int num_dims, void* data, size_t len,
    void (*deallocator)(void* data, size_t len, void* arg),
    void* deallocator_arg, TF_Status* status);

// Retrieves the address space (i.e. job, replia, task) of the local host and
// saves it in the buffer.
TF_CAPI_EXPORT extern void TFE_HostAddressSpace(TFE_Context* ctx,
                                                TF_Buffer* buf);

// Get the OpList of all OpDefs defined in this address space.
// Returns a TF_Buffer, ownership of which is transferred to the caller
// (and can be freed using TF_DeleteBuffer).
//
// The data in the buffer will be the serialized OpList proto for ops registered
// in this address space.
TF_CAPI_EXPORT extern TF_Buffer* TF_GetAllOpList(void);

  // run and have results. If is_async=true it means that all of the
  // device-specific executors have scheduled the op.
  //
  // Accepts inferred shapes for outputs (`expected_output_shapes`), which if
  // fully defined will avoid querying the shapes of the underlying
  // TensorHandles when ParallelTensor::Shape is called. This allows async
  // computation to continue without blocking.
  //

//
// Like a TF_Tensor, a TFE_TensorHandle refers to a tensor with a value, shape,
// type etc. Unlike a TF_Tensor, a TFE_TensorHandle may refer to such tensors
// placed in the memory of different devices or remote address spaces.
typedef struct TFE_TensorHandle TFE_TensorHandle;

TF_CAPI_EXPORT extern TFE_TensorHandle* TFE_NewTensorHandle(const TF_Tensor* t,
                                                            TF_Status* status);

TF_CAPI_EXPORT extern void TFE_InferShapes(
    TFE_Op* op, TF_ShapeAndTypeList* input_shapes, TF_Tensor** input_tensors,
    TF_ShapeAndTypeList* input_tensor_as_shapes,
    TF_ShapeAndTypeList** input_resource_shapes_and_types,
    TF_ShapeAndTypeList** output_shapes,
    TF_ShapeAndTypeList*** output_resource_shapes_and_types, TF_Status* status);

TF_CAPI_EXPORT extern void
TF_ImportGraphDefOptionsSetValidateColocationConstraints(

class ForwardAccumulator {
 public:
  // Does not take ownership of `vspace`, which must outlive the
  // ForwardAccumulator.
  explicit ForwardAccumulator(
      const VSpace<Gradient, BackwardFunction, TapeTensor>& vspace,
      bool use_batch)
      : vspace_(vspace), use_batch_(use_batch) {
    call_state_.emplace(nullptr, false);
  }

  virtual ~ForwardAccumulator() {

// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

#include <string.h>

#include "_cgo_export.h"

void
callback(void *f)
{
	// use some stack space
	volatile char data[64*1024];

	data[0] = 0;
	goCallback(f);
        data[sizeof(data)-1] = 0;
}

void
callGoFoo(void)
{
	extern void goFoo(void);
	goFoo();
}

void
IntoC(void)
{

  /// Plugins:
  ///   * Must set `status` to `TF_OK` if exactly `n` bytes have been written.
  ///   * Must set `status` to `TF_RESOURCE_EXHAUSTED` if fewer than `n` bytes
  ///     have been written, potentially due to quota/disk space.
  ///   * Might use any other error value for `status` to signal other errors.
  void (*append)(const TF_WritableFile* file, const char* buffer, size_t n,
                 TF_Status* status);

namespace tensorflow {

// Set the shapes and types of the output's handle.
//
// The lengths of the arrays pointed to by `shapes`, `ranks`, and `types` must
// all be equal to `num_shapes_and_types`. If `ranks[i] != -1`, (i.e., if the
// rank is known), then it must be equal to the length of `shapes[i]`; if
// `ranks[i] == 1`, then `shapes[i]` may be nullptr.
//

  const string DebugString() const;

  // Returns a map from variable names to their shapes.  Slices of a partitioned
  // tensor are combined into a single entry.
  const TensorSliceReader::VarToShapeMap& GetVariableToShapeMap() const;

  // Returns a map from variable names to their data types.  Slices of a
  // partitioned tensor are combined into a single entry.