return tmpPl, nil
				}); err != nil {
					t.Fatalf("Unable to register pre bind plugins: %s", pl.name)
				}

				configPlugins.Reserve.Enabled = append(
					configPlugins.Reserve.Enabled,
					config.Plugin{Name: pl.name},
				)
			}
			profile := config.KubeSchedulerProfile{Plugins: configPlugins}
			ctx, cancel := context.WithCancel(context.Background())

  std::vector<Node*> outside_compilation_nodes;
  std::vector<Node*> then_branch_lifted_arg_nodes;
  outside_compilation_nodes.reserve(
      then_branch_lifted_arg_nodes_and_outside_compilation_nodes.size());
  then_branch_lifted_arg_nodes.reserve(
      then_branch_lifted_arg_nodes_and_outside_compilation_nodes.size());
  for (const auto& pair :

    const FunctionBody& fbody, absl::InlinedVector<OutputTensor, 4>* arg_nodes,
    absl::InlinedVector<OutputTensor, 4>* ret_nodes,
    absl::InlinedVector<Node*, 4>* control_ret_nodes) {
  arg_nodes->reserve(fbody.arg_nodes.size());
  ret_nodes->reserve(fbody.ret_nodes.size());
  for (auto arg : fbody.arg_nodes) {
    arg_nodes->emplace_back(arg, 0);
  }
  for (auto ret : fbody.ret_nodes) {
    ret_nodes->emplace_back(ret, 0);
  }

    // Compress the operands, region arguments, and outputs.
    SmallVector<Value, 4> new_while_operands;
    SmallVector<Type, 4> new_result_types;
    new_while_operands.reserve(new_num_operands);
    new_result_types.reserve(new_num_operands);

    // Build new operands and result type.
    for (int op_idx : llvm::seq<int>(0, old_num_operands)) {
      if (removed_operand.test(op_idx)) continue;

    SmallVector<bool, 4> removed_operand(while_op.getNumOperands(), false);
    llvm::SmallVector<Type, 4> types;
    new_operands.reserve(while_op.getNumOperands());
    new_body_yield.reserve(while_op.getNumOperands());
    types.reserve(while_op.getNumOperands());

    // Remove block arguments not used in either cond or body. This leaves the
    // block arguments of body and cond matching still.

    tensorflow::shape_inference::InferenceContext* ic, int num_dims,
    const int64_t* dims) {
  if (num_dims != -1) {
    std::vector<tensorflow::shape_inference::DimensionHandle> dim_vec;
    dim_vec.reserve(num_dims);
    for (int i = 0; i < num_dims; ++i) {
      dim_vec.push_back(ic->MakeDim(dims[i]));
    }
    return ic->MakeShape(dim_vec);
  } else {
    return ic->UnknownShape();
  }
}

  }

  // If branches have incompatible input types that means that no tensor can
  // serve as input to all the functions. Hence, the op is invalid.
  int expected_num_inputs = op->getNumOperands() - 1;
  for (int i = 0; i < expected_num_inputs; ++i) {
    SmallVector<Type, 2> branch_input_i_types;
    branch_input_i_types.reserve(branches.size());
    llvm::transform(
        branch_types, std::back_inserter(branch_input_i_types),

	}{
		{
			name:       "error reserve pod",
			sendPod:    podWithID("foo", ""),
			mockResult: mockScheduleResult{ScheduleResult{SuggestedHost: testNode.Name, EvaluatedNodes: 1, FeasibleNodes: 1}, nil},
			registerPluginFuncs: []tf.RegisterPluginFunc{
				tf.RegisterReservePlugin("FakeReserve", tf.NewFakeReservePlugin(framework.NewStatus(framework.Error, "reserve error"))),
			},

    auto input_shape = input_type.getShape();

    llvm::SmallVector<int64_t, 4> start;
    llvm::SmallVector<int64_t, 4> size;
    start.reserve(explicit_padding.size() / 2);
    size.reserve(explicit_padding.size() / 2);
    for (int i = 0, e = explicit_padding.size() / 2; i < e; ++i) {
      int64_t pre_padding = explicit_padding[2 * i];

    const std::string& summary_title) {
  std::string op_str;
  llvm::raw_string_ostream os(op_str);

  std::vector<std::string> keys;
  keys.reserve(ops.size());

  std::vector<std::string> values;
  values.reserve(ops.size());

  for (auto const& op_name_and_details : ops) {
    keys.push_back(op_name_and_details.first);
    for (auto const& op_detail : op_name_and_details.second) {