On Sunday 13 January 2013 00:15:23 Vicente J. Botet Escriba wrote:
Le 12/01/13 19:50, Andrey Semashev a écrit :
On January 12, 2013 9:57:22 PM "Vicente J. Botet Escriba"
<vicente.botet@wanadoo.fr> wrote:
Andrey, the provided patch goes too far for what I was expecting. You have made a lot of changes, C++11 interface has not been preserved, and I'm not sure the algorithm is the same.
The algorithm is different for sure. What exactly do you mean by C++11 interface? Does C++11 have call_once? The only C++11 thing I saw was the once_flag definition which gave no apparent benefits.
C++11 defines the following interface. There is no STD_ONCE_INIT initializer.
struct once_flag { constexpr once_flag() noexcept; once_flag(const once_flag&) = delete; once_flag& operator=(const once_flag&) = delete; }; template<class Callable, class ...Args> void call_once(once_flag& flag, Callable func, Args&&... args);
Could this once_flag be implemented using the trick you used to mask the atomic dependency on the header file?
Hmm, I didn't know the standard defined call_once. It is possible to define once_flag as defined by the standard. But the previous implementation seem incorrect to me. once_flag *must* be either POD or have a constexpr default constructor while the previous code (as well as the code in win32/once.hpp) uses BOOST_CONSTEXPR which can be empty. BOOST_THREAD_PROVIDES_ONCE_CXX11 macro doesn't depend on constexpr or noexcept availability so it can't be used to enable C++11 implementation. So in order to implement C++11 interface there has to be a macro that depends on both constexpr or noexcept availability. Then the once_flag definition should use constexpr and noexcept unconditionally.