Floating-point exceptions require special handling. In general, you cannot retry a floating-point exception without a significant knowledge of both the floating point processor and the application that generated the exception. Because knowledge of your application is beyond the capabilities of the VisualAge for C++ library, it treats a floating-point exception as a terminating condition.
You can prevent floating point exceptions from being reported by masking them. Use the _control87 function to set the floating point bit masks. When a floating-point exception is masked, the floating-point processor performs a predetermined corrective action.
The state of the floating-point control word is unique for each thread, and changing it in one thread of a multithread program does not affect any other thread.
![]()
Signals and Exceptions
Signal and Exception Handling
![]()
Default Exception Handling (OS/2)
Default
Exception Handling (Windows)