Fixed const-qualification of dataFetcher in CUDA reduction

src/TNL/Containers/Algorithms/CudaReductionKernel.h

@@ -140,7 +140,7 @@ template< int blockSize,
 __global__ void
 __launch_bounds__( Reduction_maxThreadsPerBlock, Reduction_minBlocksPerMultiprocessor )
 CudaReductionWithArgumentKernel( const Result zero,
-                                 const DataFetcher dataFetcher,
+                                 DataFetcher dataFetcher,
                                  const Reduction reduction,
                                  const Index size,
                                  Result* output,
@@ -294,7 +294,7 @@ struct CudaReductionKernelLauncher
    template< typename DataFetcher,
              typename Reduction >
    int start( const Reduction& reduction,
-              const DataFetcher& dataFetcher,
+              DataFetcher& dataFetcher,
               const Result& zero,
               Result*& output )
    {
@@ -311,7 +311,7 @@ struct CudaReductionKernelLauncher
    template< typename DataFetcher,
              typename Reduction >
    int startWithArgument( const Reduction& reduction,
-                          const DataFetcher& dataFetcher,
+                          DataFetcher& dataFetcher,
                           const Result& zero,
                           Result*& output,
                           Index*& idxOutput )
@@ -395,7 +395,7 @@ struct CudaReductionKernelLauncher
                 typename Reduction >
       int launch( const Index size,
                   const Reduction& reduction,
-                  const DataFetcher& dataFetcher,
+                  DataFetcher& dataFetcher,
                   const Result& zero,
                   Result* output )
       {
@@ -498,7 +498,7 @@ struct CudaReductionKernelLauncher
                 typename Reduction >
       int launchWithArgument( const Index size,
                               const Reduction& reduction,
-                              const DataFetcher& dataFetcher,
+                              DataFetcher& dataFetcher,
                               const Result& zero,
                               Result* output,
                               Index* idxOutput,

src/TNL/Containers/Algorithms/CudaScanKernel.h

@@ -160,13 +160,13 @@ template< typename Real,
           typename Index >
 __global__ void
 cudaSecondPhaseBlockScan( Reduction reduction,
-                               const Index size,
-                               const int elementsInBlock,
-                               const Index gridIdx,
-                               const Index maxGridSize,
-                               const Real* auxArray,
-                               Real* data,
-                               Real shift )
+                          const Index size,
+                          const int elementsInBlock,
+                          const Index gridIdx,
+                          const Index maxGridSize,
+                          const Real* auxArray,
+                          Real* data,
+                          Real shift )
 {
    if( gridIdx > 0 || blockIdx.x > 0 )
       shift = reduction( shift, auxArray[ gridIdx * maxGridSize + blockIdx.x - 1 ] );

src/TNL/Containers/Algorithms/DistributedScan.h

 /***************************************************************************
-                          Scan.h  -  description
+                          DistributedScan.h  -  description
                              -------------------
     begin                : Aug 16, 2019
     copyright            : (C) 2019 by Tomas Oberhuber et al.

src/TNL/Containers/Algorithms/Scan.h

@@ -50,7 +50,7 @@ enum class ScanType {
  * See \ref Scan< Devices::Host, Type > and \ref Scan< Devices::Cuda, Type >.
  */
 template< typename Device,
-           ScanType Type = ScanType::Inclusive >
+          ScanType Type = ScanType::Inclusive >
 struct Scan;
 
 /**
@@ -93,7 +93,7 @@ struct Scan;
  * **Note: Segmented scan is not implemented for CUDA yet.**
  */
 template< typename Device,
-           ScanType Type = ScanType::Inclusive >
+          ScanType Type = ScanType::Inclusive >
 struct SegmentedScan;