Subj : Floating point exceptions
To   : borland.public.cpp.borlandcpp
From : lotter_NO_SpaM_
Date : Mon Jan 17 2005 03:27 pm

Hi all,

I have trouble with floating point exceptions of type 'Invalid
operation' which occure after comparing a NAN with some other value.
I have installed a signal() handler which catches and resolves the
exception (and sets the signal handler again). But after the first
occurance of such an exception, the fpu throws floating point
exceptions even on legal operations. And, the subsequent exception is
not guided to my exception handler. Instead, it is caught by the
debugger.
I have tried to call _clear87() and/or _fpreset() with no success.
Does anybody have an idea what might be going wrong?

Thanks

Erwin Lotter

.