A simple bitset implemented in c++.
For the programmer's operation requirements on the bit, the C++ library function <bitset.h> has given a more comprehensive and complete function implementation.
But in the process of actual use, I found that the efficiency of calling <bitset.h> library functions is not optimistic.
So it took me some time myself to achieve this high efficiency goal.
In actual operation, I use unsigned long type as the basic type, 4 bytes to carry 32 bits.
It basically implements the basic operations of the <bitset.h> library and the overloading of operators.
void set(u_long index);
void set(u_long index, bool tag);
void reset();
void reset(u_long index);
bool test(u_long index)const; bitmap& operator&=(const bitmap& b);
bitmap& operator^=(const bitmap& b);
bitmap& operator=(const bitmap& b);
bitmap& operator <<= (u_int pos);
bitmap& operator >>= (u_int pos);
bitmap operator&(const bitmap& b)const;
bitmap operator^(const bitmap& b)const;
bool operator==(const bitmap& b)const;
bool operator!=(const bitmap& b)const;
bitmap operator<<(u_int pos) const;
bitmap operator>>(u_int pos) const;
bool operator[](u_long index)const;
friend ostream& operator << (ostream& os, const bitmap& b); However, in the process of running, there will always be a situation where a breakpoint is triggered at the destructor call. I'm guessing it's because of an out-of-bounds situation when allocating space or modifying values, but I haven't found a bug for a long time. Therefore, in the code I omitted the destructor to make the code work properly.
If anyone can solve this problem, a PR is also welcome.
In the value operation, XOR operation, AND operation, SET operation than the C++ STL library function has a significant performance improvement.
The performance of 500,000 AND operations, XOR operations, and SET operations is as follows.
But there is still a shortcoming in the shift operation, the efficiency is lower than the STL standard library. I guess this is because of the different implementation methods. Each shift will take more time to traverse all the positions. If someone has a better implementation method, PR is also welcome.
