// used to cancel all other operations if any fails.
  //
  // Set step_id to configure the step id used for rendezvous creation. step id
  // of value -1 is reserved for global rendezvous and should not be set here.
  void StartExecute(TFE_Context* context,
                    const std::vector<ParallelTensor*>& inputs,
                    const char* operation_name, const TFE_OpAttrs* attributes,

TF_CAPI_EXPORT extern bool TFE_ExecutorIsAsync(TFE_Executor*);

// Causes the calling thread to block till all ops dispatched in this executor
// have been executed. Note that "execution" here refers to kernel execution /
// scheduling of copies, etc. Similar to sync execution, it doesn't guarantee
// that lower level device queues (like GPU streams) have been flushed.
//

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

// --------------------------------------------------------------------------
// Experimental C API for TensorFlow.
//
// The API here is subject to changes in the future.
// --------------------------------------------------------------------------

#ifdef __cplusplus
extern "C" {
#endif

// When `enable` is true, set

// adding gradients, getting zeroes, etc. Currently cannot be implemented
// without using tensorflow python code, hence left unspecified here.
//
// Gradient is the type returned by gradient functions. In Python TF it's either
// Tensor or IndexedSlices or None, which here we map to nullptr. Gradients need
// to allow their size to be computed and they need to be passable to a backward

// You must use TF_AddInput() for the first input (since it takes a
// single tensor), and TF_AddInputList() for the second input (since
// it takes a list, even if you were to pass a list with a single
// tensor), as in:
//   TF_OperationDescription* desc = TF_NewOperation(graph, "Concat", "c");
//   TF_Output concat_dim_input = {...};
//   TF_AddInput(desc, concat_dim_input);

// Macro to verify that the field `struct_size` of STRUCT_OBJ is initialized.
// `struct_size` is used for struct member compatibility check between core TF
// and plug-ins with the same C API minor version. More info here:
// https://github.com/tensorflow/community/blob/master/rfcs/20200612-stream-executor-c-api/C_API_versioning_strategy.md
#define TF_VALIDATE_STRUCT_SIZE(STRUCT_NAME, STRUCT_OBJ, SIZE_VALUE_NAME) \

// for `Env::Default()` altogether, but that's a different project, not in
// scope for now. I'm just mentioning this here as that transition will mean
// removal of the registration part from `Env` and adding it here instead: we
// will need tables to hold for each scheme the function tables that implement
// the needed functionality instead of the current `FileSystemRegistry` code in

#ifndef TENSORFLOW_C_EAGER_C_API_H_
#define TENSORFLOW_C_EAGER_C_API_H_

// C API extensions to experiment with eager execution of kernels.
// WARNING: Unlike tensorflow/c/c_api.h, the API here is not guaranteed to be
// stable and can change without notice.

#include "tensorflow/c/c_api.h"
#include "tensorflow/c/c_api_macros.h"

#ifdef __cplusplus
extern "C" {
#endif

//
// `ZerosLike` is not expected to be called and returns a nullptr. The creation
// of default zeros grads is handled by the `DefaultGradientFunction` registered
// for each op.
// TODO(srbs): We need to define `ZerosLike` here to keep the compiler happy.
// Figure out a way to avoid this.
// TODO(srbs): Should ZerosLike check-fail instead of returning nullptr?
class TapeTensor {
 public:
  explicit TapeTensor(AbstractTensorHandle* handle);

/// data in the structures here to prevent cases when padding/packing enabled by
/// different compiler options breaks compatibility. For more details about how
/// this is used, please consult next sections. Here we just wrap these tables
/// in lint warnings so that changes here cause changes to the versioning data
/// as well. Here is a short summary of what changes are allowed: