How can I use boost::thread::id as key to an unordered_map?

Question

According to the documentation, a boost::thread::id can be considered unique for each running thread and can be used in containers such as std::set and std::map (because the < operator is overridden for thread::id).

My problem is that I'd like to use thread::id as a key to an boost::unordered_map, however it requires that the key is "hashable" (ie. supports hashing to a size_t). Since all implementation detail for thread::id is hidden I don't think have anything I can use.

So my question is - is it possible to use thread::id as a key to an unordered_map?

Answer 1

The documentation says it can be written to a stream. Write it to a std::ostringstream and hash the str() result. Although the output format is unspecified, it's unique for a given ID and consistent for a given run of your program (which is as long as the thread ID would remain valid anyway).

Answer 2

You can use the streaming ability:

struct Hasher
{
  size_t operator()(const boost::thread::id& id)
  {
    std::ostringstream os; os << id; return hash(os.str());
  }
};

Little excerpt of the class, so that others may see what's possible:

class thread::id
{
public:
    id();

    bool operator==(const id& y) const;
    bool operator!=(const id& y) const;
    bool operator<(const id& y) const;
    bool operator>(const id& y) const;
    bool operator<=(const id& y) const;
    bool operator>=(const id& y) const;

    template<class charT, class traits>
    friend std::basic_ostream<charT, traits>& 
    operator<<(std::basic_ostream<charT, traits>& os, const id& x);
};

Answer 3

How many threads do you have? Unless you have more then several hundred it is unlikely that unordered_map with heavy hash (and hash is heavy especially based on std::stringstream) would be faster then std::map. Don't forger that std::map has log complexity with quite small constant.

And if you have hundreds of threads, then there is probably a problem with your application.