C++11 compile-time hash of literal strings.

lithash.cpp

#include <cstdint>

namespace detail
{
    // FNV-1a 32bit hashing algorithm.
    constexpr std::uint32_t fnv1a_32(char const* s, std::size_t count)
    {
        return ((count ? fnv1a_32(s, count - 1) : 2166136261u) ^ s[count]) * 16777619u;
    }
}    // namespace detail

constexpr std::uint32_t operator"" _hash(char const* s, std::size_t count)
{
    return detail::fnv1a_32(s, count);
}

constexpr std::uint32_t check = "0123456789ABCDEF"_hash;

int main()
{
    // Compare against precomputed value.
    static_assert(check == 141695047u, "bad hash value");
    return 0;
}

Hi, it is useful. What is license of this gist? MIT, public-domain, LGPL...?

Consider it public domain. Note however that I wrote this code 9 years ago. The 'constexpr' feature has advanced quite a bit since then (e.g. less need for recursion), so would I assume that there's newer fancier ways to do the same thing these days.

for the record, the original impl is somewhat different from the general definition of "hashing a string", as it hashes the trailing zero with the string, and rely on the fact that the underlying string is zero-terminated and 1 more byte longer than count.

It's not wrong though, both are true for UDLs, but one may expect "GET"_hash == 2531704439, while it actually provides 71161173

This is a c++11-complaint version without this problem.

inline constexpr std::uint32_t fnv1a_32(char const *s, std::size_t count) {
	return count ? (fnv1a_32(s, count - 1) ^ s[count - 1]) * 16777619u : 2166136261u;
}

A comprehensive list of general purpose hash functions and their implementations can found here:

https://www.partow.net/programming/hashfunctions/index.html