Currently, the implementation is hash(pair.first) ^ hash(pair.second) which results in 0 for any pair that satisfies pair.first == pair.second (as @roynecro pointed out in #30).
@roynecro suggests to use hash(concat(hash(pair.first), hash(pair.second))) which will have a performance toll, but will be the most correct.
Currently, the implementation is
hash(pair.first) ^ hash(pair.second)
which results in0
for any pair that satisfiespair.first == pair.second
(as @roynecro pointed out in #30).@roynecro suggests to use
hash(concat(hash(pair.first), hash(pair.second)))
which will have a performance toll, but will be the most correct.