auto outputs = parallel_device.Join(
      /*expected_output_shapes=*/{PartialTensorShape({})}, status.get());
  ASSERT_TRUE(TF_GetCode(status.get()) == TF_OK) << TF_Message(status.get());
  const std::vector<std::unique_ptr<ParallelTensor>>& handles = *outputs;
  const std::vector<int64_t>* shape;
  absl::Status s = handles[0]->Shape(&shape);
  ASSERT_TRUE(s.ok());
  EXPECT_EQ(0, shape->size());
}

  TensorReference reference;
  std::vector<int64_t> shape;
  std::vector<int64_t> strides;
  DLManagedTensor tensor;

  explicit TfDlManagedTensorCtx(const TensorReference& ref) : reference(ref) {}
};

// Gets tensor from eager tensor handle.
const Tensor* GetTensorFromHandle(TFE_TensorHandle* h, TF_Status* status) {
  if (h == nullptr) {
    status->status = tensorflow::errors::InvalidArgument("Invalid handle");

#include "tensorflow/c/eager/c_api_experimental.h"
#include "tensorflow/c/tf_status.h"

void RegisterLoggingDevice(TFE_Context* context, const char* name,
                           bool strict_scope_placement, bool* arrived_flag,
                           bool* executed_flag, TF_Status* status);
void AllocateLoggingDevice(const char* name, bool* arrived_flag,
                           bool* executed_flag, TFE_CustomDevice** device,

  EXPECT_EQ(TF_OK, TF_GetCode(status.get())) << TF_Message(status.get());

  const char* first_device = "/job:worker/replica:0/task:1/device:CPU:0";
  const char* second_device = "/job:worker/replica:0/task:2/device:CPU:0";
  const char* device_name = "/job:localhost/replica:0/task:0/device:CUSTOM:0";
  std::array<const char*, 2> underlying_devices{first_device, second_device};

void TF_AbstractOpSetOpType(TF_AbstractOp* op, const char* const op_type,
                            TF_Status* s);
// `op_name` must outlive `op`.
void TF_AbstractOpSetOpName(TF_AbstractOp* op, const char* const op_name,
                            TF_Status* s);
// `attr_name` must outlive `op`.
void TF_AbstractOpSetAttrType(TF_AbstractOp* op, const char* const attr_name,

	// Output:
	// true
}

// Use a Scanner to implement a simple word-count utility by scanning the
// input as a sequence of space-delimited tokens.
func ExampleScanner_words() {
	// An artificial input source.
	const input = "Now is the winter of our discontent,\nMade glorious summer by this sun of York.\n"
	scanner := bufio.NewScanner(strings.NewReader(input))
	// Set the split function for the scanning operation.
	scanner.Split(bufio.ScanWords)

	// Metrics v3 endpoints
	metricsV3Path = "/metrics/v3"
)

// Standard env prometheus auth type
const (
	EnvPrometheusAuthType    = "MINIO_PROMETHEUS_AUTH_TYPE"
	EnvPrometheusOpenMetrics = "MINIO_PROMETHEUS_OPEN_METRICS"
)

type prometheusAuthType string

const (
	prometheusJWT    prometheusAuthType = "jwt"
	prometheusPublic prometheusAuthType = "public"
)

// along with this program.  If not, see <http://www.gnu.org/licenses/>.

package config

// Config value separator
const (
	ValueSeparator = ","
)

// Top level common ENVs
const (
	EnvAccessKey    = "MINIO_ACCESS_KEY"
	EnvSecretKey    = "MINIO_SECRET_KEY"
	EnvRootUser     = "MINIO_ROOT_USER"
	EnvRootPassword = "MINIO_ROOT_PASSWORD"

	// Legacy files

    // requests been enqueued.
    for (int i = 0; i < 50; ++i) {
      handles_task0.push_back(TestMatrixTensorHandle(ctx));
    }
  }
  const char task1_name[] = "/job:localhost/replica:0/task:1/device:CPU:0";
  const char task2_name[] = "/job:localhost/replica:0/task:2/device:CPU:0";

  std::vector<TFE_TensorHandle*> handles_task2;
  for (auto* h_task0 : handles_task0) {
    handles_task2.push_back(

           int32_t step = 1) {
  for (int32_t i = start; i < end; i += step) {
    (*data)[i] = i;
  }
}

// Fills out_dims with the dimensions of the given tensor.
void GetDims(const TF_Tensor* t, int64_t* out_dims) {
  int num_dims = TF_NumDims(t);
  for (int i = 0; i < num_dims; i++) {
    out_dims[i] = TF_Dim(t, i);
  }
}

// Runs model as is if output is a scalar,