// cl_float_to_FF_pointer(). // General includes. #include "base/cl_sysdep.h" // Specification. #include "float/ffloat/cl_FF.h" namespace cln { // Implementation. cl_private_thing cl_float_to_FF_pointer (const float x) { var union { ffloat eksplicit; float machine_float; } u; u.machine_float = x; var ffloat val = u.eksplicit; var uintL exp = (val >> FF_mant_len) & (bit(FF_exp_len)-1); // e if (exp == 0) // e=0 ? // vorzeichenbehaftete 0.0 oder subnormale Zahl { if (!((val << 1) == 0) && underflow_allowed()) { throw floating_point_underflow_exception(); } else { return as_cl_private_thing(cl_FF_0); } // +/- 0.0 -> 0.0 } elif (exp == 255) // e=255 ? { if (!((val << (32-FF_mant_len)) == 0)) { throw floating_point_nan_exception(); } // NaN else { throw floating_point_overflow_exception(); } // Infinity, Overflow } else { // Der Exponent muß um FF_exp_mid-126 erhöht werden. if ((FF_exp_mid>126) && (exp > FF_exp_high-FF_exp_mid+126)) { throw floating_point_overflow_exception(); } // Overflow val += (FF_exp_mid - 126) << FF_mant_len; #if defined(CL_WIDE_POINTERS) return as_cl_private_thing(allocate_ffloat(val)); #else return (cl_private_thing)allocate_ffloat(val); #endif } } } // namespace cln