Update to clang 16.0.6. (#3859)

This PR updates cugraph to use clang 16.0.6. The previous version 16.0.1 has some minor formatting issues affecting several RAPIDS repos.

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - Brad Rees (https://github.com/BradReesWork)
  - Chuck Hastings (https://github.com/ChuckHastings)

URL: #3859

.pre-commit-config.yaml

@@ -33,7 +33,7 @@ repos:
         additional_dependencies:
           - flake8==6.0.0
   - repo: https://github.com/pre-commit/mirrors-clang-format
-    rev: v16.0.1
+    rev: v16.0.6
     hooks:
       - id: clang-format
         exclude: |

cpp/src/components/legacy/scc_matrix.cuh

@@ -68,7 +68,7 @@ struct SCC_Data {
   SCC_Data(size_t nrows,
            const IndexT* p_d_r_o,  // row_offsets
            const IndexT* p_d_c_i)
-    :                              // column indices
+    :  // column indices
       nrows_(nrows),
       p_d_r_o_(p_d_r_o),
       p_d_c_i_(p_d_c_i),

cpp/src/cores/core_number_impl.cuh

@@ -72,7 +72,7 @@ struct v_to_core_number_t {
 // a workaround for cudaErrorInvalidDeviceFunction error when device lambda is used
 template <typename edge_t>
 struct mult_degree_by_two_t {
-  __device__ edge_t operator()(edge_t d) const { return d* edge_t{2}; }
+  __device__ edge_t operator()(edge_t d) const { return d * edge_t{2}; }
 };
 
 }  // namespace

cpp/src/prims/per_v_random_select_transform_outgoing_e.cuh

@@ -287,7 +287,7 @@ rmm::device_uvector<edge_t> get_sampling_index_without_replacement(
 #ifndef NO_CUGRAPH_OPS
   edge_t mid_partition_degree_range_last = static_cast<edge_t>(K * 10);  // tuning parameter
   assert(mid_partition_degree_range_last > K);
-  size_t high_partition_over_sampling_K = K * 2;                         // tuning parameter
+  size_t high_partition_over_sampling_K = K * 2;  // tuning parameter
   assert(high_partition_over_sampling_K > K);
 
   rmm::device_uvector<edge_t> sample_nbr_indices(frontier_degrees.size() * K, handle.get_stream());
@@ -883,7 +883,7 @@ per_v_random_select_transform_e(raft::handle_t const& handle,
     sample_nbr_indices);  // neighbor index within an edge partition (note that each vertex's
                           // neighbors are distributed in minor_comm_size partitions)
   std::optional<rmm::device_uvector<size_t>> sample_key_indices{
-    std::nullopt};        // relevant only when (minor_comm_size > 1)
+    std::nullopt};  // relevant only when (minor_comm_size > 1)
   auto local_frontier_sample_counts        = std::vector<size_t>{};
   auto local_frontier_sample_displacements = std::vector<size_t>{};
   if (minor_comm_size > 1) {

cpp/src/sampling/random_walks.cuh

@@ -197,19 +197,19 @@ struct col_indx_extract_t {
   void operator()(
     original::device_vec_t<vertex_t> const& d_coalesced_src_v,  // in: coalesced vector of vertices
     original::device_vec_t<vertex_t> const&
-      d_v_col_indx,       // in: column indices, given by stepper's random engine
+      d_v_col_indx,  // in: column indices, given by stepper's random engine
     original::device_vec_t<vertex_t>&
       d_v_next_vertices,  // out: set of destination vertices, for next step
     original::device_vec_t<weight_t>&
-      d_v_next_weights)   // out: set of weights between src and destination vertices, for next step
+      d_v_next_weights)  // out: set of weights between src and destination vertices, for next step
     const
   {
     thrust::transform_if(
       handle_.get_thrust_policy(),
       thrust::make_counting_iterator<index_t>(0),
-      thrust::make_counting_iterator<index_t>(num_paths_),                         // input1
-      d_v_col_indx.begin(),                                                        // input2
-      out_degs_,                                                                   // stencil
+      thrust::make_counting_iterator<index_t>(num_paths_),  // input1
+      d_v_col_indx.begin(),                                 // input2
+      out_degs_,                                            // stencil
       thrust::make_zip_iterator(
         thrust::make_tuple(d_v_next_vertices.begin(), d_v_next_weights.begin())),  // output
       [max_depth         = max_depth_,
@@ -575,9 +575,9 @@ struct random_walker_t {
       d_crt_out_degs,  // |current set of vertex out degrees| = nelems,
                        // to be used as stencil (don't scatter if 0)
     original::device_vec_t<index_t> const&
-      d_sizes,         // paths sizes used to provide delta in coalesced paths;
-                       // pre-condition: assumed as updated to reflect new vertex additions;
-                       // also, this is the number of _vertices_ in each path;
+      d_sizes,  // paths sizes used to provide delta in coalesced paths;
+                // pre-condition: assumed as updated to reflect new vertex additions;
+                // also, this is the number of _vertices_ in each path;
     // hence for scattering weights this needs to be adjusted; hence the `adjust` parameter
     index_t
       stride,  // stride = coalesce block size (max_depth for vertices; max_depth-1 for weights)
@@ -762,7 +762,7 @@ random_walks_impl(
   // pre-allocate num_paths * max_depth;
   //
   original::device_vec_t<vertex_t> d_coalesced_v(num_paths * max_depth,
-                                                 stream);         // coalesced vertex set
+                                                 stream);  // coalesced vertex set
   original::device_vec_t<weight_t> d_coalesced_w(num_paths * (max_depth - 1),
                                                  stream);         // coalesced weight set
   original::device_vec_t<index_t> d_paths_sz(num_paths, stream);  // paths sizes

cpp/src/structure/renumber_edgelist_impl.cuh

@@ -86,7 +86,7 @@ struct find_unused_id_t {
     for (size_t i = worker_id; i < sorted_local_vertices.size() + size_t{1}; i += num_workers) {
       auto start = (i == size_t{0}) ? std::numeric_limits<vertex_t>::lowest()
                                     : sorted_local_vertices[i - size_t{1}];
-      if (start != std::numeric_limits<vertex_t>::max()) { ++start; };            // now inclusive
+      if (start != std::numeric_limits<vertex_t>::max()) { ++start; };  // now inclusive
       auto end = (i == sorted_local_vertices.size()) ? std::numeric_limits<vertex_t>::max()
                                                      : sorted_local_vertices[i];  // exclusive
       for (vertex_t v = start; v < end; ++v) {

cpp/src/traversal/bfs_impl.cuh

@@ -73,7 +73,7 @@ struct e_op_t {
       if (*(prev_visited_flags + packed_bool_offset(dst)) &
           packed_bool_mask(dst)) {  // check if unvisited in previous iterations
         push = false;
-      } else {                      // check if unvisited in this iteration as well
+      } else {  // check if unvisited in this iteration as well
         auto old = visited_flags.atomic_or(dst, true);
         push     = !old;
       }

cpp/tests/prims/mg_extract_transform_e.cu

@@ -157,8 +157,8 @@ class Tests_MGExtractTransformE
     // 1. create MG graph
 
     constexpr bool is_multi_gpu     = true;
-    constexpr bool renumber         = true;    // needs to be true for multi gpu case
-    constexpr bool store_transposed = false;   // needs to be false for using extract_transform_e
+    constexpr bool renumber         = true;   // needs to be true for multi gpu case
+    constexpr bool store_transposed = false;  // needs to be false for using extract_transform_e
     if (cugraph::test::g_perf) {
       RAFT_CUDA_TRY(cudaDeviceSynchronize());  // for consistent performance measurement
       handle_->get_comms().barrier();

cpp/tests/traversal/mg_sssp_test.cpp

@@ -214,7 +214,7 @@ class Tests_MGSSSP : public ::testing::TestWithParam<std::tuple<SSSP_Usecase, in
         auto h_sg_predecessors = cugraph::test::to_host(*handle_, d_sg_predecessors);
 
         auto max_weight_element = std::max_element(h_sg_weights.begin(), h_sg_weights.end());
-        auto epsilon            = *max_weight_element* weight_t{1e-6};
+        auto epsilon            = *max_weight_element * weight_t{1e-6};
         auto nearly_equal       = [epsilon](auto lhs, auto rhs) {
           return std::fabs(lhs - rhs) < epsilon;
         };

cpp/tests/traversal/sssp_test.cpp

@@ -222,7 +222,7 @@ class Tests_SSSP : public ::testing::TestWithParam<std::tuple<SSSP_Usecase, inpu
       }
 
       auto max_weight_element = std::max_element(h_weights.begin(), h_weights.end());
-      auto epsilon            = *max_weight_element* weight_t{1e-6};
+      auto epsilon            = *max_weight_element * weight_t{1e-6};
       auto nearly_equal = [epsilon](auto lhs, auto rhs) { return std::fabs(lhs - rhs) < epsilon; };
 
       ASSERT_TRUE(std::equal(h_reference_distances.begin(),

cpp/tests/utilities/test_utilities.hpp

@@ -547,7 +547,7 @@ mg_vertex_property_values_to_sg_vertex_property_values(
   std::optional<raft::device_span<vertex_t const>>
     sg_renumber_map,  // std::nullopt if the SG graph is not renumbered
   std::optional<raft::device_span<vertex_t const>>
-    mg_vertices,      // std::nullopt if the entire local vertex partition range is assumed
+    mg_vertices,  // std::nullopt if the entire local vertex partition range is assumed
   raft::device_span<value_t const> mg_values);
 
 template <typename type_t>