Use mingw-std-threads in MinGW builds

COPYRIGHT.txt

@@ -328,6 +328,11 @@ Comment: meshoptimizer
 Copyright: 2016-2022, Arseny Kapoulkine
 License: Expat
 
+Files: ./thirdparty/mingw-std-threads/
+Comment: mingw-std-threads
+Copyright: 2016, Mega Limited
+License: BSD-2-clause
+
 Files: ./thirdparty/minimp3/
 Comment: MiniMP3
 Copyright: lieff

core/math/aabb.cpp

@@ -117,6 +117,11 @@ AABB AABB::intersection(const AABB &p_aabb) const {
 	return AABB(min, max - min);
 }
 
+#ifdef MINGW_ENABLED
+#undef near
+#undef far
+#endif
+
 bool AABB::intersects_ray(const Vector3 &p_from, const Vector3 &p_dir, Vector3 *r_clip, Vector3 *r_normal) const {
 #ifdef MATH_CHECKS
 	if (unlikely(size.x < 0 || size.y < 0 || size.z < 0)) {

core/object/object.h

@@ -656,7 +656,7 @@ private:
 	friend class RefCounted;
 	bool type_is_reference = false;
 
-	std::mutex _instance_binding_mutex;
+	BinaryMutex _instance_binding_mutex;
 	struct InstanceBinding {
 		void *binding = nullptr;
 		void *token = nullptr;

core/os/condition_variable.h

@@ -31,7 +31,14 @@
 #ifndef CONDITION_VARIABLE_H
 #define CONDITION_VARIABLE_H
 
+#ifdef MINGW_ENABLED
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#include "thirdparty/mingw-std-threads/mingw.condition_variable.h"
+#define THREADING_NAMESPACE mingw_stdthread
+#else
 #include <condition_variable>
+#define THREADING_NAMESPACE std
+#endif
 
 // An object one or multiple threads can wait on a be notified by some other.
 // Normally, you want to use a semaphore for such scenarios, but when the
@@ -40,12 +47,12 @@
 // own mutex to tie the wait-notify to some other behavior, you need to use this.
 
 class ConditionVariable {
-	mutable std::condition_variable condition;
+	mutable THREADING_NAMESPACE::condition_variable condition;
 
 public:
 	template <class BinaryMutexT>
 	_ALWAYS_INLINE_ void wait(const MutexLock<BinaryMutexT> &p_lock) const {
-		condition.wait(const_cast<std::unique_lock<std::mutex> &>(p_lock.lock));
+		condition.wait(const_cast<THREADING_NAMESPACE::unique_lock<THREADING_NAMESPACE::mutex> &>(p_lock.lock));
 	}
 
 	_ALWAYS_INLINE_ void notify_one() const {

core/os/mutex.cpp

@@ -40,7 +40,7 @@ void _global_unlock() {
 	_global_mutex.unlock();
 }
 
-template class MutexImpl<std::recursive_mutex>;
-template class MutexImpl<std::mutex>;
-template class MutexLock<MutexImpl<std::recursive_mutex>>;
-template class MutexLock<MutexImpl<std::mutex>>;
+template class MutexImpl<THREADING_NAMESPACE::recursive_mutex>;
+template class MutexImpl<THREADING_NAMESPACE::mutex>;
+template class MutexLock<MutexImpl<THREADING_NAMESPACE::recursive_mutex>>;
+template class MutexLock<MutexImpl<THREADING_NAMESPACE::mutex>>;

core/os/mutex.h

@@ -34,7 +34,14 @@
 #include "core/error/error_macros.h"
 #include "core/typedefs.h"
 
+#ifdef MINGW_ENABLED
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#include "thirdparty/mingw-std-threads/mingw.mutex.h"
+#define THREADING_NAMESPACE mingw_stdthread
+#else
 #include <mutex>
+#define THREADING_NAMESPACE std
+#endif
 
 template <class MutexT>
 class MutexLock;
@@ -73,9 +80,9 @@ template <int Tag>
 class SafeBinaryMutex {
 	friend class MutexLock<SafeBinaryMutex>;
 
-	using StdMutexType = std::mutex;
+	using StdMutexType = THREADING_NAMESPACE::mutex;
 
-	mutable std::mutex mutex;
+	mutable THREADING_NAMESPACE::mutex mutex;
 	static thread_local uint32_t count;
 
 public:
@@ -115,7 +122,7 @@ template <class MutexT>
 class MutexLock {
 	friend class ConditionVariable;
 
-	std::unique_lock<typename MutexT::StdMutexType> lock;
+	THREADING_NAMESPACE::unique_lock<typename MutexT::StdMutexType> lock;
 
 public:
 	_ALWAYS_INLINE_ explicit MutexLock(const MutexT &p_mutex) :
@@ -128,7 +135,7 @@ template <int Tag>
 class MutexLock<SafeBinaryMutex<Tag>> {
 	friend class ConditionVariable;
 
-	std::unique_lock<std::mutex> lock;
+	THREADING_NAMESPACE::unique_lock<THREADING_NAMESPACE::mutex> lock;
 
 public:
 	_ALWAYS_INLINE_ explicit MutexLock(const SafeBinaryMutex<Tag> &p_mutex) :
@@ -140,12 +147,12 @@ public:
 	};
 };
 
-using Mutex = MutexImpl<std::recursive_mutex>; // Recursive, for general use
-using BinaryMutex = MutexImpl<std::mutex>; // Non-recursive, handle with care
+using Mutex = MutexImpl<THREADING_NAMESPACE::recursive_mutex>; // Recursive, for general use
+using BinaryMutex = MutexImpl<THREADING_NAMESPACE::mutex>; // Non-recursive, handle with care
 
-extern template class MutexImpl<std::recursive_mutex>;
-extern template class MutexImpl<std::mutex>;
-extern template class MutexLock<MutexImpl<std::recursive_mutex>>;
-extern template class MutexLock<MutexImpl<std::mutex>>;
+extern template class MutexImpl<THREADING_NAMESPACE::recursive_mutex>;
+extern template class MutexImpl<THREADING_NAMESPACE::mutex>;
+extern template class MutexLock<MutexImpl<THREADING_NAMESPACE::recursive_mutex>>;
+extern template class MutexLock<MutexImpl<THREADING_NAMESPACE::mutex>>;
 
 #endif // MUTEX_H

core/os/os.cpp

@@ -39,7 +39,15 @@
 #include "core/version_generated.gen.h"
 
 #include <stdarg.h>
+
+#ifdef MINGW_ENABLED
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#include "thirdparty/mingw-std-threads/mingw.thread.h"
+#define THREADING_NAMESPACE mingw_stdthread
+#else
 #include <thread>
+#define THREADING_NAMESPACE std
+#endif
 
 OS *OS::singleton = nullptr;
 uint64_t OS::target_ticks = 0;
@@ -359,7 +367,7 @@ String OS::get_unique_id() const {
 }
 
 int OS::get_processor_count() const {
-	return std::thread::hardware_concurrency();
+	return THREADING_NAMESPACE::thread::hardware_concurrency();
 }
 
 String OS::get_processor_name() const {

core/os/rw_lock.h

@@ -33,10 +33,17 @@
 
 #include "core/typedefs.h"
 
+#ifdef MINGW_ENABLED
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#include "thirdparty/mingw-std-threads/mingw.shared_mutex.h"
+#define THREADING_NAMESPACE mingw_stdthread
+#else
 #include <shared_mutex>
+#define THREADING_NAMESPACE std
+#endif
 
 class RWLock {
-	mutable std::shared_timed_mutex mutex;
+	mutable THREADING_NAMESPACE::shared_timed_mutex mutex;
 
 public:
 	// Lock the RWLock, block if locked by someone else.

core/os/semaphore.h

@@ -37,13 +37,21 @@
 #include "core/error/error_macros.h"
 #endif
 
+#ifdef MINGW_ENABLED
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#include "thirdparty/mingw-std-threads/mingw.condition_variable.h"
+#include "thirdparty/mingw-std-threads/mingw.mutex.h"
+#define THREADING_NAMESPACE mingw_stdthread
+#else
 #include <condition_variable>
 #include <mutex>
+#define THREADING_NAMESPACE std
+#endif
 
 class Semaphore {
 private:
-	mutable std::mutex mutex;
-	mutable std::condition_variable condition;
+	mutable THREADING_NAMESPACE::mutex mutex;
+	mutable THREADING_NAMESPACE::condition_variable condition;
 	mutable uint32_t count = 0; // Initialized as locked.
 #ifdef DEBUG_ENABLED
 	mutable uint32_t awaiters = 0;
@@ -57,7 +65,7 @@ public:
 	}
 
 	_ALWAYS_INLINE_ void wait() const {
-		std::unique_lock lock(mutex);
+		THREADING_NAMESPACE::unique_lock lock(mutex);
 #ifdef DEBUG_ENABLED
 		++awaiters;
 #endif
@@ -116,7 +124,7 @@ public:
 					"A Semaphore object is being destroyed while one or more threads are still waiting on it.\n"
 					"Please call post() on it as necessary to prevent such a situation and so ensure correct cleanup.");
 			// And now, the hacky countermeasure (i.e., leak the condition variable).
-			new (&condition) std::condition_variable();
+			new (&condition) THREADING_NAMESPACE::condition_variable();
 		}
 	}
 #endif

core/os/thread.cpp

@@ -69,8 +69,7 @@ void Thread::callback(ID p_caller_id, const Settings &p_settings, Callback p_cal
 Thread::ID Thread::start(Thread::Callback p_callback, void *p_user, const Settings &p_settings) {
 	ERR_FAIL_COND_V_MSG(id != UNASSIGNED_ID, UNASSIGNED_ID, "A Thread object has been re-started without wait_to_finish() having been called on it.");
 	id = id_counter.increment();
-	std::thread new_thread(&Thread::callback, id, p_settings, p_callback, p_user);
-	thread.swap(new_thread);
+	thread = THREADING_NAMESPACE::thread(&Thread::callback, id, p_settings, p_callback, p_user);
 	return id;
 }
 
@@ -82,8 +81,7 @@ void Thread::wait_to_finish() {
 	ERR_FAIL_COND_MSG(id == UNASSIGNED_ID, "Attempt of waiting to finish on a thread that was never started.");
 	ERR_FAIL_COND_MSG(id == get_caller_id(), "Threads can't wait to finish on themselves, another thread must wait.");
 	thread.join();
-	std::thread empty_thread;
-	thread.swap(empty_thread);
+	thread = THREADING_NAMESPACE::thread();
 	id = UNASSIGNED_ID;
 }
 

core/os/thread.h

@@ -42,7 +42,14 @@
 #include "core/templates/safe_refcount.h"
 #include "core/typedefs.h"
 
+#ifdef MINGW_ENABLED
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#include "thirdparty/mingw-std-threads/mingw.thread.h"
+#define THREADING_NAMESPACE mingw_stdthread
+#else
 #include <thread>
+#define THREADING_NAMESPACE std
+#endif
 
 class String;
 
@@ -82,7 +89,7 @@ private:
 	ID id = UNASSIGNED_ID;
 	static SafeNumeric<uint64_t> id_counter;
 	static thread_local ID caller_id;
-	std::thread thread;
+	THREADING_NAMESPACE::thread thread;
 
 	static void callback(ID p_caller_id, const Settings &p_settings, Thread::Callback p_callback, void *p_userdata);
 

editor/dependency_editor.h

@@ -68,6 +68,10 @@ public:
 	DependencyEditor();
 };
 
+#ifdef MINGW_ENABLED
+#undef FILE_OPEN
+#endif
+
 class DependencyEditorOwners : public AcceptDialog {
 	GDCLASS(DependencyEditorOwners, AcceptDialog);
 

editor/plugins/script_editor_plugin.h

@@ -204,6 +204,10 @@ class EditorScriptCodeCompletionCache;
 class FindInFilesDialog;
 class FindInFilesPanel;
 
+#ifdef MINGW_ENABLED
+#undef FILE_OPEN
+#endif
+
 class ScriptEditor : public PanelContainer {
 	GDCLASS(ScriptEditor, PanelContainer);
 

editor/plugins/shader_editor_plugin.h

@@ -42,6 +42,10 @@ class TextShaderEditor;
 class VisualShaderEditor;
 class WindowWrapper;
 
+#ifdef MINGW_ENABLED
+#undef FILE_OPEN
+#endif
+
 class ShaderEditorPlugin : public EditorPlugin {
 	GDCLASS(ShaderEditorPlugin, EditorPlugin);
 

editor/plugins/visual_shader_editor_plugin.h

@@ -260,6 +260,10 @@ class VisualShaderEditor : public VBoxContainer {
 		COLLAPSE_ALL
 	};
 
+#ifdef MINGW_ENABLED
+#undef DELETE
+#endif
+
 	enum NodeMenuOptions {
 		ADD,
 		SEPARATOR, // ignore

modules/gdscript/gdscript_tokenizer.h

@@ -37,6 +37,12 @@
 #include "core/templates/vector.h"
 #include "core/variant/variant.h"
 
+#ifdef MINGW_ENABLED
+#undef CONST
+#undef IN
+#undef VOID
+#endif
+
 class GDScriptTokenizer {
 public:
 	enum CursorPlace {

modules/gltf/editor/editor_scene_importer_blend.cpp

@@ -45,6 +45,11 @@
 #include "main/main.h"
 #include "scene/gui/line_edit.h"
 
+#ifdef MINGW_ENABLED
+#define near
+#define far
+#endif
+
 #ifdef WINDOWS_ENABLED
 #include <shlwapi.h>
 #endif

scene/3d/camera_3d.h

@@ -36,6 +36,11 @@
 #include "scene/resources/camera_attributes.h"
 #include "scene/resources/environment.h"
 
+#ifdef MINGW_ENABLED
+#undef near
+#undef far
+#endif
+
 class Camera3D : public Node3D {
 	GDCLASS(Camera3D, Node3D);
 

scene/debugger/scene_debugger.cpp

@@ -72,6 +72,11 @@ void SceneDebugger::deinitialize() {
 	}
 }
 
+#ifdef MINGW_ENABLED
+#undef near
+#undef far
+#endif
+
 #ifdef DEBUG_ENABLED
 Error SceneDebugger::parse_message(void *p_user, const String &p_msg, const Array &p_args, bool &r_captured) {
 	SceneTree *scene_tree = SceneTree::get_singleton();

scene/resources/camera_attributes.cpp

@@ -373,6 +373,11 @@ real_t CameraAttributesPhysical::get_fov() const {
 	return frustum_fov;
 }
 
+#ifdef MINGW_ENABLED
+#undef near
+#undef far
+#endif
+
 void CameraAttributesPhysical::_update_frustum() {
 	//https://en.wikipedia.org/wiki/Circle_of_confusion#Circle_of_confusion_diameter_limit_based_on_d/1500
 	Vector2i sensor_size = Vector2i(36, 24); // Matches high-end DSLR, could be made variable if there is demand.

servers/rendering/renderer_scene_render.cpp

@@ -47,6 +47,11 @@ void RendererSceneRender::CameraData::set_camera(const Transform3D p_transform,
 	taa_jitter = p_taa_jitter;
 }
 
+#ifdef MINGW_ENABLED
+#undef near
+#undef far
+#endif
+
 void RendererSceneRender::CameraData::set_multiview_camera(uint32_t p_view_count, const Transform3D *p_transforms, const Projection *p_projections, bool p_is_orthogonal, bool p_vaspect) {
 	ERR_FAIL_COND_MSG(p_view_count != 2, "Incorrect view count for stereoscopic view");
 

thirdparty/README.md

@@ -515,6 +515,24 @@ error metrics instead of a combination of distance and attribute errors. Patches
 for both changes can be found in the `patches` directory.
 
 
+## mingw-std-threads
+
+- Upstream: https://github.com/meganz/mingw-std-threads
+- Version: git (c931bac289dd431f1dd30fc4a5d1a7be36668073, 2023)
+- License: BSD-2-clause
+
+Files extracted from upstream repository:
+
+- `LICENSE`
+- `mingw.condition_variable.h`
+- `mingw.invoke.h`
+- `mingw.mutex.h`
+- `mingw.shared_mutex.h`
+- `mingw.thread.h`
+
+Once copied, apply `no_except.patch` (needed because Godot is built without exceptions).
+
+
 ## minimp3
 
 - Upstream: https://github.com/lieff/minimp3

thirdparty/mingw-std-threads/LICENSE

@@ -0,0 +1,24 @@
+Copyright (c) 2016, Mega Limited
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+

thirdparty/mingw-std-threads/mingw.condition_variable.h

@@ -0,0 +1,564 @@
+/**
+* @file condition_variable.h
+* @brief std::condition_variable implementation for MinGW
+*
+* (c) 2013-2016 by Mega Limited, Auckland, New Zealand
+* @author Alexander Vassilev
+*
+* @copyright Simplified (2-clause) BSD License.
+* You should have received a copy of the license along with this
+* program.
+*
+* This code is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+* @note
+* This file may become part of the mingw-w64 runtime package. If/when this happens,
+* the appropriate license will be added, i.e. this code will become dual-licensed,
+* and the current BSD 2-clause license will stay.
+*/
+
+#ifndef MINGW_CONDITIONAL_VARIABLE_H
+#define MINGW_CONDITIONAL_VARIABLE_H
+
+#if !defined(__cplusplus) || (__cplusplus < 201103L)
+#error A C++11 compiler is required!
+#endif
+//  Use the standard classes for std::, if available.
+#include <condition_variable>
+
+#include <cassert>
+#include <chrono>
+#include <system_error>
+
+#include <sdkddkver.h>  //  Detect Windows version.
+#if (WINVER < _WIN32_WINNT_VISTA)
+#include <atomic>
+#endif
+#if (defined(__MINGW32__) && !defined(__MINGW64_VERSION_MAJOR))
+#pragma message "The Windows API that MinGW-w32 provides is not fully compatible\
+ with Microsoft's API. We'll try to work around this, but we can make no\
+ guarantees. This problem does not exist in MinGW-w64."
+#include <windows.h>    //  No further granularity can be expected.
+#else
+#if (WINVER < _WIN32_WINNT_VISTA)
+#include <windef.h>
+#include <winbase.h>  //  For CreateSemaphore
+#include <handleapi.h>
+#endif
+#include <synchapi.h>
+#endif
+
+#include "mingw.mutex.h"
+#include "mingw.shared_mutex.h"
+
+#if !defined(_WIN32_WINNT) || (_WIN32_WINNT < 0x0501)
+#error To use the MinGW-std-threads library, you will need to define the macro _WIN32_WINNT to be 0x0501 (Windows XP) or higher.
+#endif
+
+namespace mingw_stdthread
+{
+#if defined(__MINGW32__ ) && !defined(_GLIBCXX_HAS_GTHREADS)
+enum class cv_status { no_timeout, timeout };
+#else
+using std::cv_status;
+#endif
+namespace xp
+{
+//    Include the XP-compatible condition_variable classes only if actually
+//  compiling for XP. The XP-compatible classes are slower than the newer
+//  versions, and depend on features not compatible with Windows Phone 8.
+#if (WINVER < _WIN32_WINNT_VISTA)
+class condition_variable_any
+{
+    recursive_mutex mMutex {};
+    std::atomic<int> mNumWaiters {0};
+    HANDLE mSemaphore;
+    HANDLE mWakeEvent {};
+public:
+    using native_handle_type = HANDLE;
+    native_handle_type native_handle()
+    {
+        return mSemaphore;
+    }
+    condition_variable_any(const condition_variable_any&) = delete;
+    condition_variable_any& operator=(const condition_variable_any&) = delete;
+    condition_variable_any()
+        :   mSemaphore(CreateSemaphoreA(NULL, 0, 0xFFFF, NULL))
+    {
+        if (mSemaphore == NULL)
+            __builtin_trap();
+        mWakeEvent = CreateEvent(NULL, FALSE, FALSE, NULL);
+        if (mWakeEvent == NULL)
+        {
+            CloseHandle(mSemaphore);
+            __builtin_trap();
+        }
+    }
+    ~condition_variable_any()
+    {
+        CloseHandle(mWakeEvent);
+        CloseHandle(mSemaphore);
+    }
+private:
+    template <class M>
+    bool wait_impl(M& lock, DWORD timeout)
+    {
+        {
+            lock_guard<recursive_mutex> guard(mMutex);
+            mNumWaiters++;
+        }
+        lock.unlock();
+        DWORD ret = WaitForSingleObject(mSemaphore, timeout);
+
+        mNumWaiters--;
+        SetEvent(mWakeEvent);
+        lock.lock();
+        if (ret == WAIT_OBJECT_0)
+            return true;
+        else if (ret == WAIT_TIMEOUT)
+            return false;
+//2 possible cases:
+//1)The point in notify_all() where we determine the count to
+//increment the semaphore with has not been reached yet:
+//we just need to decrement mNumWaiters, but setting the event does not hurt
+//
+//2)Semaphore has just been released with mNumWaiters just before
+//we decremented it. This means that the semaphore count
+//after all waiters finish won't be 0 - because not all waiters
+//woke up by acquiring the semaphore - we woke up by a timeout.
+//The notify_all() must handle this gracefully
+//
+        else
+        {
+            using namespace std;
+            __builtin_trap();
+        }
+    }
+public:
+    template <class M>
+    void wait(M& lock)
+    {
+        wait_impl(lock, INFINITE);
+    }
+    template <class M, class Predicate>
+    void wait(M& lock, Predicate pred)
+    {
+        while(!pred())
+        {
+            wait(lock);
+        };
+    }
+
+    void notify_all() noexcept
+    {
+        lock_guard<recursive_mutex> lock(mMutex); //block any further wait requests until all current waiters are unblocked
+        if (mNumWaiters.load() <= 0)
+            return;
+
+        ReleaseSemaphore(mSemaphore, mNumWaiters, NULL);
+        while(mNumWaiters > 0)
+        {
+            auto ret = WaitForSingleObject(mWakeEvent, 1000);
+            if (ret == WAIT_FAILED || ret == WAIT_ABANDONED)
+                std::terminate();
+        }
+        assert(mNumWaiters == 0);
+//in case some of the waiters timed out just after we released the
+//semaphore by mNumWaiters, it won't be zero now, because not all waiters
+//woke up by acquiring the semaphore. So we must zero the semaphore before
+//we accept waiters for the next event
+//See _wait_impl for details
+        while(WaitForSingleObject(mSemaphore, 0) == WAIT_OBJECT_0);
+    }
+    void notify_one() noexcept
+    {
+        lock_guard<recursive_mutex> lock(mMutex);
+        int targetWaiters = mNumWaiters.load() - 1;
+        if (targetWaiters <= -1)
+            return;
+        ReleaseSemaphore(mSemaphore, 1, NULL);
+        while(mNumWaiters > targetWaiters)
+        {
+            auto ret = WaitForSingleObject(mWakeEvent, 1000);
+            if (ret == WAIT_FAILED || ret == WAIT_ABANDONED)
+                std::terminate();
+        }
+        assert(mNumWaiters == targetWaiters);
+    }
+    template <class M, class Rep, class Period>
+    cv_status wait_for(M& lock,
+                       const std::chrono::duration<Rep, Period>& rel_time)
+    {
+        using namespace std::chrono;
+        auto timeout = duration_cast<milliseconds>(rel_time).count();
+        DWORD waittime = (timeout < INFINITE) ? ((timeout < 0) ? 0 : static_cast<DWORD>(timeout)) : (INFINITE - 1);
+        bool ret = wait_impl(lock, waittime) || (timeout >= INFINITE);
+        return ret?cv_status::no_timeout:cv_status::timeout;
+    }
+
+    template <class M, class Rep, class Period, class Predicate>
+    bool wait_for(M& lock,
+                  const std::chrono::duration<Rep, Period>& rel_time, Predicate pred)
+    {
+        return wait_until(lock, std::chrono::steady_clock::now()+rel_time, pred);
+    }
+    template <class M, class Clock, class Duration>
+    cv_status wait_until (M& lock,
+                          const std::chrono::time_point<Clock,Duration>& abs_time)
+    {
+        return wait_for(lock, abs_time - Clock::now());
+    }
+    template <class M, class Clock, class Duration, class Predicate>
+    bool wait_until (M& lock,
+                     const std::chrono::time_point<Clock, Duration>& abs_time,
+                     Predicate pred)
+    {
+        while (!pred())
+        {
+            if (wait_until(lock, abs_time) == cv_status::timeout)
+            {
+                return pred();
+            }
+        }
+        return true;
+    }
+};
+class condition_variable: condition_variable_any
+{
+    using base = condition_variable_any;
+public:
+    using base::native_handle_type;
+    using base::native_handle;
+    using base::base;
+    using base::notify_all;
+    using base::notify_one;
+    void wait(unique_lock<mutex> &lock)
+    {
+        base::wait(lock);
+    }
+    template <class Predicate>
+    void wait(unique_lock<mutex>& lock, Predicate pred)
+    {
+        base::wait(lock, pred);
+    }
+    template <class Rep, class Period>
+    cv_status wait_for(unique_lock<mutex>& lock, const std::chrono::duration<Rep, Period>& rel_time)
+    {
+        return base::wait_for(lock, rel_time);
+    }
+    template <class Rep, class Period, class Predicate>
+    bool wait_for(unique_lock<mutex>& lock, const std::chrono::duration<Rep, Period>& rel_time, Predicate pred)
+    {
+        return base::wait_for(lock, rel_time, pred);
+    }
+    template <class Clock, class Duration>
+    cv_status wait_until (unique_lock<mutex>& lock, const std::chrono::time_point<Clock,Duration>& abs_time)
+    {
+        return base::wait_until(lock, abs_time);
+    }
+    template <class Clock, class Duration, class Predicate>
+    bool wait_until (unique_lock<mutex>& lock, const std::chrono::time_point<Clock, Duration>& abs_time, Predicate pred)
+    {
+        return base::wait_until(lock, abs_time, pred);
+    }
+};
+#endif  //  Compiling for XP
+} //  Namespace mingw_stdthread::xp
+
+#if (WINVER >= _WIN32_WINNT_VISTA)
+namespace vista
+{
+//  If compiling for Vista or higher, use the native condition variable.
+class condition_variable
+{
+    static constexpr DWORD kInfinite = 0xffffffffl;
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wzero-as-null-pointer-constant"
+    CONDITION_VARIABLE cvariable_ = CONDITION_VARIABLE_INIT;
+#pragma GCC diagnostic pop
+
+    friend class condition_variable_any;
+
+#if STDMUTEX_RECURSION_CHECKS
+    template<typename MTX>
+    inline static void before_wait (MTX * pmutex)
+    {
+        pmutex->mOwnerThread.checkSetOwnerBeforeUnlock();
+    }
+    template<typename MTX>
+    inline static void after_wait (MTX * pmutex)
+    {
+        pmutex->mOwnerThread.setOwnerAfterLock(GetCurrentThreadId());
+    }
+#else
+    inline static void before_wait (void *) { }
+    inline static void after_wait (void *) { }
+#endif
+
+    bool wait_impl (unique_lock<xp::mutex> & lock, DWORD time)
+    {
+        using mutex_handle_type = typename xp::mutex::native_handle_type;
+        static_assert(std::is_same<mutex_handle_type, PCRITICAL_SECTION>::value,
+                      "Native Win32 condition variable requires std::mutex to \
+use native Win32 critical section objects.");
+        xp::mutex * pmutex = lock.release();
+        before_wait(pmutex);
+        BOOL success = SleepConditionVariableCS(&cvariable_,
+                                                pmutex->native_handle(),
+                                                time);
+        after_wait(pmutex);
+        lock = unique_lock<xp::mutex>(*pmutex, adopt_lock);
+        return success;
+    }
+
+    bool wait_unique (windows7::mutex * pmutex, DWORD time)
+    {
+        before_wait(pmutex);
+        BOOL success = SleepConditionVariableSRW( native_handle(),
+                                                  pmutex->native_handle(),
+                                                  time,
+//    CONDITION_VARIABLE_LOCKMODE_SHARED has a value not specified by
+//  Microsoft's Dev Center, but is known to be (convertible to) a ULONG. To
+//  ensure that the value passed to this function is not equal to Microsoft's
+//  constant, we can either use a static_assert, or simply generate an
+//  appropriate value.
+                                           !CONDITION_VARIABLE_LOCKMODE_SHARED);
+        after_wait(pmutex);
+        return success;
+    }
+    bool wait_impl (unique_lock<windows7::mutex> & lock, DWORD time)
+    {
+        windows7::mutex * pmutex = lock.release();
+        bool success = wait_unique(pmutex, time);
+        lock = unique_lock<windows7::mutex>(*pmutex, adopt_lock);
+        return success;
+    }
+public:
+    using native_handle_type = PCONDITION_VARIABLE;
+    native_handle_type native_handle (void)
+    {
+        return &cvariable_;
+    }
+
+    condition_variable (void) = default;
+    ~condition_variable (void) = default;
+
+    condition_variable (const condition_variable &) = delete;
+    condition_variable & operator= (const condition_variable &) = delete;
+
+    void notify_one (void) noexcept
+    {
+        WakeConditionVariable(&cvariable_);
+    }
+
+    void notify_all (void) noexcept
+    {
+        WakeAllConditionVariable(&cvariable_);
+    }
+
+    void wait (unique_lock<mutex> & lock)
+    {
+        wait_impl(lock, kInfinite);
+    }
+
+    template<class Predicate>
+    void wait (unique_lock<mutex> & lock, Predicate pred)
+    {
+        while (!pred())
+            wait(lock);
+    }
+
+    template <class Rep, class Period>
+    cv_status wait_for(unique_lock<mutex>& lock,
+                       const std::chrono::duration<Rep, Period>& rel_time)
+    {
+        using namespace std::chrono;
+        auto timeout = duration_cast<milliseconds>(rel_time).count();
+        DWORD waittime = (timeout < kInfinite) ? ((timeout < 0) ? 0 : static_cast<DWORD>(timeout)) : (kInfinite - 1);
+        bool result = wait_impl(lock, waittime) || (timeout >= kInfinite);
+        return result ? cv_status::no_timeout : cv_status::timeout;
+    }
+
+    template <class Rep, class Period, class Predicate>
+    bool wait_for(unique_lock<mutex>& lock,
+                  const std::chrono::duration<Rep, Period>& rel_time,
+                  Predicate pred)
+    {
+        return wait_until(lock,
+                          std::chrono::steady_clock::now() + rel_time,
+                          std::move(pred));
+    }
+    template <class Clock, class Duration>
+    cv_status wait_until (unique_lock<mutex>& lock,
+                          const std::chrono::time_point<Clock,Duration>& abs_time)
+    {
+        return wait_for(lock, abs_time - Clock::now());
+    }
+    template <class Clock, class Duration, class Predicate>
+    bool wait_until  (unique_lock<mutex>& lock,
+                      const std::chrono::time_point<Clock, Duration>& abs_time,
+                      Predicate pred)
+    {
+        while (!pred())
+        {
+            if (wait_until(lock, abs_time) == cv_status::timeout)
+            {
+                return pred();
+            }
+        }
+        return true;
+    }
+};
+
+class condition_variable_any
+{
+    static constexpr DWORD kInfinite = 0xffffffffl;
+    using native_shared_mutex = windows7::shared_mutex;
+
+    condition_variable internal_cv_ {};
+//    When available, the SRW-based mutexes should be faster than the
+//  CriticalSection-based mutexes. Only try_lock will be unavailable in Vista,
+//  and try_lock is not used by condition_variable_any.
+    windows7::mutex internal_mutex_ {};
+
+    template<class L>
+    bool wait_impl (L & lock, DWORD time)
+    {
+        unique_lock<decltype(internal_mutex_)> internal_lock(internal_mutex_);
+        lock.unlock();
+        bool success = internal_cv_.wait_impl(internal_lock, time);
+        lock.lock();
+        return success;
+    }
+//    If the lock happens to be called on a native Windows mutex, skip any extra
+//  contention.
+    inline bool wait_impl (unique_lock<mutex> & lock, DWORD time)
+    {
+        return internal_cv_.wait_impl(lock, time);
+    }
+//    Some shared_mutex functionality is available even in Vista, but it's not
+//  until Windows 7 that a full implementation is natively possible. The class
+//  itself is defined, with missing features, at the Vista feature level.
+    bool wait_impl (unique_lock<native_shared_mutex> & lock, DWORD time)
+    {
+        native_shared_mutex * pmutex = lock.release();
+        bool success = internal_cv_.wait_unique(pmutex, time);
+        lock = unique_lock<native_shared_mutex>(*pmutex, adopt_lock);
+        return success;
+    }
+    bool wait_impl (shared_lock<native_shared_mutex> & lock, DWORD time)
+    {
+        native_shared_mutex * pmutex = lock.release();
+        BOOL success = SleepConditionVariableSRW(native_handle(),
+                       pmutex->native_handle(), time,
+                       CONDITION_VARIABLE_LOCKMODE_SHARED);
+        lock = shared_lock<native_shared_mutex>(*pmutex, adopt_lock);
+        return success;
+    }
+public:
+    using native_handle_type = typename condition_variable::native_handle_type;
+
+    native_handle_type native_handle (void)
+    {
+        return internal_cv_.native_handle();
+    }
+
+    void notify_one (void) noexcept
+    {
+        internal_cv_.notify_one();
+    }
+
+    void notify_all (void) noexcept
+    {
+        internal_cv_.notify_all();
+    }
+
+    condition_variable_any (void) = default;
+    ~condition_variable_any (void) = default;
+
+    template<class L>
+    void wait (L & lock)
+    {
+        wait_impl(lock, kInfinite);
+    }
+
+    template<class L, class Predicate>
+    void wait (L & lock, Predicate pred)
+    {
+        while (!pred())
+            wait(lock);
+    }
+
+    template <class L, class Rep, class Period>
+    cv_status wait_for(L& lock, const std::chrono::duration<Rep,Period>& period)
+    {
+        using namespace std::chrono;
+        auto timeout = duration_cast<milliseconds>(period).count();
+        DWORD waittime = (timeout < kInfinite) ? ((timeout < 0) ? 0 : static_cast<DWORD>(timeout)) : (kInfinite - 1);
+        bool result = wait_impl(lock, waittime) || (timeout >= kInfinite);
+        return result ? cv_status::no_timeout : cv_status::timeout;
+    }
+
+    template <class L, class Rep, class Period, class Predicate>
+    bool wait_for(L& lock, const std::chrono::duration<Rep, Period>& period,
+                  Predicate pred)
+    {
+        return wait_until(lock, std::chrono::steady_clock::now() + period,
+                          std::move(pred));
+    }
+    template <class L, class Clock, class Duration>
+    cv_status wait_until (L& lock,
+                          const std::chrono::time_point<Clock,Duration>& abs_time)
+    {
+        return wait_for(lock, abs_time - Clock::now());
+    }
+    template <class L, class Clock, class Duration, class Predicate>
+    bool wait_until  (L& lock,
+                      const std::chrono::time_point<Clock, Duration>& abs_time,
+                      Predicate pred)
+    {
+        while (!pred())
+        {
+            if (wait_until(lock, abs_time) == cv_status::timeout)
+            {
+                return pred();
+            }
+        }
+        return true;
+    }
+};
+} //  Namespace vista
+#endif
+#if WINVER < 0x0600
+using xp::condition_variable;
+using xp::condition_variable_any;
+#else
+using vista::condition_variable;
+using vista::condition_variable_any;
+#endif
+} //  Namespace mingw_stdthread
+
+//  Push objects into std, but only if they are not already there.
+namespace std
+{
+//    Because of quirks of the compiler, the common "using namespace std;"
+//  directive would flatten the namespaces and introduce ambiguity where there
+//  was none. Direct specification (std::), however, would be unaffected.
+//    Take the safe option, and include only in the presence of MinGW's win32
+//  implementation.
+#if defined(__MINGW32__ ) && !defined(_GLIBCXX_HAS_GTHREADS)
+using mingw_stdthread::cv_status;
+using mingw_stdthread::condition_variable;
+using mingw_stdthread::condition_variable_any;
+#elif !defined(MINGW_STDTHREAD_REDUNDANCY_WARNING)  //  Skip repetition
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#pragma message "This version of MinGW seems to include a win32 port of\
+ pthreads, and probably already has C++11 std threading classes implemented,\
+ based on pthreads. These classes, found in namespace std, are not overridden\
+ by the mingw-std-thread library. If you would still like to use this\
+ implementation (as it is more lightweight), use the classes provided in\
+ namespace mingw_stdthread."
+#endif
+}
+#endif // MINGW_CONDITIONAL_VARIABLE_H

thirdparty/mingw-std-threads/mingw.invoke.h

@@ -0,0 +1,109 @@
+/// \file mingw.invoke.h
+/// \brief Lightweight `invoke` implementation, for C++11 and C++14.
+///
+/// (c) 2018-2019 by Nathaniel J. McClatchey, San Jose, CA, United States
+/// \author Nathaniel J. McClatchey, PhD
+///
+/// \copyright Simplified (2-clause) BSD License.
+///
+/// \note This file may become part of the mingw-w64 runtime package. If/when
+/// this happens, the appropriate license will be added, i.e. this code will
+/// become dual-licensed, and the current BSD 2-clause license will stay.
+
+#ifndef MINGW_INVOKE_H_
+#define MINGW_INVOKE_H_
+
+#include <type_traits>  //  For std::result_of, etc.
+#include <utility>      //  For std::forward
+#include <functional>   //  For std::reference_wrapper
+
+namespace mingw_stdthread
+{
+namespace detail
+{
+//  For compatibility, implement std::invoke for C++11 and C++14
+#if __cplusplus < 201703L
+  template<bool PMemFunc, bool PMemData>
+  struct Invoker
+  {
+    template<class F, class... Args>
+    inline static typename std::result_of<F(Args...)>::type invoke (F&& f, Args&&... args)
+    {
+      return std::forward<F>(f)(std::forward<Args>(args)...);
+    }
+  };
+  template<bool>
+  struct InvokerHelper;
+
+  template<>
+  struct InvokerHelper<false>
+  {
+    template<class T1>
+    inline static auto get (T1&& t1) -> decltype(*std::forward<T1>(t1))
+    {
+      return *std::forward<T1>(t1);
+    }
+
+    template<class T1>
+    inline static auto get (const std::reference_wrapper<T1>& t1) -> decltype(t1.get())
+    {
+      return t1.get();
+    }
+  };
+
+  template<>
+  struct InvokerHelper<true>
+  {
+    template<class T1>
+    inline static auto get (T1&& t1) -> decltype(std::forward<T1>(t1))
+    {
+      return std::forward<T1>(t1);
+    }
+  };
+
+  template<>
+  struct Invoker<true, false>
+  {
+    template<class T, class F, class T1, class... Args>
+    inline static auto invoke (F T::* f, T1&& t1, Args&&... args) ->\
+      decltype((InvokerHelper<std::is_base_of<T,typename std::decay<T1>::type>::value>::get(std::forward<T1>(t1)).*f)(std::forward<Args>(args)...))
+    {
+      return (InvokerHelper<std::is_base_of<T,typename std::decay<T1>::type>::value>::get(std::forward<T1>(t1)).*f)(std::forward<Args>(args)...);
+    }
+  };
+
+  template<>
+  struct Invoker<false, true>
+  {
+    template<class T, class F, class T1, class... Args>
+    inline static auto invoke (F T::* f, T1&& t1, Args&&... args) ->\
+      decltype(InvokerHelper<std::is_base_of<T,typename std::decay<T1>::type>::value>::get(t1).*f)
+    {
+      return InvokerHelper<std::is_base_of<T,typename std::decay<T1>::type>::value>::get(t1).*f;
+    }
+  };
+
+  template<class F, class... Args>
+  struct InvokeResult
+  {
+    typedef Invoker<std::is_member_function_pointer<typename std::remove_reference<F>::type>::value,
+                    std::is_member_object_pointer<typename std::remove_reference<F>::type>::value &&
+                    (sizeof...(Args) == 1)> invoker;
+    inline static auto invoke (F&& f, Args&&... args) -> decltype(invoker::invoke(std::forward<F>(f), std::forward<Args>(args)...))
+    {
+      return invoker::invoke(std::forward<F>(f), std::forward<Args>(args)...);
+    }
+  };
+
+  template<class F, class...Args>
+  auto invoke (F&& f, Args&&... args) -> decltype(InvokeResult<F, Args...>::invoke(std::forward<F>(f), std::forward<Args>(args)...))
+  {
+    return InvokeResult<F, Args...>::invoke(std::forward<F>(f), std::forward<Args>(args)...);
+  }
+#else
+    using std::invoke;
+#endif
+} //  Namespace "detail"
+} //  Namespace "mingw_stdthread"
+
+#endif

thirdparty/mingw-std-threads/mingw.mutex.h

@@ -0,0 +1,500 @@
+/**
+* @file mingw.mutex.h
+* @brief std::mutex et al implementation for MinGW
+** (c) 2013-2016 by Mega Limited, Auckland, New Zealand
+* @author Alexander Vassilev
+*
+* @copyright Simplified (2-clause) BSD License.
+* You should have received a copy of the license along with this
+* program.
+*
+* This code is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+* @note
+* This file may become part of the mingw-w64 runtime package. If/when this happens,
+* the appropriate license will be added, i.e. this code will become dual-licensed,
+* and the current BSD 2-clause license will stay.
+*/
+
+#ifndef WIN32STDMUTEX_H
+#define WIN32STDMUTEX_H
+
+#if !defined(__cplusplus) || (__cplusplus < 201103L)
+#error A C++11 compiler is required!
+#endif
+// Recursion checks on non-recursive locks have some performance penalty, and
+// the C++ standard does not mandate them. The user might want to explicitly
+// enable or disable such checks. If the user has no preference, enable such
+// checks in debug builds, but not in release builds.
+#ifdef STDMUTEX_RECURSION_CHECKS
+#elif defined(NDEBUG)
+#define STDMUTEX_RECURSION_CHECKS 0
+#else
+#define STDMUTEX_RECURSION_CHECKS 1
+#endif
+
+#include <chrono>
+#include <system_error>
+#include <atomic>
+#include <mutex> //need for call_once()
+
+#if STDMUTEX_RECURSION_CHECKS || !defined(NDEBUG)
+#include <cstdio>
+#endif
+
+#include <sdkddkver.h>  //  Detect Windows version.
+
+#if (defined(__MINGW32__) && !defined(__MINGW64_VERSION_MAJOR))
+#pragma message "The Windows API that MinGW-w32 provides is not fully compatible\
+ with Microsoft's API. We'll try to work around this, but we can make no\
+ guarantees. This problem does not exist in MinGW-w64."
+#include <windows.h>    //  No further granularity can be expected.
+#else
+#if STDMUTEX_RECURSION_CHECKS
+#include <processthreadsapi.h>  //  For GetCurrentThreadId
+#endif
+#include <synchapi.h> //  For InitializeCriticalSection, etc.
+#include <errhandlingapi.h> //  For GetLastError
+#include <handleapi.h>
+#endif
+
+//  Need for the implementation of invoke
+#include "mingw.invoke.h"
+
+#if !defined(_WIN32_WINNT) || (_WIN32_WINNT < 0x0501)
+#error To use the MinGW-std-threads library, you will need to define the macro _WIN32_WINNT to be 0x0501 (Windows XP) or higher.
+#endif
+
+namespace mingw_stdthread
+{
+//    The _NonRecursive class has mechanisms that do not play nice with direct
+//  manipulation of the native handle. This forward declaration is part of
+//  a friend class declaration.
+#if STDMUTEX_RECURSION_CHECKS
+namespace vista
+{
+class condition_variable;
+}
+#endif
+//    To make this namespace equivalent to the thread-related subset of std,
+//  pull in the classes and class templates supplied by std but not by this
+//  implementation.
+using std::lock_guard;
+using std::unique_lock;
+using std::adopt_lock_t;
+using std::defer_lock_t;
+using std::try_to_lock_t;
+using std::adopt_lock;
+using std::defer_lock;
+using std::try_to_lock;
+
+class recursive_mutex
+{
+    CRITICAL_SECTION mHandle;
+public:
+    typedef LPCRITICAL_SECTION native_handle_type;
+    native_handle_type native_handle() {return &mHandle;}
+    recursive_mutex() noexcept : mHandle()
+    {
+        InitializeCriticalSection(&mHandle);
+    }
+    recursive_mutex (const recursive_mutex&) = delete;
+    recursive_mutex& operator=(const recursive_mutex&) = delete;
+    ~recursive_mutex() noexcept
+    {
+        DeleteCriticalSection(&mHandle);
+    }
+    void lock()
+    {
+        EnterCriticalSection(&mHandle);
+    }
+    void unlock()
+    {
+        LeaveCriticalSection(&mHandle);
+    }
+    bool try_lock()
+    {
+        return (TryEnterCriticalSection(&mHandle)!=0);
+    }
+};
+
+#if STDMUTEX_RECURSION_CHECKS
+struct _OwnerThread
+{
+//    If this is to be read before locking, then the owner-thread variable must
+//  be atomic to prevent a torn read from spuriously causing errors.
+    std::atomic<DWORD> mOwnerThread;
+    constexpr _OwnerThread () noexcept : mOwnerThread(0) {}
+    static void on_deadlock (void)
+    {
+        using namespace std;
+        fprintf(stderr, "FATAL: Recursive locking of non-recursive mutex\
+ detected. Throwing system exception\n");
+        fflush(stderr);
+        __builtin_trap();
+    }
+    DWORD checkOwnerBeforeLock() const
+    {
+        DWORD self = GetCurrentThreadId();
+        if (mOwnerThread.load(std::memory_order_relaxed) == self)
+            on_deadlock();
+        return self;
+    }
+    void setOwnerAfterLock(DWORD id)
+    {
+        mOwnerThread.store(id, std::memory_order_relaxed);
+    }
+    void checkSetOwnerBeforeUnlock()
+    {
+        DWORD self = GetCurrentThreadId();
+        if (mOwnerThread.load(std::memory_order_relaxed) != self)
+            on_deadlock();
+        mOwnerThread.store(0, std::memory_order_relaxed);
+    }
+};
+#endif
+
+//    Though the Slim Reader-Writer (SRW) locks used here are not complete until
+//  Windows 7, implementing partial functionality in Vista will simplify the
+//  interaction with condition variables.
+
+//Define SRWLOCK_INIT.
+ 
+#if !defined(SRWLOCK_INIT)
+#pragma message "SRWLOCK_INIT macro is not defined. Defining automatically."
+#define SRWLOCK_INIT {0}
+#endif
+ 
+#if defined(_WIN32) && (WINVER >= _WIN32_WINNT_VISTA)
+namespace windows7
+{
+class mutex
+{
+    SRWLOCK mHandle;
+//  Track locking thread for error checking.
+#if STDMUTEX_RECURSION_CHECKS
+    friend class vista::condition_variable;
+    _OwnerThread mOwnerThread {};
+#endif
+public:
+    typedef PSRWLOCK native_handle_type;
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wzero-as-null-pointer-constant"
+    constexpr mutex () noexcept : mHandle(SRWLOCK_INIT) { }
+#pragma GCC diagnostic pop
+    mutex (const mutex&) = delete;
+    mutex & operator= (const mutex&) = delete;
+    void lock (void)
+    {
+//  Note: Undefined behavior if called recursively.
+#if STDMUTEX_RECURSION_CHECKS
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+#endif
+        AcquireSRWLockExclusive(&mHandle);
+#if STDMUTEX_RECURSION_CHECKS
+        mOwnerThread.setOwnerAfterLock(self);
+#endif
+    }
+    void unlock (void)
+    {
+#if STDMUTEX_RECURSION_CHECKS
+        mOwnerThread.checkSetOwnerBeforeUnlock();
+#endif
+        ReleaseSRWLockExclusive(&mHandle);
+    }
+//  TryAcquireSRW functions are a Windows 7 feature.
+#if (WINVER >= _WIN32_WINNT_WIN7)
+    bool try_lock (void)
+    {
+#if STDMUTEX_RECURSION_CHECKS
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+#endif
+        BOOL ret = TryAcquireSRWLockExclusive(&mHandle);
+#if STDMUTEX_RECURSION_CHECKS
+        if (ret)
+            mOwnerThread.setOwnerAfterLock(self);
+#endif
+        return ret;
+    }
+#endif
+    native_handle_type native_handle (void)
+    {
+        return &mHandle;
+    }
+};
+} //  Namespace windows7
+#endif  //  Compiling for Vista
+namespace xp
+{
+class mutex
+{
+    CRITICAL_SECTION mHandle;
+    std::atomic_uchar mState;
+//  Track locking thread for error checking.
+#if STDMUTEX_RECURSION_CHECKS
+    friend class vista::condition_variable;
+    _OwnerThread mOwnerThread {};
+#endif
+public:
+    typedef PCRITICAL_SECTION native_handle_type;
+    constexpr mutex () noexcept : mHandle(), mState(2) { }
+    mutex (const mutex&) = delete;
+    mutex & operator= (const mutex&) = delete;
+    ~mutex() noexcept
+    {
+//    Undefined behavior if the mutex is held (locked) by any thread.
+//    Undefined behavior if a thread terminates while holding ownership of the
+//  mutex.
+        DeleteCriticalSection(&mHandle);
+    }
+    void lock (void)
+    {
+        unsigned char state = mState.load(std::memory_order_acquire);
+        while (state) {
+            if ((state == 2) && mState.compare_exchange_weak(state, 1, std::memory_order_acquire))
+            {
+                InitializeCriticalSection(&mHandle);
+                mState.store(0, std::memory_order_release);
+                break;
+            }
+            if (state == 1)
+            {
+                Sleep(0);
+                state = mState.load(std::memory_order_acquire);
+            }
+        }
+#if STDMUTEX_RECURSION_CHECKS
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+#endif
+        EnterCriticalSection(&mHandle);
+#if STDMUTEX_RECURSION_CHECKS
+        mOwnerThread.setOwnerAfterLock(self);
+#endif
+    }
+    void unlock (void)
+    {
+#if STDMUTEX_RECURSION_CHECKS
+        mOwnerThread.checkSetOwnerBeforeUnlock();
+#endif
+        LeaveCriticalSection(&mHandle);
+    }
+    bool try_lock (void)
+    {
+        unsigned char state = mState.load(std::memory_order_acquire);
+        if ((state == 2) && mState.compare_exchange_strong(state, 1, std::memory_order_acquire))
+        {
+            InitializeCriticalSection(&mHandle);
+            mState.store(0, std::memory_order_release);
+        }
+        if (state == 1)
+            return false;
+#if STDMUTEX_RECURSION_CHECKS
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+#endif
+        BOOL ret = TryEnterCriticalSection(&mHandle);
+#if STDMUTEX_RECURSION_CHECKS
+        if (ret)
+            mOwnerThread.setOwnerAfterLock(self);
+#endif
+        return ret;
+    }
+    native_handle_type native_handle (void)
+    {
+        return &mHandle;
+    }
+};
+} //  Namespace "xp"
+#if (WINVER >= _WIN32_WINNT_WIN7)
+using windows7::mutex;
+#else
+using xp::mutex;
+#endif
+
+class recursive_timed_mutex
+{
+    static constexpr DWORD kWaitAbandoned = 0x00000080l;
+    static constexpr DWORD kWaitObject0 = 0x00000000l;
+    static constexpr DWORD kInfinite = 0xffffffffl;
+    inline bool try_lock_internal (DWORD ms) noexcept
+    {
+        DWORD ret = WaitForSingleObject(mHandle, ms);
+#ifndef NDEBUG
+        if (ret == kWaitAbandoned)
+        {
+            using namespace std;
+            fprintf(stderr, "FATAL: Thread terminated while holding a mutex.");
+            terminate();
+        }
+#endif
+        return (ret == kWaitObject0) || (ret == kWaitAbandoned);
+    }
+protected:
+    HANDLE mHandle;
+//    Track locking thread for error checking of non-recursive timed_mutex. For
+//  standard compliance, this must be defined in same class and at the same
+//  access-control level as every other variable in the timed_mutex.
+#if STDMUTEX_RECURSION_CHECKS
+    friend class vista::condition_variable;
+    _OwnerThread mOwnerThread {};
+#endif
+public:
+    typedef HANDLE native_handle_type;
+    native_handle_type native_handle() const {return mHandle;}
+    recursive_timed_mutex(const recursive_timed_mutex&) = delete;
+    recursive_timed_mutex& operator=(const recursive_timed_mutex&) = delete;
+    recursive_timed_mutex(): mHandle(CreateMutex(NULL, FALSE, NULL)) {}
+    ~recursive_timed_mutex()
+    {
+        CloseHandle(mHandle);
+    }
+    void lock()
+    {
+        DWORD ret = WaitForSingleObject(mHandle, kInfinite);
+//    If (ret == WAIT_ABANDONED), then the thread that held ownership was
+//  terminated. Behavior is undefined, but Windows will pass ownership to this
+//  thread.
+#ifndef NDEBUG
+        if (ret == kWaitAbandoned)
+        {
+            using namespace std;
+            fprintf(stderr, "FATAL: Thread terminated while holding a mutex.");
+            terminate();
+        }
+#endif
+        if ((ret != kWaitObject0) && (ret != kWaitAbandoned))
+        {
+            __builtin_trap();
+        }
+    }
+    void unlock()
+    {
+        if (!ReleaseMutex(mHandle))
+            __builtin_trap();
+    }
+    bool try_lock()
+    {
+        return try_lock_internal(0);
+    }
+    template <class Rep, class Period>
+    bool try_lock_for(const std::chrono::duration<Rep,Period>& dur)
+    {
+        using namespace std::chrono;
+        auto timeout = duration_cast<milliseconds>(dur).count();
+        while (timeout > 0)
+        {
+          constexpr auto kMaxStep = static_cast<decltype(timeout)>(kInfinite-1);
+          auto step = (timeout < kMaxStep) ? timeout : kMaxStep;
+          if (try_lock_internal(static_cast<DWORD>(step)))
+            return true;
+          timeout -= step;
+        }
+        return false;
+    }
+    template <class Clock, class Duration>
+    bool try_lock_until(const std::chrono::time_point<Clock,Duration>& timeout_time)
+    {
+        return try_lock_for(timeout_time - Clock::now());
+    }
+};
+
+//  Override if, and only if, it is necessary for error-checking.
+#if STDMUTEX_RECURSION_CHECKS
+class timed_mutex: recursive_timed_mutex
+{
+public:
+    timed_mutex() = default;
+    timed_mutex(const timed_mutex&) = delete;
+    timed_mutex& operator=(const timed_mutex&) = delete;
+    void lock()
+    {
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+        recursive_timed_mutex::lock();
+        mOwnerThread.setOwnerAfterLock(self);
+    }
+    void unlock()
+    {
+        mOwnerThread.checkSetOwnerBeforeUnlock();
+        recursive_timed_mutex::unlock();
+    }
+    template <class Rep, class Period>
+    bool try_lock_for(const std::chrono::duration<Rep,Period>& dur)
+    {
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+        bool ret = recursive_timed_mutex::try_lock_for(dur);
+        if (ret)
+            mOwnerThread.setOwnerAfterLock(self);
+        return ret;
+    }
+    template <class Clock, class Duration>
+    bool try_lock_until(const std::chrono::time_point<Clock,Duration>& timeout_time)
+    {
+        return try_lock_for(timeout_time - Clock::now());
+    }
+    bool try_lock ()
+    {
+        return try_lock_for(std::chrono::milliseconds(0));
+    }
+};
+#else
+typedef recursive_timed_mutex timed_mutex;
+#endif
+
+class once_flag
+{
+//    When available, the SRW-based mutexes should be faster than the
+//  CriticalSection-based mutexes. Only try_lock will be unavailable in Vista,
+//  and try_lock is not used by once_flag.
+#if (_WIN32_WINNT == _WIN32_WINNT_VISTA)
+    windows7::mutex mMutex;
+#else
+    mutex mMutex;
+#endif
+    std::atomic_bool mHasRun;
+    once_flag(const once_flag&) = delete;
+    once_flag& operator=(const once_flag&) = delete;
+    template<class Callable, class... Args>
+    friend void call_once(once_flag& once, Callable&& f, Args&&... args);
+public:
+    constexpr once_flag() noexcept: mMutex(), mHasRun(false) {}
+};
+
+template<class Callable, class... Args>
+void call_once(once_flag& flag, Callable&& func, Args&&... args)
+{
+    if (flag.mHasRun.load(std::memory_order_acquire))
+        return;
+    lock_guard<decltype(flag.mMutex)> lock(flag.mMutex);
+    if (flag.mHasRun.load(std::memory_order_relaxed))
+        return;
+    detail::invoke(std::forward<Callable>(func),std::forward<Args>(args)...);
+    flag.mHasRun.store(true, std::memory_order_release);
+}
+} //  Namespace mingw_stdthread
+
+//  Push objects into std, but only if they are not already there.
+namespace std
+{
+//    Because of quirks of the compiler, the common "using namespace std;"
+//  directive would flatten the namespaces and introduce ambiguity where there
+//  was none. Direct specification (std::), however, would be unaffected.
+//    Take the safe option, and include only in the presence of MinGW's win32
+//  implementation.
+#if defined(__MINGW32__ ) && !defined(_GLIBCXX_HAS_GTHREADS)
+using mingw_stdthread::recursive_mutex;
+using mingw_stdthread::mutex;
+using mingw_stdthread::recursive_timed_mutex;
+using mingw_stdthread::timed_mutex;
+using mingw_stdthread::once_flag;
+using mingw_stdthread::call_once;
+#elif !defined(MINGW_STDTHREAD_REDUNDANCY_WARNING)  //  Skip repetition
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#pragma message "This version of MinGW seems to include a win32 port of\
+ pthreads, and probably already has C++11 std threading classes implemented,\
+ based on pthreads. These classes, found in namespace std, are not overridden\
+ by the mingw-std-thread library. If you would still like to use this\
+ implementation (as it is more lightweight), use the classes provided in\
+ namespace mingw_stdthread."
+#endif
+}
+#endif // WIN32STDMUTEX_H

thirdparty/mingw-std-threads/mingw.shared_mutex.h

@@ -0,0 +1,503 @@
+/// \file mingw.shared_mutex.h
+/// \brief Standard-compliant shared_mutex for MinGW
+///
+/// (c) 2017 by Nathaniel J. McClatchey, Athens OH, United States
+/// \author Nathaniel J. McClatchey
+///
+/// \copyright Simplified (2-clause) BSD License.
+///
+/// \note This file may become part of the mingw-w64 runtime package. If/when
+/// this happens, the appropriate license will be added, i.e. this code will
+/// become dual-licensed, and the current BSD 2-clause license will stay.
+/// \note Target Windows version is determined by WINVER, which is determined in
+/// <windows.h> from _WIN32_WINNT, which can itself be set by the user.
+
+//  Notes on the namespaces:
+//  - The implementation can be accessed directly in the namespace
+//    mingw_stdthread.
+//  - Objects will be brought into namespace std by a using directive. This
+//    will cause objects declared in std (such as MinGW's implementation) to
+//    hide this implementation's definitions.
+//  - To avoid poluting the namespace with implementation details, all objects
+//    to be pushed into std will be placed in mingw_stdthread::visible.
+//  The end result is that if MinGW supplies an object, it is automatically
+//  used. If MinGW does not supply an object, this implementation's version will
+//  instead be used.
+
+#ifndef MINGW_SHARED_MUTEX_H_
+#define MINGW_SHARED_MUTEX_H_
+
+#if !defined(__cplusplus) || (__cplusplus < 201103L)
+#error A C++11 compiler is required!
+#endif
+
+#include <cassert>
+//  For descriptive errors.
+#include <system_error>
+//    Implementing a shared_mutex without OS support will require atomic read-
+//  modify-write capacity.
+#include <atomic>
+//  For timing in shared_lock and shared_timed_mutex.
+#include <chrono>
+#include <limits>
+
+//    Use MinGW's shared_lock class template, if it's available. Requires C++14.
+//  If unavailable (eg. because this library is being used in C++11), then an
+//  implementation of shared_lock is provided by this header.
+#if (__cplusplus >= 201402L)
+#include <shared_mutex>
+#endif
+
+//  For defer_lock_t, adopt_lock_t, and try_to_lock_t
+#include "mingw.mutex.h"
+//  For this_thread::yield.
+//#include "mingw.thread.h"
+
+//  Might be able to use native Slim Reader-Writer (SRW) locks.
+#ifdef _WIN32
+#include <sdkddkver.h>  //  Detect Windows version.
+#if (defined(__MINGW32__) && !defined(__MINGW64_VERSION_MAJOR))
+#pragma message "The Windows API that MinGW-w32 provides is not fully compatible\
+ with Microsoft's API. We'll try to work around this, but we can make no\
+ guarantees. This problem does not exist in MinGW-w64."
+#include <windows.h>    //  No further granularity can be expected.
+#else
+#include <synchapi.h>
+#endif
+#endif
+
+namespace mingw_stdthread
+{
+//  Define a portable atomics-based shared_mutex
+namespace portable
+{
+class shared_mutex
+{
+    typedef uint_fast16_t counter_type;
+    std::atomic<counter_type> mCounter {0};
+    static constexpr counter_type kWriteBit = 1 << (std::numeric_limits<counter_type>::digits - 1);
+
+#if STDMUTEX_RECURSION_CHECKS
+//  Runtime checker for verifying owner threads. Note: Exclusive mode only.
+    _OwnerThread mOwnerThread {};
+#endif
+public:
+    typedef shared_mutex * native_handle_type;
+
+    shared_mutex () = default;
+
+//  No form of copying or moving should be allowed.
+    shared_mutex (const shared_mutex&) = delete;
+    shared_mutex & operator= (const shared_mutex&) = delete;
+
+    ~shared_mutex ()
+    {
+//  Terminate if someone tries to destroy an owned mutex.
+        assert(mCounter.load(std::memory_order_relaxed) == 0);
+    }
+
+    void lock_shared (void)
+    {
+        counter_type expected = mCounter.load(std::memory_order_relaxed);
+        do
+        {
+//  Delay if writing or if too many readers are attempting to read.
+            if (expected >= kWriteBit - 1)
+            {
+                using namespace std;
+                expected = mCounter.load(std::memory_order_relaxed);
+                continue;
+            }
+            if (mCounter.compare_exchange_weak(expected,
+                                               static_cast<counter_type>(expected + 1),
+                                               std::memory_order_acquire,
+                                               std::memory_order_relaxed))
+                break;
+        }
+        while (true);
+    }
+
+    bool try_lock_shared (void)
+    {
+        counter_type expected = mCounter.load(std::memory_order_relaxed) & static_cast<counter_type>(~kWriteBit);
+        if (expected + 1 == kWriteBit)
+            return false;
+        else
+            return mCounter.compare_exchange_strong( expected,
+                                                    static_cast<counter_type>(expected + 1),
+                                                    std::memory_order_acquire,
+                                                    std::memory_order_relaxed);
+    }
+
+    void unlock_shared (void)
+    {
+        using namespace std;
+#ifndef NDEBUG
+        if (!(mCounter.fetch_sub(1, memory_order_release) & static_cast<counter_type>(~kWriteBit)))
+            __builtin_trap();
+#else
+        mCounter.fetch_sub(1, memory_order_release);
+#endif
+    }
+
+//  Behavior is undefined if a lock was previously acquired.
+    void lock (void)
+    {
+#if STDMUTEX_RECURSION_CHECKS
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+#endif
+        using namespace std;
+//  Might be able to use relaxed memory order...
+//  Wait for the write-lock to be unlocked, then claim the write slot.
+        counter_type current;
+        while ((current = mCounter.fetch_or(kWriteBit, std::memory_order_acquire)) & kWriteBit);
+            //this_thread::yield();
+//  Wait for readers to finish up.
+        while (current != kWriteBit)
+        {
+            //this_thread::yield();
+            current = mCounter.load(std::memory_order_acquire);
+        }
+#if STDMUTEX_RECURSION_CHECKS
+        mOwnerThread.setOwnerAfterLock(self);
+#endif
+    }
+
+    bool try_lock (void)
+    {
+#if STDMUTEX_RECURSION_CHECKS
+        DWORD self = mOwnerThread.checkOwnerBeforeLock();
+#endif
+        counter_type expected = 0;
+        bool ret = mCounter.compare_exchange_strong(expected, kWriteBit,
+                                                    std::memory_order_acquire,
+                                                    std::memory_order_relaxed);
+#if STDMUTEX_RECURSION_CHECKS
+        if (ret)
+            mOwnerThread.setOwnerAfterLock(self);
+#endif
+        return ret;
+    }
+
+    void unlock (void)
+    {
+#if STDMUTEX_RECURSION_CHECKS
+        mOwnerThread.checkSetOwnerBeforeUnlock();
+#endif
+        using namespace std;
+#ifndef NDEBUG
+        if (mCounter.load(memory_order_relaxed) != kWriteBit)
+            __builtin_trap();
+#endif
+        mCounter.store(0, memory_order_release);
+    }
+
+    native_handle_type native_handle (void)
+    {
+        return this;
+    }
+};
+
+} //  Namespace portable
+
+//    The native shared_mutex implementation primarily uses features of Windows
+//  Vista, but the features used for try_lock and try_lock_shared were not
+//  introduced until Windows 7. To allow limited use while compiling for Vista,
+//  I define the class without try_* functions in that case.
+//    Only fully-featured implementations will be placed into namespace std.
+#if defined(_WIN32) && (WINVER >= _WIN32_WINNT_VISTA)
+namespace vista
+{
+class condition_variable_any;
+}
+
+namespace windows7
+{
+//  We already #include "mingw.mutex.h". May as well reduce redundancy.
+class shared_mutex : windows7::mutex
+{
+//    Allow condition_variable_any (and only condition_variable_any) to treat a
+//  shared_mutex as its base class.
+    friend class vista::condition_variable_any;
+public:
+    using windows7::mutex::native_handle_type;
+    using windows7::mutex::lock;
+    using windows7::mutex::unlock;
+    using windows7::mutex::native_handle;
+
+    void lock_shared (void)
+    {
+        AcquireSRWLockShared(native_handle());
+    }
+
+    void unlock_shared (void)
+    {
+        ReleaseSRWLockShared(native_handle());
+    }
+
+//  TryAcquireSRW functions are a Windows 7 feature.
+#if (WINVER >= _WIN32_WINNT_WIN7)
+    bool try_lock_shared (void)
+    {
+        return TryAcquireSRWLockShared(native_handle()) != 0;
+    }
+
+    using windows7::mutex::try_lock;
+#endif
+};
+
+} //  Namespace windows7
+#endif  //  Compiling for Vista
+#if (defined(_WIN32) && (WINVER >= _WIN32_WINNT_WIN7))
+using windows7::shared_mutex;
+#else
+using portable::shared_mutex;
+#endif
+
+class shared_timed_mutex : shared_mutex
+{
+    typedef shared_mutex Base;
+public:
+    using Base::lock;
+    using Base::try_lock;
+    using Base::unlock;
+    using Base::lock_shared;
+    using Base::try_lock_shared;
+    using Base::unlock_shared;
+
+    template< class Clock, class Duration >
+    bool try_lock_until ( const std::chrono::time_point<Clock,Duration>& cutoff )
+    {
+        do
+        {
+            if (try_lock())
+                return true;
+        }
+        while (std::chrono::steady_clock::now() < cutoff);
+        return false;
+    }
+
+    template< class Rep, class Period >
+    bool try_lock_for (const std::chrono::duration<Rep,Period>& rel_time)
+    {
+        return try_lock_until(std::chrono::steady_clock::now() + rel_time);
+    }
+
+    template< class Clock, class Duration >
+    bool try_lock_shared_until ( const std::chrono::time_point<Clock,Duration>& cutoff )
+    {
+        do
+        {
+            if (try_lock_shared())
+                return true;
+        }
+        while (std::chrono::steady_clock::now() < cutoff);
+        return false;
+    }
+
+    template< class Rep, class Period >
+    bool try_lock_shared_for (const std::chrono::duration<Rep,Period>& rel_time)
+    {
+        return try_lock_shared_until(std::chrono::steady_clock::now() + rel_time);
+    }
+};
+
+#if __cplusplus >= 201402L
+using std::shared_lock;
+#else
+//    If not supplied by shared_mutex (eg. because C++14 is not supported), I
+//  supply the various helper classes that the header should have defined.
+template<class Mutex>
+class shared_lock
+{
+    Mutex * mMutex;
+    bool mOwns;
+//  Reduce code redundancy
+    void verify_lockable (void)
+    {
+        using namespace std;
+        if (mMutex == nullptr)
+            __builtin_trap();
+        if (mOwns)
+            __builtin_trap();
+    }
+public:
+    typedef Mutex mutex_type;
+
+    shared_lock (void) noexcept
+        : mMutex(nullptr), mOwns(false)
+    {
+    }
+
+    shared_lock (shared_lock<Mutex> && other) noexcept
+        : mMutex(other.mutex_), mOwns(other.owns_)
+    {
+        other.mMutex = nullptr;
+        other.mOwns = false;
+    }
+
+    explicit shared_lock (mutex_type & m)
+        : mMutex(&m), mOwns(true)
+    {
+        mMutex->lock_shared();
+    }
+
+    shared_lock (mutex_type & m, defer_lock_t) noexcept
+        : mMutex(&m), mOwns(false)
+    {
+    }
+
+    shared_lock (mutex_type & m, adopt_lock_t)
+        : mMutex(&m), mOwns(true)
+    {
+    }
+
+    shared_lock (mutex_type & m, try_to_lock_t)
+        : mMutex(&m), mOwns(m.try_lock_shared())
+    {
+    }
+
+    template< class Rep, class Period >
+    shared_lock( mutex_type& m, const std::chrono::duration<Rep,Period>& timeout_duration )
+        : mMutex(&m), mOwns(m.try_lock_shared_for(timeout_duration))
+    {
+    }
+
+    template< class Clock, class Duration >
+    shared_lock( mutex_type& m, const std::chrono::time_point<Clock,Duration>& timeout_time )
+        : mMutex(&m), mOwns(m.try_lock_shared_until(timeout_time))
+    {
+    }
+
+    shared_lock& operator= (shared_lock<Mutex> && other) noexcept
+    {
+        if (&other != this)
+        {
+            if (mOwns)
+                mMutex->unlock_shared();
+            mMutex = other.mMutex;
+            mOwns = other.mOwns;
+            other.mMutex = nullptr;
+            other.mOwns = false;
+        }
+        return *this;
+    }
+
+
+    ~shared_lock (void)
+    {
+        if (mOwns)
+            mMutex->unlock_shared();
+    }
+
+    shared_lock (const shared_lock<Mutex> &) = delete;
+    shared_lock& operator= (const shared_lock<Mutex> &) = delete;
+
+//  Shared locking
+    void lock (void)
+    {
+        verify_lockable();
+        mMutex->lock_shared();
+        mOwns = true;
+    }
+
+    bool try_lock (void)
+    {
+        verify_lockable();
+        mOwns = mMutex->try_lock_shared();
+        return mOwns;
+    }
+
+    template< class Clock, class Duration >
+    bool try_lock_until( const std::chrono::time_point<Clock,Duration>& cutoff )
+    {
+        verify_lockable();
+        do
+        {
+            mOwns = mMutex->try_lock_shared();
+            if (mOwns)
+                return mOwns;
+        }
+        while (std::chrono::steady_clock::now() < cutoff);
+        return false;
+    }
+
+    template< class Rep, class Period >
+    bool try_lock_for (const std::chrono::duration<Rep,Period>& rel_time)
+    {
+        return try_lock_until(std::chrono::steady_clock::now() + rel_time);
+    }
+
+    void unlock (void)
+    {
+        using namespace std;
+        if (!mOwns)
+            __builtin_trap();
+        mMutex->unlock_shared();
+        mOwns = false;
+    }
+
+//  Modifiers
+    void swap (shared_lock<Mutex> & other) noexcept
+    {
+        using namespace std;
+        swap(mMutex, other.mMutex);
+        swap(mOwns, other.mOwns);
+    }
+
+    mutex_type * release (void) noexcept
+    {
+        mutex_type * ptr = mMutex;
+        mMutex = nullptr;
+        mOwns = false;
+        return ptr;
+    }
+//  Observers
+    mutex_type * mutex (void) const noexcept
+    {
+        return mMutex;
+    }
+
+    bool owns_lock (void) const noexcept
+    {
+        return mOwns;
+    }
+
+    explicit operator bool () const noexcept
+    {
+        return owns_lock();
+    }
+};
+
+template< class Mutex >
+void swap( shared_lock<Mutex>& lhs, shared_lock<Mutex>& rhs ) noexcept
+{
+    lhs.swap(rhs);
+}
+#endif  //  C++11
+} //  Namespace mingw_stdthread
+
+namespace std
+{
+//    Because of quirks of the compiler, the common "using namespace std;"
+//  directive would flatten the namespaces and introduce ambiguity where there
+//  was none. Direct specification (std::), however, would be unaffected.
+//    Take the safe option, and include only in the presence of MinGW's win32
+//  implementation.
+#if (__cplusplus < 201703L) || (defined(__MINGW32__ ) && !defined(_GLIBCXX_HAS_GTHREADS))
+using mingw_stdthread::shared_mutex;
+#endif
+#if (__cplusplus < 201402L) || (defined(__MINGW32__ ) && !defined(_GLIBCXX_HAS_GTHREADS))
+using mingw_stdthread::shared_timed_mutex;
+using mingw_stdthread::shared_lock;
+#elif !defined(MINGW_STDTHREAD_REDUNDANCY_WARNING)  //  Skip repetition
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#pragma message "This version of MinGW seems to include a win32 port of\
+ pthreads, and probably already has C++ std threading classes implemented,\
+ based on pthreads. These classes, found in namespace std, are not overridden\
+ by the mingw-std-thread library. If you would still like to use this\
+ implementation (as it is more lightweight), use the classes provided in\
+ namespace mingw_stdthread."
+#endif
+} //  Namespace std
+#endif // MINGW_SHARED_MUTEX_H_

thirdparty/mingw-std-threads/mingw.thread.h

@@ -0,0 +1,360 @@
+/**
+* @file mingw.thread.h
+* @brief std::thread implementation for MinGW
+* (c) 2013-2016 by Mega Limited, Auckland, New Zealand
+* @author Alexander Vassilev
+*
+* @copyright Simplified (2-clause) BSD License.
+* You should have received a copy of the license along with this
+* program.
+*
+* This code is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+* @note
+* This file may become part of the mingw-w64 runtime package. If/when this happens,
+* the appropriate license will be added, i.e. this code will become dual-licensed,
+* and the current BSD 2-clause license will stay.
+*/
+
+#ifndef WIN32STDTHREAD_H
+#define WIN32STDTHREAD_H
+
+#if !defined(__cplusplus) || (__cplusplus < 201103L)
+#error A C++11 compiler is required!
+#endif
+
+//  Use the standard classes for std::, if available.
+#include <thread>
+
+#include <cstddef>      //  For std::size_t
+#include <cerrno>       //  Detect error type.
+#include <exception>    //  For std::terminate
+#include <system_error> //  For std::system_error
+#include <functional>   //  For std::hash
+#include <tuple>        //  For std::tuple
+#include <chrono>       //  For sleep timing.
+#include <memory>       //  For std::unique_ptr
+#include <iosfwd>       //  Stream output for thread ids.
+#include <utility>      //  For std::swap, std::forward
+
+#include "mingw.invoke.h"
+
+#if (defined(__MINGW32__) && !defined(__MINGW64_VERSION_MAJOR))
+#pragma message "The Windows API that MinGW-w32 provides is not fully compatible\
+ with Microsoft's API. We'll try to work around this, but we can make no\
+ guarantees. This problem does not exist in MinGW-w64."
+#include <windows.h>    //  No further granularity can be expected.
+#else
+#include <synchapi.h>   //  For WaitForSingleObject
+#include <handleapi.h>  //  For CloseHandle, etc.
+#include <sysinfoapi.h> //  For GetNativeSystemInfo
+#include <processthreadsapi.h>  //  For GetCurrentThreadId
+#endif
+#include <process.h>  //  For _beginthreadex
+
+#ifndef NDEBUG
+#include <cstdio>
+#endif
+
+#if !defined(_WIN32_WINNT) || (_WIN32_WINNT < 0x0501)
+#error To use the MinGW-std-threads library, you will need to define the macro _WIN32_WINNT to be 0x0501 (Windows XP) or higher.
+#endif
+
+//  Instead of INVALID_HANDLE_VALUE, _beginthreadex returns 0.
+namespace mingw_stdthread
+{
+namespace detail
+{
+    template<std::size_t...>
+    struct IntSeq {};
+
+    template<std::size_t N, std::size_t... S>
+    struct GenIntSeq : GenIntSeq<N-1, N-1, S...> { };
+
+    template<std::size_t... S>
+    struct GenIntSeq<0, S...> { typedef IntSeq<S...> type; };
+
+//    Use a template specialization to avoid relying on compiler optimization
+//  when determining the parameter integer sequence.
+    template<class Func, class T, typename... Args>
+    class ThreadFuncCall;
+// We can't define the Call struct in the function - the standard forbids template methods in that case
+    template<class Func, std::size_t... S, typename... Args>
+    class ThreadFuncCall<Func, detail::IntSeq<S...>, Args...>
+    {
+        static_assert(sizeof...(S) == sizeof...(Args), "Args must match.");
+        using Tuple = std::tuple<typename std::decay<Args>::type...>;
+        typename std::decay<Func>::type mFunc;
+        Tuple mArgs;
+
+    public:
+        ThreadFuncCall(Func&& aFunc, Args&&... aArgs)
+          : mFunc(std::forward<Func>(aFunc)),
+            mArgs(std::forward<Args>(aArgs)...)
+        {
+        }
+
+        void callFunc()
+        {
+            detail::invoke(std::move(mFunc), std::move(std::get<S>(mArgs)) ...);
+        }
+    };
+
+//  Allow construction of threads without exposing implementation.
+    class ThreadIdTool;
+} //  Namespace "detail"
+
+class thread
+{
+public:
+    class id
+    {
+        DWORD mId = 0;
+        friend class thread;
+        friend class std::hash<id>;
+        friend class detail::ThreadIdTool;
+        explicit id(DWORD aId) noexcept : mId(aId){}
+    public:
+        id (void) noexcept = default;
+        friend bool operator==(id x, id y) noexcept {return x.mId == y.mId; }
+        friend bool operator!=(id x, id y) noexcept {return x.mId != y.mId; }
+        friend bool operator< (id x, id y) noexcept {return x.mId <  y.mId; }
+        friend bool operator<=(id x, id y) noexcept {return x.mId <= y.mId; }
+        friend bool operator> (id x, id y) noexcept {return x.mId >  y.mId; }
+        friend bool operator>=(id x, id y) noexcept {return x.mId >= y.mId; }
+
+        template<class _CharT, class _Traits>
+        friend std::basic_ostream<_CharT, _Traits>&
+        operator<<(std::basic_ostream<_CharT, _Traits>& __out, id __id)
+        {
+            if (__id.mId == 0)
+            {
+                return __out << "(invalid std::thread::id)";
+            }
+            else
+            {
+                return __out << __id.mId;
+            }
+        }
+    };
+private:
+    static constexpr HANDLE kInvalidHandle = nullptr;
+    static constexpr DWORD kInfinite = 0xffffffffl;
+    HANDLE mHandle;
+    id mThreadId;
+
+    template <class Call>
+    static unsigned __stdcall threadfunc(void* arg)
+    {
+        std::unique_ptr<Call> call(static_cast<Call*>(arg));
+        call->callFunc();
+        return 0;
+    }
+
+    static unsigned int _hardware_concurrency_helper() noexcept
+    {
+        SYSTEM_INFO sysinfo;
+//    This is one of the few functions used by the library which has a nearly-
+//  equivalent function defined in earlier versions of Windows. Include the
+//  workaround, just as a reminder that it does exist.
+#if defined(_WIN32_WINNT) && (_WIN32_WINNT >= 0x0501)
+        ::GetNativeSystemInfo(&sysinfo);
+#else
+        ::GetSystemInfo(&sysinfo);
+#endif
+        return sysinfo.dwNumberOfProcessors;
+    }
+public:
+    typedef HANDLE native_handle_type;
+    id get_id() const noexcept {return mThreadId;}
+    native_handle_type native_handle() const {return mHandle;}
+    thread(): mHandle(kInvalidHandle), mThreadId(){}
+
+    thread(thread&& other)
+    :mHandle(other.mHandle), mThreadId(other.mThreadId)
+    {
+        other.mHandle = kInvalidHandle;
+        other.mThreadId = id{};
+    }
+
+    thread(const thread &other)=delete;
+
+    template<class Func, typename... Args>
+    explicit thread(Func&& func, Args&&... args) : mHandle(), mThreadId()
+    {
+        using ArgSequence = typename detail::GenIntSeq<sizeof...(Args)>::type;
+        using Call = detail::ThreadFuncCall<Func, ArgSequence, Args...>;
+        auto call = new Call(
+            std::forward<Func>(func), std::forward<Args>(args)...);
+        unsigned id_receiver;
+        auto int_handle = _beginthreadex(NULL, 0, threadfunc<Call>,
+            static_cast<LPVOID>(call), 0, &id_receiver);
+        if (int_handle == 0)
+        {
+            mHandle = kInvalidHandle;
+            int errnum = errno;
+            delete call;
+//  Note: Should only throw EINVAL, EAGAIN, EACCES
+            __builtin_trap();
+        } else {
+            mThreadId.mId = id_receiver;
+            mHandle = reinterpret_cast<HANDLE>(int_handle);
+        }
+    }
+
+    bool joinable() const {return mHandle != kInvalidHandle;}
+
+//    Note: Due to lack of synchronization, this function has a race condition
+//  if called concurrently, which leads to undefined behavior. The same applies
+//  to all other member functions of this class, but this one is mentioned
+//  explicitly.
+    void join()
+    {
+        using namespace std;
+        if (get_id() == id(GetCurrentThreadId()))
+            __builtin_trap();
+        if (mHandle == kInvalidHandle)
+            __builtin_trap();
+        if (!joinable())
+            __builtin_trap();
+        WaitForSingleObject(mHandle, kInfinite);
+        CloseHandle(mHandle);
+        mHandle = kInvalidHandle;
+        mThreadId = id{};
+    }
+
+    ~thread()
+    {
+        if (joinable())
+        {
+#ifndef NDEBUG
+            std::printf("Error: Must join() or detach() a thread before \
+destroying it.\n");
+#endif
+            std::terminate();
+        }
+    }
+    thread& operator=(const thread&) = delete;
+    thread& operator=(thread&& other) noexcept
+    {
+        if (joinable())
+        {
+#ifndef NDEBUG
+            std::printf("Error: Must join() or detach() a thread before \
+moving another thread to it.\n");
+#endif
+            std::terminate();
+        }
+        swap(std::forward<thread>(other));
+        return *this;
+    }
+    void swap(thread&& other) noexcept
+    {
+        std::swap(mHandle, other.mHandle);
+        std::swap(mThreadId.mId, other.mThreadId.mId);
+    }
+
+    static unsigned int hardware_concurrency() noexcept
+    {
+        static unsigned int cached = _hardware_concurrency_helper();
+        return cached;
+    }
+
+    void detach()
+    {
+        if (!joinable())
+        {
+            using namespace std;
+            __builtin_trap();
+        }
+        if (mHandle != kInvalidHandle)
+        {
+            CloseHandle(mHandle);
+            mHandle = kInvalidHandle;
+        }
+        mThreadId = id{};
+    }
+};
+
+namespace detail
+{
+    class ThreadIdTool
+    {
+    public:
+        static thread::id make_id (DWORD base_id) noexcept
+        {
+            return thread::id(base_id);
+        }
+    };
+} //  Namespace "detail"
+
+namespace this_thread
+{
+    inline thread::id get_id() noexcept
+    {
+        return detail::ThreadIdTool::make_id(GetCurrentThreadId());
+    }
+    inline void yield() noexcept {Sleep(0);}
+    template< class Rep, class Period >
+    void sleep_for( const std::chrono::duration<Rep,Period>& sleep_duration)
+    {
+        static constexpr DWORD kInfinite = 0xffffffffl;
+        using namespace std::chrono;
+        using rep = milliseconds::rep;
+        rep ms = duration_cast<milliseconds>(sleep_duration).count();
+        while (ms > 0)
+        {
+            constexpr rep kMaxRep = static_cast<rep>(kInfinite - 1);
+            auto sleepTime = (ms < kMaxRep) ? ms : kMaxRep;
+            Sleep(static_cast<DWORD>(sleepTime));
+            ms -= sleepTime;
+        }
+    }
+    template <class Clock, class Duration>
+    void sleep_until(const std::chrono::time_point<Clock,Duration>& sleep_time)
+    {
+        sleep_for(sleep_time-Clock::now());
+    }
+}
+} //  Namespace mingw_stdthread
+
+namespace std
+{
+//    Because of quirks of the compiler, the common "using namespace std;"
+//  directive would flatten the namespaces and introduce ambiguity where there
+//  was none. Direct specification (std::), however, would be unaffected.
+//    Take the safe option, and include only in the presence of MinGW's win32
+//  implementation.
+#if defined(__MINGW32__ ) && !defined(_GLIBCXX_HAS_GTHREADS)
+using mingw_stdthread::thread;
+//    Remove ambiguity immediately, to avoid problems arising from the above.
+//using std::thread;
+namespace this_thread
+{
+using namespace mingw_stdthread::this_thread;
+}
+#elif !defined(MINGW_STDTHREAD_REDUNDANCY_WARNING)  //  Skip repetition
+#define MINGW_STDTHREAD_REDUNDANCY_WARNING
+#pragma message "This version of MinGW seems to include a win32 port of\
+ pthreads, and probably already has C++11 std threading classes implemented,\
+ based on pthreads. These classes, found in namespace std, are not overridden\
+ by the mingw-std-thread library. If you would still like to use this\
+ implementation (as it is more lightweight), use the classes provided in\
+ namespace mingw_stdthread."
+#endif
+
+//    Specialize hash for this implementation's thread::id, even if the
+//  std::thread::id already has a hash.
+template<>
+struct hash<mingw_stdthread::thread::id>
+{
+    typedef mingw_stdthread::thread::id argument_type;
+    typedef size_t result_type;
+    size_t operator() (const argument_type & i) const noexcept
+    {
+        return i.mId;
+    }
+};
+}
+#endif // WIN32STDTHREAD_H

thirdparty/mingw-std-threads/no_except.patch

@@ -0,0 +1,137 @@
+diff --git a/thirdparty/mingw-std-threads/mingw.condition_variable.h b/thirdparty/mingw-std-threads/mingw.condition_variable.h
+index 50c5ebd6df..f9e248c154 100644
+--- a/thirdparty/mingw-std-threads/mingw.condition_variable.h
++++ b/thirdparty/mingw-std-threads/mingw.condition_variable.h
+@@ -87,12 +87,12 @@ public:
+         :   mSemaphore(CreateSemaphoreA(NULL, 0, 0xFFFF, NULL))
+     {
+         if (mSemaphore == NULL)
+-            throw std::system_error(GetLastError(), std::generic_category());
++            __builtin_trap();
+         mWakeEvent = CreateEvent(NULL, FALSE, FALSE, NULL);
+         if (mWakeEvent == NULL)
+         {
+             CloseHandle(mSemaphore);
+-            throw std::system_error(GetLastError(), std::generic_category());
++            __builtin_trap();
+         }
+     }
+     ~condition_variable_any()
+@@ -132,7 +132,7 @@ private:
+         else
+         {
+             using namespace std;
+-            throw system_error(make_error_code(errc::protocol_error));
++            __builtin_trap();
+         }
+     }
+ public:
+diff --git a/thirdparty/mingw-std-threads/mingw.mutex.h b/thirdparty/mingw-std-threads/mingw.mutex.h
+index 03efa13f8b..73698d13cb 100644
+--- a/thirdparty/mingw-std-threads/mingw.mutex.h
++++ b/thirdparty/mingw-std-threads/mingw.mutex.h
+@@ -132,7 +132,7 @@ struct _OwnerThread
+         fprintf(stderr, "FATAL: Recursive locking of non-recursive mutex\
+  detected. Throwing system exception\n");
+         fflush(stderr);
+-        throw system_error(make_error_code(errc::resource_deadlock_would_occur));
++        __builtin_trap();
+     }
+     DWORD checkOwnerBeforeLock() const
+     {
+@@ -364,13 +364,13 @@ public:
+ #endif
+         if ((ret != kWaitObject0) && (ret != kWaitAbandoned))
+         {
+-            throw std::system_error(GetLastError(), std::system_category());
++            __builtin_trap();
+         }
+     }
+     void unlock()
+     {
+         if (!ReleaseMutex(mHandle))
+-            throw std::system_error(GetLastError(), std::system_category());
++            __builtin_trap();
+     }
+     bool try_lock()
+     {
+diff --git a/thirdparty/mingw-std-threads/mingw.shared_mutex.h b/thirdparty/mingw-std-threads/mingw.shared_mutex.h
+index ff1ac65135..5375b0fbd1 100644
+--- a/thirdparty/mingw-std-threads/mingw.shared_mutex.h
++++ b/thirdparty/mingw-std-threads/mingw.shared_mutex.h
+@@ -134,7 +134,7 @@ public:
+         using namespace std;
+ #ifndef NDEBUG
+         if (!(mCounter.fetch_sub(1, memory_order_release) & static_cast<counter_type>(~kWriteBit)))
+-            throw system_error(make_error_code(errc::operation_not_permitted));
++            __builtin_trap();
+ #else
+         mCounter.fetch_sub(1, memory_order_release);
+ #endif
+@@ -187,7 +187,7 @@ public:
+         using namespace std;
+ #ifndef NDEBUG
+         if (mCounter.load(memory_order_relaxed) != kWriteBit)
+-            throw system_error(make_error_code(errc::operation_not_permitted));
++            __builtin_trap();
+ #endif
+         mCounter.store(0, memory_order_release);
+     }
+@@ -317,9 +317,9 @@ class shared_lock
+     {
+         using namespace std;
+         if (mMutex == nullptr)
+-            throw system_error(make_error_code(errc::operation_not_permitted));
++            __builtin_trap();
+         if (mOwns)
+-            throw system_error(make_error_code(errc::resource_deadlock_would_occur));
++            __builtin_trap();
+     }
+ public:
+     typedef Mutex mutex_type;
+@@ -432,7 +432,7 @@ public:
+     {
+         using namespace std;
+         if (!mOwns)
+-            throw system_error(make_error_code(errc::operation_not_permitted));
++            __builtin_trap();
+         mMutex->unlock_shared();
+         mOwns = false;
+     }
+diff --git a/thirdparty/mingw-std-threads/mingw.thread.h b/thirdparty/mingw-std-threads/mingw.thread.h
+index bcdd1a36a8..7ca09e25f5 100644
+--- a/thirdparty/mingw-std-threads/mingw.thread.h
++++ b/thirdparty/mingw-std-threads/mingw.thread.h
+@@ -196,7 +196,7 @@ public:
+             int errnum = errno;
+             delete call;
+ //  Note: Should only throw EINVAL, EAGAIN, EACCES
+-            throw std::system_error(errnum, std::generic_category());
++            __builtin_trap();
+         } else {
+             mThreadId.mId = id_receiver;
+             mHandle = reinterpret_cast<HANDLE>(int_handle);
+@@ -213,11 +213,11 @@ public:
+     {
+         using namespace std;
+         if (get_id() == id(GetCurrentThreadId()))
+-            throw system_error(make_error_code(errc::resource_deadlock_would_occur));
++            __builtin_trap();
+         if (mHandle == kInvalidHandle)
+-            throw system_error(make_error_code(errc::no_such_process));
++            __builtin_trap();
+         if (!joinable())
+-            throw system_error(make_error_code(errc::invalid_argument));
++            __builtin_trap();
+         WaitForSingleObject(mHandle, kInfinite);
+         CloseHandle(mHandle);
+         mHandle = kInvalidHandle;
+@@ -266,7 +266,7 @@ moving another thread to it.\n");
+         if (!joinable())
+         {
+             using namespace std;
+-            throw system_error(make_error_code(errc::invalid_argument));
++            __builtin_trap();
+         }
+         if (mHandle != kInvalidHandle)
+         {