使用unsigned int而不是std :: vector <bool> 或std :: bitset推荐的做法？

Question

I've seen some coding examples that recommend using an unsigned int to represent a bitmap:

unsigned int zero_rows {0};
for (auto i = 0; i < n_rows; ++i) {
  zero_rows |= (1 << i);
  …
}

Does this provide any benefit over using std::vector<bool> :

std::vector<bool> zero_rows(n_rows, false);
for (auto i = 0; i < n_rows; ++i) {
  zero_rows[i] = true;
  …
}

Another option I guess could be std::bitset , but I'm not really sure about the pros and cons of each yet. I'd just like to know what is the recommended practice.

Answer 1

The use of some size of unsigned integer to represent a fixed-size sequence of bits was the only option in C++ before the 1998 Standard, when std::vector<bool> and std::bitset were introduced. The practice was inherited from C, in which it is considered a competent programmer's proficiency and it remains so considered in C++.

std::vector<bool> has come to be regarded with regret. See eg vector<bool> : More Problems, Better Solutions and Effective STL Item 18 . std:::bitset is considered fit for purpose.

The unsigned integer practice inherently represents a hand-rolled simulation of of a fixed-size bit-sequence by overt artifice, constrained and complicated by the fact that there are only a few sizes of unsigned integer (even if the chosen size is made precise). If you require a fixed-size bit-sequence to be subject to operations that are all supported by the interface of std::bitset then other things being equal the Standard Library's provision is to be preferred to hand-rolled code for the all of the reasons to which the Library owes its existence.

You must be the judge whether, in the context of your application, other things are equal.