Where to implement the hash function?

Question

I am using an object as a key in an unordered_map , so I need to define a hash function. My question is, where should the hash function be implemented. Should I put it with the class implementation or should I implement it close to where I need it.

UPDATE: If it makes a difference, all of this is based in a framework

Answer 1

If you anticipate you'll need to reuse it in many unordered_map s, put it somewhere visible, like in the class.

If you just need it for a one-off unordered_map , put it close to where you use it. You can even use a lambda .

Answer 2

I'd put it with the class definition, at least if you're using == as the equality function in the unordered_map . The implementation of the hash function depends on the implementation of the equality comparison, and there is a definite advantage in keeping both together, to reduce the probability of someone not changing the hash function if they change == .

If you're also defining a special equality function for the map, then the two functions should be defined together, probably close to where they will be used to instantiate the map.

Answer 3

In my opinion if the hash function is basic as below it should be method of the class and also should be inline.

int hashFunction(long x){
    return (int) (x % N);
}

If it is a little more complex hash function it should be a method of this class.Because you will need a "N" which will be spesific to that class .