What's the difference between set<pair> and map in C++?

Question

There are two ways in which I can easily make a key,value attribution in C++ STL: maps and sets of pairs. For instance, I might have

map<key_class,value_class> 

or

set<pair<key_class,value_class> > 

In terms of algorithm complexity and coding style, what are the differences between these usages?

Answer 1

They are semantically different. Consider:

#include <set> #include <map> #include <utility> #include <iostream>  using namespace std;  int main() {   pair<int, int> p1(1, 1);   pair<int, int> p2(1, 2);   set< pair<int, int> > s;   s.insert(p1);   s.insert(p2);   map<int, int> m;   m.insert(p1);   m.insert(p2);   cout << "Set size = " << s.size() << endl;   cout << "Map size = " << m.size() << endl; } 

http://ideone.com/cZ8Vjr

Output:

Set size = 2
Map size = 1

Answer 2

Set elements cannot be modified while they are in the set. set's iterator and const_iterator are equivalent. Therefore, with set<pair<key_class,value_class> >, you cannot modify the value_class in-place. You must remove the old value from the set and add the new value. However, if value_class is a pointer, this doesn't prevent you from modifying the object it points to.

With map<key_class,value_class>, you can modify the value_class in-place, assuming you have a non-const reference to the map.