About how to determine whether the unsigned integer addition overflows

Recently, many unsigned integer binary operations have been used in the code, so I am afraid that the addition operation will overflow.

Therefore, it is necessary to add an overflow detection.

About overflow, http://www.phrack.com/issues.html? Issue = 60 & id = 10. This article is very clear.

It is easier to detect the overflow of unsigned integers:

First, in the unsigned representation, A + 2 ^ n =;

If two unsigned integers A and B are not exceeded: A <2 ^ N, B <2 ^ N, and a + B> 2 ^ N,

Then sum = a + B-2 ^ N, that is, Sum-A = B-2 ^ n <0 => sum <A, similarly, sum <B.

This method can be used to determine whether the sum of two unsigned integers overflows. How can we determine the sum of multiple unsigned integers? The stupid method I think of now is to sum up and judge ......

The code above is ugly ......

 1 //! @brief check whether the usinged add will result in overflow 2 template<typename input_iterator> 3 int is_unsinged_add_overflow(input_iterator begin, 4                              input_iterator last) 5 { 6   typedef typename Iterator_Traits<input_iterator>::value_type value_type; 7   if(!(boost::is_same<value_type,size_t>::value) 8       && !(boost::is_same<value_type,unsigned int>::value) 9       && !(boost::is_same<value_type,unsigned long>::value)10       && !(boost::is_same<value_type,unsigned short>::value)){11     return -1;12   }13   std::stack<value_type> temp_sum;14   input_iterator it = begin;15   while(it != last){16     value_type a = *it++;17     if(it == last) {18       temp_sum.push(a);19       break;20     }21     value_type b = *it++;22     value_type sum = a + b;23     if(sum < a || sum < b) return 0; // overflow24     temp_sum.push(sum);25   }26   while(!temp_sum.empty()){27     if(temp_sum.size() == 1) return 1;// no overflow28     std::stack<value_type> forwad_sum;29     while(!temp_sum.empty()){30       value_type a = temp_sum.top();31       temp_sum.pop();32       if(temp_sum.empty()) forwad_sum.push(a);33       else{34         value_type b = temp_sum.top();35         temp_sum.pop();36         value_type sum = a + b;37         if(sum < a || sum < b) return 0; // overflow38         forwad_sum.push(sum);39       }40     }41    swap(temp_sum,forwad_sum);42   }43 }

Troublesome traits, making up your own special features ......

 1 template <typename T> 2 struct Iterator_Traits 3 { 4   typedef typename T::value_type value_type; 5 }; 6  7 template <typename type> 8 struct Iterator_Traits<type*> 9 {10   typedef type value_type;11 };

The code is compiled in Ubuntu 11.04 64bit, G ++-4.5.2, and is poorly written ~