#ifndef TENSORFLOW_C_EAGER_PARALLEL_DEVICE_PARALLEL_DEVICE_H_
#define TENSORFLOW_C_EAGER_PARALLEL_DEVICE_PARALLEL_DEVICE_H_

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

namespace tensorflow {
namespace parallel_device {

// Allocate a parallel device named `device_name` which forwards operations to

#ifndef TENSORFLOW_C_EAGER_PARALLEL_DEVICE_PARALLEL_DEVICE_TESTLIB_H_
#define TENSORFLOW_C_EAGER_PARALLEL_DEVICE_PARALLEL_DEVICE_TESTLIB_H_

#include <array>

#include "tensorflow/c/c_api.h"
#include "tensorflow/c/c_api_experimental.h"
#include "tensorflow/c/eager/c_api.h"
#include "tensorflow/c/eager/c_api_experimental.h"
#include "tensorflow/c/eager/parallel_device/parallel_device.h"

limitations under the License.
==============================================================================*/

#ifndef TENSORFLOW_C_EAGER_PARALLEL_DEVICE_PARALLEL_DEVICE_LIB_H_
#define TENSORFLOW_C_EAGER_PARALLEL_DEVICE_PARALLEL_DEVICE_LIB_H_

#include <memory>
#include <optional>
#include <string>
#include <utility>
#include <vector>

#include "absl/types/optional.h"

  TF_TriState pin_to_host_optimization;
  TF_TriState layout_optimizer;
  TF_TriState remapping;
  TF_TriState loop_optimization;
  TF_TriState dependency_optimization;
  TF_TriState auto_parallel;
  TF_TriState memory_optimization;
  TF_TriState scoped_allocator_optimization;
} TP_OptimizerConfigs;

#define TP_OPTIMIZER_CONFIGS_STRUCT_SIZE \

typedef struct TFE_Executor TFE_Executor;

// Creates a new eager Executor. Nodes in one executor are guaranteed to be
// executed in sequence. Assigning nodes to different executors allows executing
// nodes in parallel.
// in_flight_nodes_limit: when is_async is true, this value controls the
// maximum number of in flight async nodes. Enqueuing of additional async ops
// after the limit is reached blocks until some inflight nodes finishes.

  // Returns size of specified dimension
  //
  // -1 indicates an unknown axis length; this is unreachable for most standard
  // ImmediateExecutionTensorHandles, but comes up for example when computing
  // the shape of a parallel tensor with component shapes differing across
  // devices.
  virtual Status Dim(int dim_index, int64_t* dim) const = 0;

  // Returns the device which created the handle.

  ///   * Must set `status` to `TF_OK` if all matches were returned.
  ///   * Might use any other error value for `status` to signal other errors.
  ///
  /// DEFAULT IMPLEMENTATION: Scans the directory tree (in parallel if possible)
  /// and fills `*entries`. Needs `get_children` and `is_directory`.
  int (*get_matching_paths)(const TF_Filesystem* filesystem, const char* glob,
                            char*** entries, TF_Status* status);