Uglify par_backend namespace

[rodgert]

Committed by other, non-phab, means

include/pstl/internal/algorithm_impl.h

 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _Function, class _IsVector>
 void
 __pattern_walk1(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _Function __f,
                 _IsVector __is_vector,
                 /*parallel=*/std::true_type)
 {
     internal::__except_handler([&]() {
-        par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+        __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                       [__f, __is_vector](_ForwardIterator __i, _ForwardIterator __j) {
                                           internal::__brick_walk1(__i, __j, __f, __is_vector);
                                       });
     });
 }
 #endif
 
 template <class _ExecutionPolicy, class _ForwardIterator, class _Brick>
 void
 __pattern_walk_brick(_ExecutionPolicy&&, _ForwardIterator __first, _ForwardIterator __last, _Brick __brick,
                      /*parallel=*/std::false_type) noexcept
 {
     __brick(__first, __last);
 }
 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _Brick>
 void
 __pattern_walk_brick(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _Brick __brick,
                      /*parallel=*/std::true_type)
 {
     internal::__except_handler([&]() {
-        par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+        __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                       [__brick](_ForwardIterator __i, _ForwardIterator __j) { __brick(__i, __j); });
     });
 }
 #endif
 
 //------------------------------------------------------------------------
 // walk1_n
 //------------------------------------------------------------------------
 template <class _ForwardIterator, class _Size, class _Function>

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator, class _Size, class _Brick>
 _RandomAccessIterator
 __pattern_walk_brick_n(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _Size __n, _Brick __brick,
                        /*is_parallel=*/std::true_type)
 {
     return internal::__except_handler([&]() {
-        par_backend::__parallel_for(
+        __par_backend::__parallel_for(
             std::forward<_ExecutionPolicy>(__exec), __first, __first + __n,
             [__brick](_RandomAccessIterator __i, _RandomAccessIterator __j) { __brick(__i, __j - __i); });
         return __first + __n;
     });
 }
 #endif
 
 //------------------------------------------------------------------------

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator1, class _ForwardIterator2, class _Function, class _IsVector>
 _ForwardIterator2
 __pattern_walk2(_ExecutionPolicy&& __exec, _ForwardIterator1 __first1, _ForwardIterator1 __last1,
                 _ForwardIterator2 __first2, _Function __f, _IsVector __is_vector, /*parallel=*/std::true_type)
 {
     return internal::__except_handler([&]() {
-        par_backend::__parallel_for(
+        __par_backend::__parallel_for(
             std::forward<_ExecutionPolicy>(__exec), __first1, __last1,
             [__f, __first1, __first2, __is_vector](_ForwardIterator1 __i, _ForwardIterator1 __j) {
                 internal::__brick_walk2(__i, __j, __first2 + (__i - __first1), __f, __is_vector);
             });
         return __first2 + (__last1 - __first1);
     });
 }
 #endif

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator1, class _RandomAccessIterator2, class _Brick>
 _RandomAccessIterator2
 __pattern_walk2_brick(_ExecutionPolicy&& __exec, _RandomAccessIterator1 __first1, _RandomAccessIterator1 __last1,
                       _RandomAccessIterator2 __first2, _Brick __brick, /*parallel=*/std::true_type)
 {
     return __except_handler([&]() {
-        par_backend::__parallel_for(
+        __par_backend::__parallel_for(
             std::forward<_ExecutionPolicy>(__exec), __first1, __last1,
             [__first1, __first2, __brick](_RandomAccessIterator1 __i, _RandomAccessIterator1 __j) {
                 __brick(__i, __j, __first2 + (__i - __first1));
             });
         return __first2 + (__last1 - __first1);
     });
 }
 #endif
 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator1, class _Size, class _RandomAccessIterator2, class _Brick>
 _RandomAccessIterator2
 __pattern_walk2_brick_n(_ExecutionPolicy&& __exec, _RandomAccessIterator1 __first1, _Size __n,
                         _RandomAccessIterator2 __first2, _Brick __brick, /*parallel=*/std::true_type)
 {
     return __except_handler([&]() {
-        par_backend::__parallel_for(
+        __par_backend::__parallel_for(
             std::forward<_ExecutionPolicy>(__exec), __first1, __first1 + __n,
             [__first1, __first2, __brick](_RandomAccessIterator1 __i, _RandomAccessIterator1 __j) {
                 __brick(__i, __j - __i, __first2 + (__i - __first1));
             });
         return __first2 + __n;
     });
 }
 #endif

 template <class _ExecutionPolicy, class _RandomAccessIterator1, class _RandomAccessIterator2,
           class _RandomAccessIterator3, class _Function, class _IsVector>
 _RandomAccessIterator3
 __pattern_walk3(_ExecutionPolicy&& __exec, _RandomAccessIterator1 __first1, _RandomAccessIterator1 __last1,
                 _RandomAccessIterator2 __first2, _RandomAccessIterator3 __first3, _Function __f, _IsVector __is_vector,
                 /*parallel=*/std::true_type)
 {
     return internal::__except_handler([&]() {
-        par_backend::__parallel_for(
+        __par_backend::__parallel_for(
             std::forward<_ExecutionPolicy>(__exec), __first1, __last1,
             [__f, __first1, __first2, __first3, __is_vector](_RandomAccessIterator1 __i, _RandomAccessIterator1 __j) {
                 internal::__brick_walk3(__i, __j, __first2 + (__i - __first1), __first3 + (__i - __first1), __f,
                                         __is_vector);
             });
         return __first3 + (__last1 - __first1);
     });
 }

 _OutputIterator
 __pattern_copy_if(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                   _OutputIterator __result, _UnaryPredicate __pred, _IsVector __is_vector, /*parallel=*/std::true_type)
 {
     typedef typename std::iterator_traits<_RandomAccessIterator>::difference_type _DifferenceType;
     const _DifferenceType __n = __last - __first;
     if (_DifferenceType(1) < __n)
     {
-        par_backend::__buffer<bool> __mask_buf(__n);
+        __par_backend::__buffer<bool> __mask_buf(__n);
         return __except_handler([&__exec, __n, __first, __result, __is_vector, __pred, &__mask_buf]() {
             bool* __mask = __mask_buf.get();
             _DifferenceType __m{};
-            par_backend::parallel_strict_scan(
+            __par_backend::parallel_strict_scan(
                 std::forward<_ExecutionPolicy>(__exec), __n, _DifferenceType(0),
                 [=](_DifferenceType __i, _DifferenceType __len) { // Reduce
                     return __brick_calc_mask_1<_DifferenceType>(__first + __i, __first + (__i + __len), __mask + __i,
                                                                 __pred, __is_vector)
                         .first;
                 },
                 std::plus<_DifferenceType>(),                                                // Combine
                 [=](_DifferenceType __i, _DifferenceType __len, _DifferenceType __initial) { // Scan

 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _Predicate, class _IsVector>
 typename std::iterator_traits<_ForwardIterator>::difference_type
 __pattern_count(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _Predicate __pred,
                 /* is_parallel */ std::true_type, _IsVector __is_vector)
 {
     typedef typename std::iterator_traits<_ForwardIterator>::difference_type _SizeType;
     return __except_handler([&]() {
-        return par_backend::__parallel_reduce(
+        return __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first, __last, _SizeType(0),
             [__pred, __is_vector](_ForwardIterator __begin, _ForwardIterator __end, _SizeType __value) -> _SizeType {
                 return __value + __brick_count(__begin, __end, __pred, __is_vector);
             },
             std::plus<_SizeType>());
     });
 }
 #endif

 template <class _ExecutionPolicy, class _ForwardIterator, class _CalcMask, class _IsVector>
 _ForwardIterator
 remove_elements(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _CalcMask __calc_mask,
                 _IsVector __is_vector)
 {
     typedef typename std::iterator_traits<_ForwardIterator>::difference_type _DifferenceType;
     typedef typename std::iterator_traits<_ForwardIterator>::value_type _Tp;
     _DifferenceType __n = __last - __first;
-    par_backend::__buffer<bool> __mask_buf(__n);
+    __par_backend::__buffer<bool> __mask_buf(__n);
     // 1. find a first iterator that should be removed
     return __except_handler([&]() {
         bool* __mask = __mask_buf.get();
-        _DifferenceType __min = par_backend::__parallel_reduce(
+        _DifferenceType __min = __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), _DifferenceType(0), __n, __n,
             [__first, __mask, &__calc_mask, __is_vector](_DifferenceType __i, _DifferenceType __j,
                                                          _DifferenceType __local_min) -> _DifferenceType {
                 // Create mask
                 __calc_mask(__mask + __i, __mask + __j, __first + __i);
 
                 // if minimum was found in a previous range we shouldn't do anymore
                 if (__local_min < __i)
return __except_handler([&]() {
         // No elements to remove - exit
         if (__min == __n)
         {
             return __last;
         }
         __n -= __min;
         __first += __min;
 
-        par_backend::__buffer<_Tp> __buf(__n);
+        __par_backend::__buffer<_Tp> __buf(__n);
         _Tp* __result = __buf.get();
         __mask += __min;
         _DifferenceType __m{};
         // 2. Elements that doesn't satisfy pred are moved to result
-        par_backend::parallel_strict_scan(
+        __par_backend::parallel_strict_scan(
             std::forward<_ExecutionPolicy>(__exec), __n, _DifferenceType(0),
             [__mask, __is_vector](_DifferenceType __i, _DifferenceType __len) {
                 return __brick_count(__mask + __i, __mask + __i + __len, [](bool __val) { return __val; }, __is_vector);
             },
             std::plus<_DifferenceType>(),
             [=](_DifferenceType __i, _DifferenceType __len, _DifferenceType __initial) {
                 __brick_copy_by_mask(__first + __i, __first + __i + __len, __result + __initial, __mask + __i,
                                      [](_ForwardIterator __x, _Tp* __z) {
                                          __invoke_if_else(std::is_trivial<_Tp>(), [&]() { *__z = std::move(*__x); },
                                                           [&]() { ::new (std::addressof(*__z)) _Tp(std::move(*__x)); });
                                      },
                                      __is_vector);
             },
             [&__m](_DifferenceType __total) { __m = __total; });
 
         // 3. Elements from result are moved to [first, last)
-        par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __result, __result + __m,
+        __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __result, __result + __m,
                                       [__result, __first, __is_vector](_Tp* __i, _Tp* __j) {
                                           __brick_move(__i, __j, __first + (__i - __result), __is_vector);
                                       });
         return __first + __m;
     });
 }
 #endif
 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _BinaryPredicate, class _IsVector>
 _ForwardIterator

 __pattern_unique_copy(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                       _OutputIterator __result, _BinaryPredicate __pred, _IsVector __is_vector,
                       /*parallel=*/std::true_type)
 {
     typedef typename std::iterator_traits<_RandomAccessIterator>::difference_type _DifferenceType;
     const _DifferenceType __n = __last - __first;
     if (_DifferenceType(2) < __n)
     {
-        par_backend::__buffer<bool> __mask_buf(__n);
+        __par_backend::__buffer<bool> __mask_buf(__n);
         if (_DifferenceType(2) < __n)
         {
             return internal::__except_handler([&__exec, __n, __first, __result, __pred, __is_vector, &__mask_buf]() {
                 bool* __mask = __mask_buf.get();
                 _DifferenceType __m{};
-                par_backend::parallel_strict_scan(
+                __par_backend::parallel_strict_scan(
                     std::forward<_ExecutionPolicy>(__exec), __n, _DifferenceType(0),
                     [=](_DifferenceType __i, _DifferenceType __len) -> _DifferenceType { // Reduce
                         _DifferenceType __extra = 0;
                         if (__i == 0)
                         {
                             // Special boundary case
                             __mask[__i] = true;
                             if (--__len == 0)

 }
 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _BidirectionalIterator, class _IsVector>
 void
 __pattern_reverse(_ExecutionPolicy&& __exec, _BidirectionalIterator __first, _BidirectionalIterator __last,
                   _IsVector __is_vector, /*is_parallel=*/std::true_type)
 {
-    par_backend::__parallel_for(
+    __par_backend::__parallel_for(
         std::forward<_ExecutionPolicy>(__exec), __first, __first + (__last - __first) / 2,
         [__is_vector, __first, __last](_BidirectionalIterator __inner_first, _BidirectionalIterator __inner_last) {
             __brick_reverse(__inner_first, __inner_last, __last - (__inner_first - __first), __is_vector);
         });
 }
 #endif
 
 //------------------------------------------------------------------------

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _BidirectionalIterator, class _OutputIterator, class _IsVector>
 _OutputIterator
 __pattern_reverse_copy(_ExecutionPolicy&& __exec, _BidirectionalIterator __first, _BidirectionalIterator __last,
                        _OutputIterator __d_first, _IsVector __is_vector, /*is_parallel=*/std::true_type)
 {
     auto __len = __last - __first;
-    par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+    __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                   [__is_vector, __first, __len, __d_first](_BidirectionalIterator __inner_first,
                                                                            _BidirectionalIterator __inner_last) {
                                       __brick_reverse_copy(__inner_first, __inner_last,
                                                            __d_first + (__len - (__inner_last - __first)), __is_vector);
                                   });
     return __d_first + __len;
 }
 #endif
 
 //------------------------------------------------------------------------
 // rotate
 //------------------------------------------------------------------------
 template <class _ForwardIterator>

 __pattern_rotate(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __middle,
                  _ForwardIterator __last, _IsVector __is_vector, /*is_parallel=*/std::true_type)
 {
     typedef typename std::iterator_traits<_ForwardIterator>::value_type _Tp;
     auto __n = __last - __first;
     auto __m = __middle - __first;
     if (__m <= __n / 2)
     {
-        par_backend::__buffer<_Tp> __buf(__n - __m);
+        __par_backend::__buffer<_Tp> __buf(__n - __m);
         return __except_handler([&__exec, __n, __m, __first, __middle, __last, __is_vector, &__buf]() {
             _Tp* __result = __buf.get();
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __middle, __last,
-                                        [__middle, __result, __is_vector](_ForwardIterator __b, _ForwardIterator __e) {
-                                            __brick_uninitialized_move(__b, __e, __result + (__b - __middle),
-                                                                       __is_vector);
-                                        });
+            __par_backend::__parallel_for(
+                std::forward<_ExecutionPolicy>(__exec), __middle, __last,
+                    __brick_uninitialized_move(__b, __e, __result + (__b - __middle), __is_vector);
 
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __middle,
+            __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __middle,
                                           [__last, __middle, __is_vector](_ForwardIterator __b, _ForwardIterator __e) {
                                               __brick_move(__b, __e, __b + (__last - __middle), __is_vector);
                                           });
 
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __result, __result + (__n - __m),
+            __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __result, __result + (__n - __m),
                                           [__first, __result, __is_vector](_Tp* __b, _Tp* __e) {
                                               __brick_move(__b, __e, __first + (__b - __result), __is_vector);
                                           });
 
             return __first + (__last - __middle);
         });
     }
     else
     {
-        par_backend::__buffer<_Tp> __buf(__m);
+        __par_backend::__buffer<_Tp> __buf(__m);
         return __except_handler([&__exec, __n, __m, __first, __middle, __last, __is_vector, &__buf]() {
             _Tp* __result = __buf.get();
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __middle,
+            __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __middle,
                                           [__first, __result, __is_vector](_ForwardIterator __b, _ForwardIterator __e) {
                                               __brick_uninitialized_move(__b, __e, __result + (__b - __first),
                                                                          __is_vector);
                                           });
 
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __middle, __last,
+            __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __middle, __last,
                                           [__first, __middle, __is_vector](_ForwardIterator __b, _ForwardIterator __e) {
                                               __brick_move(__b, __e, __first + (__b - __middle), __is_vector);
                                           });
 
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __result, __result + __m,
+            __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __result, __result + __m,
                                           [__n, __m, __first, __result, __is_vector](_Tp* __b, _Tp* __e) {
                                               __brick_move(__b, __e, __first + ((__n - __m) + (__b - __result)),
                                                            __is_vector);
                                           });
 
             return __first + (__last - __middle);
         });
     }
 }
 #endif
 
 //------------------------------------------------------------------------

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _OutputIterator, class _IsVector>
 _OutputIterator
 __pattern_rotate_copy(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __middle,
                       _ForwardIterator __last, _OutputIterator __result, _IsVector __is_vector,
                       /*is_parallel=*/std::true_type)
 {
-    par_backend::__parallel_for(
+    __par_backend::__parallel_for(
         std::forward<_ExecutionPolicy>(__exec), __first, __last,
         [__first, __last, __middle, __result, __is_vector](_ForwardIterator __b, _ForwardIterator __e) {
             if (__b > __middle)
             {
                 __brick_copy(__b, __e, __result + (__b - __middle), __is_vector);
             }
             else
             {
else
             _ReduceType __init = __not_init;
 
             // Array with states that we'll have when state from the left branch is merged with state from the right branch.
             // State is calculated by formula: new_state = table[left_state * 4 + right_state]
             _ReduceType __table[] = {__broken,     __broken,     __broken,     __broken, __broken,    __all_true,
                                      __true_false, __true_false, __broken,     __broken, __all_false, __broken,
                                      __broken,     __broken,     __true_false, __broken};
 
-            __init = par_backend::__parallel_reduce(
+            __init = __par_backend::__parallel_reduce(
                 std::forward<_ExecutionPolicy>(__exec), __first, __last, __init,
                 [&__pred, &__table, __is_vector](_ForwardIterator __i, _ForwardIterator __j,
                                                  _ReduceType __value) -> _ReduceType {
                     if (__value == __broken)
                     {
                         return __broken;
                     }
                     _ReduceType __res = __not_init;
return __except_handler([&]() {
             if (__val1.__end == __val1.__pivot)
             {
                 return {__new_begin, __val2.__pivot, __val2.__end};
             }
             // if true part of right range greater than false part of left range
             // then we should swap the false part of left range and last part of true part of right range
             else if (__size2 > __size1)
             {
-                par_backend::__parallel_for(
+                __par_backend::__parallel_for(
                     std::forward<_ExecutionPolicy>(__exec), __val1.__pivot, __val1.__pivot + __size1,
                     [__val1, __val2, __size1, __is_vector](_ForwardIterator __i, _ForwardIterator __j) {
                         __brick_swap_ranges(__i, __j, (__val2.__pivot - __size1) + (__i - __val1.__pivot), __is_vector);
                     });
                 return {__new_begin, __val2.__pivot - __size1, __val2.__end};
             }
             // else we should swap the first part of false part of left range and true part of right range
             else
             {
-                par_backend::__parallel_for(
+                __par_backend::__parallel_for(
                     std::forward<_ExecutionPolicy>(__exec), __val1.__pivot, __val1.__pivot + __size2,
                     [__val1, __val2, __is_vector](_ForwardIterator __i, _ForwardIterator __j) {
                         __brick_swap_ranges(__i, __j, __val2.__begin + (__i - __val1.__pivot), __is_vector);
                     });
                 return {__new_begin, __val1.__pivot + __size2, __val2.__end};
             }
         };
 
-        _PartitionRange __result = par_backend::__parallel_reduce(
+        _PartitionRange __result = __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first, __last, __init,
             [__pred, __is_vector, __reductor](_ForwardIterator __i, _ForwardIterator __j,
                                               _PartitionRange __value) -> _PartitionRange {
                 //1. serial partition
                 _ForwardIterator __pivot = __brick_partition(__i, __j, __pred, __is_vector);
 
                 // 2. merging of two ranges (left and right respectively)
                 return __reductor(__value, {__i, __pivot, __j});
return __except_handler([&]() {
             // then we should swap the false part of left range and last part of true part of right range
             else
             {
                 __brick_rotate(__val1.__pivot, __val2.__begin, __val2.__pivot, __is_vector);
                 return {__new_begin, __val2.__pivot - __size1, __val2.__end};
             }
         };
 
-        _PartitionRange __result = par_backend::__parallel_reduce(
+        _PartitionRange __result = __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first, __last, __init,
             [&__pred, __is_vector, __reductor](_BidirectionalIterator __i, _BidirectionalIterator __j,
                                                _PartitionRange __value) -> _PartitionRange {
                 //1. serial stable_partition
                 _BidirectionalIterator __pivot = __brick_stable_partition(__i, __j, __pred, __is_vector);
 
                 // 2. merging of two ranges (left and right respectively)
                 return __reductor(__value, {__i, __pivot, __j});
__pattern_partition_copy(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                          _OutputIterator1 __out_true, _OutputIterator2 __out_false, _UnaryPredicate __pred,
                          _IsVector __is_vector, /*is_parallelization=*/std::true_type)
 {
     typedef typename std::iterator_traits<_RandomAccessIterator>::difference_type _DifferenceType;
     typedef std::pair<_DifferenceType, _DifferenceType> _ReturnType;
     const _DifferenceType __n = __last - __first;
     if (_DifferenceType(1) < __n)
     {
-        par_backend::__buffer<bool> __mask_buf(__n);
+        __par_backend::__buffer<bool> __mask_buf(__n);
         return internal::__except_handler(
             [&__exec, __n, __first, __out_true, __out_false, __is_vector, __pred, &__mask_buf]() {
                 bool* __mask = __mask_buf.get();
                 _ReturnType __m{};
-                par_backend::parallel_strict_scan(
+                __par_backend::parallel_strict_scan(
                     std::forward<_ExecutionPolicy>(__exec), __n, std::make_pair(_DifferenceType(0), _DifferenceType(0)),
                     [=](_DifferenceType __i, _DifferenceType __len) { // Reduce
                         return internal::__brick_calc_mask_1<_DifferenceType>(__first + __i, __first + (__i + __len),
                                                                               __mask + __i, __pred, __is_vector);
                     },
                     [](const _ReturnType& __x, const _ReturnType& __y) -> _ReturnType {
                         return std::make_pair(__x.first + __y.first, __x.second + __y.second);
                     },                                                                       // Combine

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator, class _Compare, class _IsVector>
 void
 __pattern_sort(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last, _Compare __comp,
                _IsVector /*is_vector*/, /*is_parallel=*/std::true_type, /*is_move_constructible=*/std::true_type)
 {
     __except_handler([&]() {
-        par_backend::__parallel_stable_sort(std::forward<_ExecutionPolicy>(__exec), __first, __last, __comp,
+        __par_backend::__parallel_stable_sort(std::forward<_ExecutionPolicy>(__exec), __first, __last, __comp,
                                               [](_RandomAccessIterator __first, _RandomAccessIterator __last,
                                                  _Compare __comp) { std::sort(__first, __last, __comp); },
                                               __last - __first);
     });
 }
 #endif
 
 //------------------------------------------------------------------------
 // stable_sort
 //------------------------------------------------------------------------
 
 template <class _ExecutionPolicy, class _RandomAccessIterator, class _Compare, class _IsVector>
 void
 __pattern_stable_sort(_ExecutionPolicy&&, _RandomAccessIterator __first, _RandomAccessIterator __last, _Compare __comp,
                       _IsVector /*is_vector*/, /*is_parallel=*/std::false_type) noexcept
 {
     std::stable_sort(__first, __last, __comp);
 }
 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator, class _Compare, class _IsVector>
 void
 __pattern_stable_sort(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                       _Compare __comp, _IsVector /*is_vector*/, /*is_parallel=*/std::true_type)
 {
     internal::__except_handler([&]() {
-        par_backend::__parallel_stable_sort(std::forward<_ExecutionPolicy>(__exec), __first, __last, __comp,
+        __par_backend::__parallel_stable_sort(std::forward<_ExecutionPolicy>(__exec), __first, __last, __comp,
                                               [](_RandomAccessIterator __first, _RandomAccessIterator __last,
                                                  _Compare __comp) { std::stable_sort(__first, __last, __comp); });
     });
 }
 #endif
 
 //------------------------------------------------------------------------
 // partial_sort
 //------------------------------------------------------------------------
 

 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator, class _Compare, class _IsVector>
 void
 __pattern_partial_sort(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __middle,
                        _RandomAccessIterator __last, _Compare __comp, _IsVector, /*is_parallel=*/std::true_type)
 {
     const auto __n = __middle - __first;
     __except_handler([&]() {
-        par_backend::__parallel_stable_sort(
+        __par_backend::__parallel_stable_sort(
             std::forward<_ExecutionPolicy>(__exec), __first, __last, __comp,
             [__n](_RandomAccessIterator __begin, _RandomAccessIterator __end, _Compare __comp) {
                 if (__n < __end - __begin)
                     std::partial_sort(__begin, __begin + __n, __end, __comp);
                 else
                     std::sort(__begin, __end, __comp);
             },
             __n);
__pattern_partial_sort_copy(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last,
     {
         return __d_first;
     }
     auto __n1 = __last - __first;
     auto __n2 = __d_last - __d_first;
     return __except_handler([&]() {
         if (__n2 >= __n1)
         {
-            par_backend::__parallel_stable_sort(
+            __par_backend::__parallel_stable_sort(
                 std::forward<_ExecutionPolicy>(__exec), __d_first, __d_first + __n1, __comp,
                 [__first, __d_first, __is_vector](_RandomAccessIterator __i, _RandomAccessIterator __j,
                                                   _Compare __comp) {
                     _ForwardIterator __i1 = __first + (__i - __d_first);
                     _ForwardIterator __j1 = __first + (__j - __d_first);
 
                 // 1. Copy elements from input to output
 #if !__PSTL_ICC_18_OMP_SIMD_BROKEN
                     __brick_copy(__i1, __j1, __i, __is_vector);
 #else
                     std::copy(__i1, __j1, __i);
 #endif
                     // 2. Sort elements in output sequence
                     std::sort(__i, __j, __comp);
                 },
                 __n1);
             return __d_first + __n1;
         }
         else
         {
             typedef typename std::iterator_traits<_ForwardIterator>::value_type _T1;
             typedef typename std::iterator_traits<_RandomAccessIterator>::value_type _T2;
-            par_backend::__buffer<_T1> __buf(__n1);
+            __par_backend::__buffer<_T1> __buf(__n1);
             _T1* __r = __buf.get();
 
-            par_backend::__parallel_stable_sort(std::forward<_ExecutionPolicy>(__exec), __r, __r + __n1, __comp,
+            __par_backend::__parallel_stable_sort(std::forward<_ExecutionPolicy>(__exec), __r, __r + __n1, __comp,
                                                   [__n2, __first, __r](_T1* __i, _T1* __j, _Compare __comp) {
                                                       _ForwardIterator __it = __first + (__i - __r);
 
                                                       // 1. Copy elements from input to raw memory
                                                       for (_T1* __k = __i; __k != __j; ++__k, ++__it)
                                                       {
                                                           ::new (__k) _T2(*__it);
                                                       }
 
                                                       // 2. Sort elements in temporary __buffer
                                                       if (__n2 < __j - __i)
                                                           std::partial_sort(__i, __i + __n2, __j, __comp);
                                                       else
                                                           std::sort(__i, __j, __comp);
                                                   },
                                                   __n2);
 
             // 3. Move elements from temporary __buffer to output
-            par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __r, __r + __n2,
+            __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __r, __r + __n2,
                                           [__r, __d_first, __is_vector](_T1* __i, _T1* __j) {
                                               __brick_move(__i, __j, __d_first + (__i - __r), __is_vector);
                                           });
             return __d_first + __n2;
         }
     });
 }
 #endif
 
 //------------------------------------------------------------------------
 // adjacent_find

 __pattern_adjacent_find(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                         _BinaryPredicate __pred, /* is_parallel */ std::true_type, _IsVector __is_vector,
                         bool __or_semantic)
 {
     if (__last - __first < 2)
         return __last;
 
     return internal::__except_handler([&]() {
-        return par_backend::__parallel_reduce(
+        return __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first, __last, __last,
             [__last, __pred, __is_vector, __or_semantic](_RandomAccessIterator __begin, _RandomAccessIterator __end,
                                                          _RandomAccessIterator __value) -> _RandomAccessIterator {
                 // TODO: investigate performance benefits from the use of shared variable for the result,
                 // checking (compare_and_swap idiom) its __value at __first.
                 if (__or_semantic && __value < __last)
                 { //found
-                    par_backend::__cancel_execution();
+                    __par_backend::__cancel_execution();
                     return __value;
                 }
 
                 if (__value > __begin)
                 {
                     // modify __end to check the predicate on the boundary __values;
                     // TODO: to use a custom range with boundaries overlapping
                     // TODO: investigate what if we remove "if" below and run algorithm on range [__first, __last-1)

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _Tp, class _IsVector>
 _ForwardIterator
 __pattern_fill(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, const _Tp& __value,
                /*is_parallel=*/std::true_type, _IsVector __is_vector)
 {
     return __except_handler([&__exec, __first, __last, &__value, __is_vector]() {
-        par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+        __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                       [&__value, __is_vector](_ForwardIterator __begin, _ForwardIterator __end) {
                                           internal::__brick_fill(__begin, __end, __value, __is_vector);
                                       });
         return __last;
     });
 }
 #endif
 
 template <class _OutputIterator, class _Size, class _Tp>
 _OutputIterator
 __brick_fill_n(_OutputIterator __first, _Size __count, const _Tp& __value, /* __is_vector = */ std::true_type) noexcept

 
 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _ForwardIterator, class _Generator, class _IsVector>
 _ForwardIterator
 __pattern_generate(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _Generator __g,
                    /*is_parallel=*/std::true_type, _IsVector __is_vector)
 {
     return internal::__except_handler([&]() {
-        par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+        __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                       [__g, __is_vector](_ForwardIterator __begin, _ForwardIterator __end) {
                                           internal::__brick_generate(__begin, __end, __g, __is_vector);
                                       });
         return __last;
     });
 }
 #endif
 
 template <class OutputIterator, class Size, class _Generator>
 OutputIterator
 __brick_generate_n(OutputIterator __first, Size __count, _Generator __g, /* is_vector = */ std::true_type) noexcept

 #if __PSTL_USE_PAR_POLICIES
 template <class _ExecutionPolicy, class _RandomAccessIterator1, class _RandomAccessIterator2, class _OutputIterator,
           class _Compare, class _IsVector>
 _OutputIterator
 __pattern_merge(_ExecutionPolicy&& __exec, _RandomAccessIterator1 __first1, _RandomAccessIterator1 __last1,
                 _RandomAccessIterator2 __first2, _RandomAccessIterator2 __last2, _OutputIterator __d_first,
                 _Compare __comp, _IsVector __is_vector, /* is_parallel = */ std::true_type)
 {
-    par_backend::__parallel_merge(
+    __par_backend::__parallel_merge(
         std::forward<_ExecutionPolicy>(__exec), __first1, __last1, __first2, __last2, __d_first, __comp,
         [__is_vector](_RandomAccessIterator1 __f1, _RandomAccessIterator1 __l1, _RandomAccessIterator2 __f2,
                       _RandomAccessIterator2 __l2, _OutputIterator __f3,
                       _Compare __comp) { return __brick_merge(__f1, __l1, __f2, __l2, __f3, __comp, __is_vector); });
     return __d_first + (__last1 - __first1) + (__last2 - __first2);
 }
 #endif
 
__pattern_inplace_merge(_ExecutionPolicy&& __exec, _BidirectionalIterator __first, _BidirectionalIterator __middle,
                         /*is_parallel=*/std::true_type)
 {
     if (__first == __last || __first == __middle || __middle == __last)
     {
         return;
     }
     typedef typename std::iterator_traits<_BidirectionalIterator>::value_type _Tp;
     auto __n = __last - __first;
-    par_backend::__buffer<_Tp> __buf(__n);
+    __par_backend::__buffer<_Tp> __buf(__n);
     _Tp* __r = __buf.get();
     __except_handler([&]() {
         auto __move_values = [](_BidirectionalIterator __x, _Tp* __z) {
             __invoke_if_else(std::is_trivial<_Tp>(), [&]() { *__z = std::move(*__x); },
                              [&]() { ::new (std::addressof(*__z)) _Tp(std::move(*__x)); });
         };
 
         auto __move_sequences = [](_BidirectionalIterator __first1, _BidirectionalIterator __last1, _Tp* __first2) {
             return __brick_uninitialized_move(__first1, __last1, __first2, _IsVector());
         };
 
-        par_backend::__parallel_merge(
+        __par_backend::__parallel_merge(
             std::forward<_ExecutionPolicy>(__exec), __first, __middle, __middle, __last, __r, __comp,
             [__n, __move_values, __move_sequences](_BidirectionalIterator __f1, _BidirectionalIterator __l1,
                                                    _BidirectionalIterator __f2, _BidirectionalIterator __l2, _Tp* __f3,
                                                    _Compare __comp) {
-                auto __func = par_backend::__serial_move_merge<decltype(__move_values), decltype(__move_sequences)>(
+                auto __func = __par_backend::__serial_move_merge<decltype(__move_values), decltype(__move_sequences)>(
                     __n, __move_values, __move_sequences);
                 __func(__f1, __l1, __f2, __l2, __f3, __comp);
                 return __f3 + (__l1 - __f1) + (__l2 - __f2);
             });
-        par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __r, __r + __n,
+        __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __r, __r + __n,
                                       [__r, __first, __is_vector](_Tp* __i, _Tp* __j) {
                                           __brick_move(__i, __j, __first + (__i - __r), __is_vector);
                                       });
     });
 }
 #endif
 
 //------------------------------------------------------------------------
 // includes
 //------------------------------------------------------------------------
 
struct _SetRange
         {
             return __len == 0;
         }
     };
 
     const _DifferenceType __n1 = __last1 - __first1;
     const _DifferenceType __n2 = __last2 - __first2;
 
-    par_backend::__buffer<_T> __buf(__size_func(__n1, __n2));
+    __par_backend::__buffer<_T> __buf(__size_func(__n1, __n2));
 
     return __except_handler([&__exec, __n1, __first1, __last1, __first2, __last2, __result, __is_vector, __comp,
                              __size_func, __set_op, &__buf]() {
         auto __buffer = __buf.get();
         _DifferenceType __m{};
         auto __scan = [=](_DifferenceType, _DifferenceType, const _SetRange& __s) { // Scan
             if (!__s.empty())
                 __brick_move(__buffer + __s.__buf_pos, __buffer + (__s.__buf_pos + __s.__len), __result + __s.__pos,
                              __is_vector);
         };
-        par_backend::parallel_strict_scan(
+        __par_backend::parallel_strict_scan(
             std::forward<_ExecutionPolicy>(__exec), __n1, _SetRange{0, 0, 0}, //-1, 0},
             [=](_DifferenceType __i, _DifferenceType __len) {                 // Reduce
                 //[__b; __e) - a subrange of the first sequence, to reduce
                 _ForwardIterator1 __b = __first1 + __i, __e = __first1 + (__i + __len);
 
                 //try searching for the first element which not equal to *__b
                 if (__b != __first1)
                     __b = std::upper_bound(__b, __last1, *__b, __comp);
if (__n1 == 0)
                                      std::true_type());
 
     // testing  whether the sequences are intersected
     _ForwardIterator1 __left_bound_seq_1 = lower_bound(__first1, __last1, *__first2, __comp);
 
     if (__left_bound_seq_1 == __last1)
     {
         //{1} < {2}: seq2 is wholly greater than seq1, so, do parallel copying seq1 and seq2
-        par_backend::__parallel_invoke(std::forward<_ExecutionPolicy>(__exec),
+        __par_backend::__parallel_invoke(std::forward<_ExecutionPolicy>(__exec),
                                          [=] {
                                              __pattern_walk2_brick(std::forward<_ExecutionPolicy>(__exec), __first1,
                                                                    __last1, __result, copy_range1, std::true_type());
                                          },
                                          [=] {
                                              __pattern_walk2_brick(std::forward<_ExecutionPolicy>(__exec), __first2,
                                                                    __last2, __result + __n1, copy_range2,
                                                                    std::true_type());
                                          });
         return __result + __n1 + __n2;
     }
 
     // testing  whether the sequences are intersected
     _ForwardIterator2 __left_bound_seq_2 = lower_bound(__first2, __last2, *__first1, __comp);
 
     if (__left_bound_seq_2 == __last2)
     {
         //{2} < {1}: seq2 is wholly greater than seq1, so, do parallel copying seq1 and seq2
-        par_backend::__parallel_invoke(std::forward<_ExecutionPolicy>(__exec),
+        __par_backend::__parallel_invoke(std::forward<_ExecutionPolicy>(__exec),
                                          [=] {
                                              __pattern_walk2_brick(std::forward<_ExecutionPolicy>(__exec), __first2,
                                                                    __last2, __result, copy_range2, std::true_type());
                                          },
                                          [=] {
                                              __pattern_walk2_brick(std::forward<_ExecutionPolicy>(__exec), __first1,
                                                                    __last1, __result + __n2, copy_range1,
                                                                    std::true_type());
                                          });
         return __result + __n1 + __n2;
     }
 
     const auto __m1 = __left_bound_seq_1 - __first1;
     if (__m1 > __set_algo_cut_off)
     {
         auto __res_or = __result;
         __result += __m1; //we know proper offset due to [first1; left_bound_seq_1) < [first2; last2)
-        par_backend::__parallel_invoke(
+        __par_backend::__parallel_invoke(
             std::forward<_ExecutionPolicy>(__exec),
             //do parallel copying of [first1; left_bound_seq_1)
             [=] {
                 __pattern_walk2_brick(std::forward<_ExecutionPolicy>(__exec), __first1, __left_bound_seq_1, __res_or,
                                       copy_range1, std::true_type());
             },
             [=, &__result] {
                 __result = __parallel_set_op(std::forward<_ExecutionPolicy>(__exec), __left_bound_seq_1, __last1,
                                              __first2, __last2, __result, __comp,
                                              [](_DifferenceType __n, _DifferenceType __m) { return __n + __m; },
                                              __set_union_op, __is_vector);
             });
         return __result;
     }
 
     const auto __m2 = __left_bound_seq_2 - __first2;
     assert(__m1 == 0 || __m2 == 0);
     if (__m2 > __set_algo_cut_off)
     {
         auto __res_or = __result;
         __result += __m2; //we know proper offset due to [first2; left_bound_seq_2) < [first1; last1)
-        par_backend::__parallel_invoke(
+        __par_backend::__parallel_invoke(
             std::forward<_ExecutionPolicy>(__exec),
             //do parallel copying of [first2; left_bound_seq_2)
             [=] {
                 __pattern_walk2_brick(std::forward<_ExecutionPolicy>(__exec), __first2, __left_bound_seq_2, __res_or,
                                       copy_range2, std::true_type());
             },
             [=, &__result] {
                 __result = __parallel_set_op(std::forward<_ExecutionPolicy>(__exec), __first1, __last1,

 _RandomAccessIterator
 __pattern_min_element(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                       _Compare __comp, _IsVector __is_vector, /* is_parallel = */ std::true_type)
 {
     if (__first == __last)
         return __last;
 
     return internal::__except_handler([&]() {
-        return par_backend::__parallel_reduce(
+        return __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first + 1, __last, __first,
             [=](_RandomAccessIterator __begin, _RandomAccessIterator __end,
                 _RandomAccessIterator __init) -> _RandomAccessIterator {
                 const _RandomAccessIterator subresult = __brick_min_element(__begin, __end, __comp, __is_vector);
                 return internal::__cmp_iterators_by_values(__init, subresult, __comp);
             },
             [=](_RandomAccessIterator __it1, _RandomAccessIterator __it2) -> _RandomAccessIterator {
                 return internal::__cmp_iterators_by_values(__it1, __it2, __comp);
__pattern_minmax_element(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _Compare __comp,
                          _IsVector __is_vector, /* is_parallel = */ std::true_type)
 {
     if (__first == __last)
         return std::make_pair(__first, __first);
 
     return internal::__except_handler([&]() {
         typedef std::pair<_ForwardIterator, _ForwardIterator> _Result;
 
-        return par_backend::__parallel_reduce(
+        return __par_backend::__parallel_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first + 1, __last, std::make_pair(__first, __first),
             [=](_ForwardIterator __begin, _ForwardIterator __end, _Result __init) -> _Result {
                 const _Result __subresult = __brick_minmax_element(__begin, __end, __comp, __is_vector);
                 return std::make_pair(internal::__cmp_iterators_by_values(__subresult.first, __init.first, __comp),
                                       internal::__cmp_iterators_by_values(__init.second, __subresult.second,
                                                                           internal::__not_pred<_Compare>(__comp)));
             },
             [=](_Result __p1, _Result __p2) -> _Result {

include/pstl/internal/numeric_impl.h

 template <class _ExecutionPolicy, class _RandomAccessIterator1, class _RandomAccessIterator2, class _Tp,
           class _BinaryOperation1, class _BinaryOperation2, class _IsVector>
 _Tp
 __pattern_transform_reduce(_ExecutionPolicy&& __exec, _RandomAccessIterator1 __first1, _RandomAccessIterator1 __last1,
                            _RandomAccessIterator2 __first2, _Tp __init, _BinaryOperation1 __binary_op1,
                            _BinaryOperation2 __binary_op2, _IsVector __is_vector, /*is_parallel=*/std::true_type)
 {
     return internal::__except_handler([&]() {
-        return par_backend::__parallel_transform_reduce(
+        return __par_backend::__parallel_transform_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first1, __last1,
             [__first1, __first2, __binary_op2](_RandomAccessIterator1 __i) mutable {
                 return __binary_op2(*__i, *(__first2 + (__i - __first1)));
             },
             __init,
             __binary_op1, // Combine
             [__first1, __first2, __binary_op1, __binary_op2,
              __is_vector](_RandomAccessIterator1 __i, _RandomAccessIterator1 __j, _Tp __init) -> _Tp {

 template <class _ExecutionPolicy, class _ForwardIterator, class _Tp, class _BinaryOperation, class _UnaryOperation,
           class _IsVector>
 _Tp
 __pattern_transform_reduce(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator __last, _Tp __init,
                            _BinaryOperation __binary_op, _UnaryOperation __unary_op, _IsVector __is_vector,
                            /*is_parallel=*/std::true_type)
 {
     return __except_handler([&]() {
-        return par_backend::__parallel_transform_reduce(
+        return __par_backend::__parallel_transform_reduce(
             std::forward<_ExecutionPolicy>(__exec), __first, __last,
             [__unary_op](_ForwardIterator __i) mutable { return __unary_op(*__i); }, __init, __binary_op,
             [__unary_op, __binary_op, __is_vector](_ForwardIterator __i, _ForwardIterator __j, _Tp __init) {
                 return __brick_transform_reduce(__i, __j, __init, __binary_op, __unary_op, __is_vector);
             });
     });
 }
 #endif

 typename std::enable_if<!std::is_floating_point<_Tp>::value, _OutputIterator>::type
 __pattern_transform_scan(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
                          _OutputIterator __result, _UnaryOperation __unary_op, _Tp __init, _BinaryOperation __binary_op,
                          _Inclusive, _IsVector __is_vector, /*is_parallel=*/std::true_type)
 {
     typedef typename std::iterator_traits<_RandomAccessIterator>::difference_type _DifferenceType;
 
     return internal::__except_handler([&]() {
-        par_backend::__parallel_transform_scan(
+        __par_backend::__parallel_transform_scan(
             std::forward<_ExecutionPolicy>(__exec), __last - __first,
             [__first, __unary_op](_DifferenceType __i) mutable { return __unary_op(__first[__i]); }, __init,
             __binary_op,
             [__first, __unary_op, __binary_op](_DifferenceType __i, _DifferenceType __j, _Tp __init) {
                 // Execute serial __brick_transform_reduce, due to the explicit SIMD vectorization (reduction) requires a commutative operation for the guarantee of correct scan.
                 return internal::__brick_transform_reduce(__first + __i, __first + __j, __init, __binary_op, __unary_op,
                                                           /*__is_vector*/ std::false_type());
             },
__pattern_transform_scan(_ExecutionPolicy&& __exec, _RandomAccessIterator __first, _RandomAccessIterator __last,
     typedef typename std::iterator_traits<_RandomAccessIterator>::difference_type _DifferenceType;
     _DifferenceType __n = __last - __first;
 
     if (__n <= 0)
     {
         return __result;
     }
     return __except_handler([&]() {
-        par_backend::parallel_strict_scan(
+        __par_backend::parallel_strict_scan(
             std::forward<_ExecutionPolicy>(__exec), __n, __init,
             [__first, __unary_op, __binary_op, __result, __is_vector](_DifferenceType __i, _DifferenceType __len) {
                 return __brick_transform_scan(__first + __i, __first + (__i + __len), __result + __i, __unary_op, _Tp{},
                                               __binary_op, _Inclusive(), __is_vector)
                     .second;
             },
             __binary_op,
             [__result, &__binary_op](_DifferenceType __i, _DifferenceType __len, _Tp __initial) {
__pattern_adjacent_difference(_ExecutionPolicy&& __exec, _ForwardIterator1 __first, _ForwardIterator1 __last,
                               _ForwardIterator2 __d_first, _BinaryOperation __op, _IsVector __is_vector,
                               /*is_parallel=*/std::true_type)
 {
     assert(__first != __last);
     typedef typename std::iterator_traits<_ForwardIterator1>::reference _ReferenceType1;
     typedef typename std::iterator_traits<_ForwardIterator2>::reference _ReferenceType2;
 
     *__d_first = *__first;
-    par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last - 1,
-                                [&__op, __is_vector, __d_first, __first](_ForwardIterator1 __b, _ForwardIterator1 __e) {
-                                    _ForwardIterator2 __d_b = __d_first + (__b - __first);
-                                    __brick_walk3(__b, __e, __b + 1, __d_b + 1,
-                                                  [&__op](_ReferenceType1 __x, _ReferenceType1 __y,
-                                                          _ReferenceType2 __z) { __z = __op(__y, __x); },
-                                                  __is_vector);
-                                });
+    __par_backend::__parallel_for(
+        std::forward<_ExecutionPolicy>(__exec), __first, __last - 1,
+            __brick_walk3(
+                __b, __e, __b + 1, __d_b + 1,
+                [&__op](_ReferenceType1 __x, _ReferenceType1 __y, _ReferenceType2 __z) { __z = __op(__y, __x); },
     return __d_first + (__last - __first);
 }
 #endif
 
 } // namespace internal
 } // namespace __pstl
 
 #endif /* __PSTL_numeric_impl_H */

include/pstl/internal/parallel_backend_tbb.h

 #include <tbb/tbb_allocator.h>
 
 #if TBB_INTERFACE_VERSION < 10000
 #error Intel(R) Threading Building Blocks 2018 is required; older versions are not supported.
 #endif
 
 namespace __pstl
 {
-namespace par_backend
+namespace __par_backend
 {
 
 //! Raw memory buffer with automatic freeing and no exceptions.
 /** Some of our algorithms need to start with raw memory buffer,
 not an initialize array, because initialization/destruction
 would make the span be at least O(N). */
 // tbb::allocator can improve performance in some cases.
 template <typename _Tp>

 __upsweep(_Index __i, _Index __m, _Index __tilesize, _Tp* __r, _Index __lastsize, _Rp __reduce, _Cp __combine)
 {
     if (__m == 1)
         __r[0] = __reduce(__i * __tilesize, __lastsize);
     else
     {
         _Index __k = __split(__m);
         tbb::parallel_invoke(
-            [=] { par_backend::__upsweep(__i, __k, __tilesize, __r, __tilesize, __reduce, __combine); },
-            [=] {
-                par_backend::__upsweep(__i + __k, __m - __k, __tilesize, __r + __k, __lastsize, __reduce, __combine);
-            });
+            [=] { __upsweep(__i, __k, __tilesize, __r, __tilesize, __reduce, __combine); },
+            [=] { __upsweep(__i + __k, __m - __k, __tilesize, __r + __k, __lastsize, __reduce, __combine); });
         if (__m == 2 * __k)
             __r[__m - 1] = __combine(__r[__k - 1], __r[__m - 1]);
     }
 }
 
 template <typename _Index, typename _Tp, typename _Cp, typename _Sp>
 void
 __downsweep(_Index __i, _Index __m, _Index __tilesize, _Tp* __r, _Index __lastsize, _Tp __initial, _Cp __combine,
             _Sp __scan)
 {
     if (__m == 1)
         __scan(__i * __tilesize, __lastsize, __initial);
     else
     {
-        const _Index __k = par_backend::__split(__m);
-        tbb::parallel_invoke(
-            [=] { par_backend::__downsweep(__i, __k, __tilesize, __r, __tilesize, __initial, __combine, __scan); },
+        const _Index __k = __split(__m);
+        tbb::parallel_invoke([=] { __downsweep(__i, __k, __tilesize, __r, __tilesize, __initial, __combine, __scan); },
                              // Assumes that __combine never throws.
                              //TODO: Consider adding a requirement for user functors to be constant.
                              [=, &__combine] {
-                par_backend::__downsweep(__i + __k, __m - __k, __tilesize, __r + __k, __lastsize,
+                                 __downsweep(__i + __k, __m - __k, __tilesize, __r + __k, __lastsize,
                                              __combine(__initial, __r[__k - 1]), __combine, __scan);
                              });
     }
 }
 
 // Adapted from Intel(R) Cilk(TM) version from cilkpub.
 // Let i:len denote a counted interval of length n starting at i.  s denotes a generalized-sum value.
 // Expected actions of the functors are:
 //     reduce(i,len) -> s  -- return reduction value of i:len.
 //     combine(s1,s2) -> s -- return merged sum
tbb::this_task_arena::isolate([=, &__combine]() {
         if (__n > 1)
         {
             _Index __p = tbb::this_task_arena::max_concurrency();
             const _Index __slack = 4;
             _Index __tilesize = (__n - 1) / (__slack * __p) + 1;
             _Index __m = (__n - 1) / __tilesize;
             __buffer<_Tp> __buf(__m + 1);
             _Tp* __r = __buf.get();
-            par_backend::__upsweep(_Index(0), _Index(__m + 1), __tilesize, __r, __n - __m * __tilesize, __reduce,
-                                   __combine);
+            __upsweep(_Index(0), _Index(__m + 1), __tilesize, __r, __n - __m * __tilesize, __reduce, __combine);
             // When __apex is a no-op and __combine has no side effects, a good optimizer
             // should be able to eliminate all code between here and __apex.
             // Alternatively, provide a default value for __apex that can be
             // recognized by metaprogramming that conditionlly executes the following.
             size_t __k = __m + 1;
             _Tp __t = __r[__k - 1];
             while ((__k &= __k - 1))
                 __t = __combine(__r[__k - 1], __t);
             __apex(__combine(__initial, __t));
-            par_backend::__downsweep(_Index(0), _Index(__m + 1), __tilesize, __r, __n - __m * __tilesize, __initial,
-                                     __combine, __scan);
+            __downsweep(_Index(0), _Index(__m + 1), __tilesize, __r, __n - __m * __tilesize, __initial, __combine,
+                        __scan);
             return;
         }
         // Fewer than 2 elements in sequence, or out of memory.  Handle has single block.
         _Tp __sum = __initial;
         if (__n)
             __sum = __combine(__sum, __reduce(_Index(0), __n));
         __apex(__sum);
         if (__n)
__parallel_merge(_ExecutionPolicy&&, _RandomAccessIterator1 __xs, _RandomAccessIterator1 __xe,
     {
         // Fall back on serial merge
         __leaf_merge(__xs, __xe, __ys, __ye, __zs, __comp);
     }
     else
     {
         tbb::this_task_arena::isolate([=]() {
             typedef __merge_task<_RandomAccessIterator1, _RandomAccessIterator2, _RandomAccessIterator3, _Compare,
-                                 par_backend::__binary_no_op, _LeafMerge>
+                                 __binary_no_op, _LeafMerge>
                 _TaskType;
             tbb::task::spawn_root_and_wait(*new (tbb::task::allocate_root()) _TaskType(
-                __xs, __xe, __ys, __ye, __zs, __comp, par_backend::__binary_no_op(), __leaf_merge));
+                __xs, __xe, __ys, __ye, __zs, __comp, __binary_no_op(), __leaf_merge));
         });
     }
 }
 
 //------------------------------------------------------------------------
 // parallel_invoke
 //------------------------------------------------------------------------
 template <class _ExecutionPolicy, typename _F1, typename _F2>
 void
 __parallel_invoke(_ExecutionPolicy&&, _F1&& __f1, _F2&& __f2)
 {
     //TODO: a version of tbb::this_task_arena::isolate with variadic arguments pack should be added in the future
     tbb::this_task_arena::isolate([&]() { tbb::parallel_invoke(std::forward<_F1>(__f1), std::forward<_F2>(__f2)); });
 }
 
-} // namespace par_backend
+} // namespace __par_backend
 } // namespace __pstl
 
 #endif /* __PSTL_parallel_backend_tbb_H */

include/pstl/internal/parallel_backend_utils.h

 #define __PSTL_parallel_backend_utils_H
 
 #include <iterator>
 #include <utility>
 #include "utils.h"
 
 namespace __pstl
 {
-namespace par_backend
+namespace __par_backend
 {
 
 //! Destroy sequence [xs,xe)
 struct __serial_destroy
 {
     template <typename _RandomAccessIterator>
     void
     operator()(_RandomAccessIterator __zs, _RandomAccessIterator __ze)
public:
     pop()
     {
         assert(_M_ptr > _M_buf.get());
         --_M_ptr;
         (*_M_ptr).~_ValueType();
     }
 };
 
-} // namespace par_backend
+} // namespace __par_backend
 } // namespace __pstl
 
 #endif /* __PSTL_parallel_backend_utils_H */

include/pstl/internal/parallel_impl.h

 _Index
 __parallel_find(_ExecutionPolicy&& __exec, _Index __first, _Index __last, _Brick __f, _Compare __comp, bool __b_first)
 {
     typedef typename std::iterator_traits<_Index>::difference_type _DifferenceType;
     const _DifferenceType __n = __last - __first;
     _DifferenceType __initial_dist = __b_first ? __n : -1;
     std::atomic<_DifferenceType> __extremum(__initial_dist);
     // TODO: find out what is better here: parallel_for or parallel_reduce
-    par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+    __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                   [__comp, __f, __first, &__extremum](_Index __i, _Index __j) {
                                       // See "Reducing Contention Through Priority Updates", PPoPP '13, for discussion of
                                       // why using a shared variable scales fairly well in this situation.
                                       if (__comp(__i - __first, __extremum))
                                       {
                                           _Index __res = __f(__i, __j);
                                           // If not '__last' returned then we found what we want so put this to extremum
                                           if (__res != __j)
                                           {
                                               const _DifferenceType __k = __res - __first;
                                               for (_DifferenceType __old = __extremum; __comp(__k, __old);
                                                    __old = __extremum)
                                               {
                                                   __extremum.compare_exchange_weak(__old, __k);
                                               }
                                           }
                                       }
                                   });
     return __extremum != __initial_dist ? __first + __extremum : __last;
 }
 
 //------------------------------------------------------------------------
 // parallel_or
 //------------------------------------------------------------------------
 //! Return true if brick f[i,j) returns true for some subrange [i,j) of [first,last)
 template <class _ExecutionPolicy, class _Index, class _Brick>
 bool
 __parallel_or(_ExecutionPolicy&& __exec, _Index __first, _Index __last, _Brick __f)
 {
     std::atomic<bool> __found(false);
-    par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
+    __par_backend::__parallel_for(std::forward<_ExecutionPolicy>(__exec), __first, __last,
                                   [__f, &__found](_Index __i, _Index __j) {
                                       if (!__found.load(std::memory_order_relaxed) && __f(__i, __j))
                                       {
                                           __found.store(true, std::memory_order_relaxed);
-                                        par_backend::__cancel_execution();
+                                          __par_backend::__cancel_execution();
                                       }
                                   });
     return __found;
 }
 
 } // namespace internal
 } // namespace __pstl
 
 #endif /* __PSTL_parallel_impl_H */