, search

Completely incorrect descriptions.

First please remove "the result is +INF or -INF" like sentence. These value is alternate value IF FPU masks related floating-point-exceptions. 　This description confused cause and effect.

And please divide three topics for different related math-exceptions.

For EInvalidOp

Title: Floating-Point Exception: Invalid operation

These exceptions result from invalid instructions and abnormal condition for Floating-point calculation. If this exception is masked, the result is depended the hardware implementation. Almost case, NaN (Not a number) is stored.

For EOverflow

Title: Floating-Point Exception: Overflow

These exceptions result from a floating-point operation for Overflow. The result is larger than the target type. such as calculating 10^200 * 10^200 to a double type variable. If this exception is masked, the result is +INF or -INF exactly.

For EZeroDivide

Title: Floating-Point Exception: Divided by 0

These exceptions result from a floating-point operation for Divide by 0. such as 1.0/0.0. If this exception is masked, the result is +INF or -INF exactly.

Also please add the hyper-link to related exception class (for example EInvalidOp)

Retrieved from "http://docwiki.embarcadero.com/RADStudio/Rio/e/index.php?title=Talk:Floating-Point_Exception:_Divide_by_0,_Domain_or_Overflow&oldid=239203"