FIX: A floating-point exception may occur when an inline function in a C++ application returns an uninitialized floating-point variable in Visual Studio 2005
Symptoms
Consider the following scenario:In Microsoft Visual Studio 2005, you create a C++ application.You do not specify the /fp:except option. Therefore, floating-point exceptions are not enabled in the C++ application.You do not specify the /fp:strict option. Therefore, the strict floating-point model is not enabled in the C++ application.An inline function in the C++ application returns an uninitialized floating-point variable.You compile the C++ application.In this scenario, a floating-point exception may occur when you run the C++ application. Additionally, you may receive the following error message:
0xC0000090: Floating-point invalid operation.Note This problem may occur in debug builds and in release builds.
Resolution
This problem occurs when the uninitialized floating-point variable loads a memory location that contains a signaling not-a-number (NaN) value.
