C++ unions vs. reinterpret_cast -

it appears other stackoverflow questions , reading §9.5.1 of iso/iec draft c++ standard standard use of unions literal reinterpret_cast of data undefined behavior.

consider code below. goal take integer value of 0xffff , literally interpret series of bits in ieee 754 floating point. (binary convert shows visually how done.)

#include <iostream> using namespace std;  union uniontype {     int myint;     float myfloat; };  int main() {      int = 0xffff;      uniontype u;     u.myint = i;      cout << "size of int    " << sizeof(int) << endl;     cout << "size of float  " << sizeof(float) << endl;      cout << "myint          " << u.myint << endl;     cout << "myfloat        " << u.myfloat << endl;      float thefloat = *reinterpret_cast<float*>(&i);     cout << "thefloat       " << thefloat << endl;      return 0; } 

the output of code, using both gcc , clang compilers expected.

size of int    4 size of float  4 myint          65535 myfloat        9.18341e-41 thefloat       9.18341e-41 

my question is, standard preclude value of myfloat being deterministic? use of reinterpret_cast better in way perform type of conversion?

the standard states following in §9.5.1:

in union, at 1 of non-static data members can active @ time, is, value of @ 1 of non-static data members can stored in union @ time. [...] size of union sufficient contain largest of non-static data members. each non-static data member allocated if sole member of struct. all non-static data members of union object have same address.

the last sentence, guaranteeing non-static members have same address, seems indicate use of union guaranteed identical use of reinterpret_cast, earlier statement active data members seems preclude guarantee.

so construct more correct?

edit: using intel's icpc compiler, above code produces more interesting results:

$ icpc union.cpp $ ./a.out size of int    4 size of float  4 myint          65535 myfloat        0 thefloat       0 

the reason it's undefined because there's no guarantee value representations of int , float are. c++ standard doesn't float stored ieee 754 single-precision floating point number. should standard treating int object value 0xffff float? doesn't other fact undefined.

practically, however, purpose of reinterpret_cast - tell compiler ignore knows types of objects , trust int float. it's used machine-specific bit-level jiggery-pokery. c++ standard doesn't guarantee once it. @ point, it's understand compiler , machine in situation.

this true both union , reinterpret_cast approaches. suggest reinterpret_cast "better" task, since makes intent clearer. however, keeping code well-defined best approach.