namespace internal {
struct ImmediateExecutionOperationDeleter {
  void operator()(ImmediateExecutionOperation* p) const {
    if (p != nullptr) {
      p->Release();
    }
  }
};
}  // namespace internal

using ImmediateOpPtr =
    std::unique_ptr<ImmediateExecutionOperation,
                    internal::ImmediateExecutionOperationDeleter>;

}  // namespace tensorflow

};

namespace internal {
struct AbstractContextDeleter {
  void operator()(AbstractContext* p) const {
    if (p != nullptr) {
      p->Release();
    }
  }
};
}  // namespace internal

using AbstractContextPtr =
    std::unique_ptr<AbstractContext, internal::AbstractContextDeleter>;

}  // namespace tensorflow

//
// `device_name` must not name an existing physical or custom device (see
// the documentation for TFE_RegisterCustomDevice for more information).
//
// Tensors may be copied on or off the device explicitly using
// TPUReplicatedInput and TPUReplicatedOutput respectively. For example, with
// two component devices, running `x = TPUReplicatedInput(inputs=[a, b])` on the
// parallel device creates a parallel tensor `x` with `a` on the first of

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

// It's going to be hard to include a whole real JVM to test this.
// So we'll simulate a really easy JVM using just the parts we need.

// This is the relevant part of jni.h.

// On Android NDK16, jobject is defined like this in C and C++
typedef void* jobject;

typedef jobject jclass;
typedef jobject jthrowable;

  // Returns a shared reference to the wrapped function.
  virtual absl::StatusOr<core::RefCountPtr<FunctionRecord>>
  GetFunctionRecord() = 0;

 private:
  const AbstractFunctionKind kind_;
};

using AbstractFunctionPtr =
    tensorflow::core::IntrusivePtr<tensorflow::AbstractFunction>;

}  // namespace tensorflow

  using TFE_MonitoringGauge::TFE_MonitoringGauge;
};
struct TFE_MonitoringStringGauge1 : TFE_MonitoringGauge<tensorflow::string, 1> {
  using TFE_MonitoringGauge::TFE_MonitoringGauge;
};
struct TFE_MonitoringStringGauge2 : TFE_MonitoringGauge<tensorflow::string, 2> {
  using TFE_MonitoringGauge::TFE_MonitoringGauge;
};
struct TFE_MonitoringStringGauge3 : TFE_MonitoringGauge<tensorflow::string, 3> {

#include "tensorflow/c/eager/abstract_function.h"
#include "tensorflow/core/framework/function.h"
#include "tensorflow/core/platform/refcount.h"
namespace tensorflow {
namespace tracing {
namespace graph {
using tensorflow::AbstractFunction;
// Thin wrapper around a FunctionDef.
class GraphFunction : public AbstractFunction {
 public:
  explicit GraphFunction(FunctionDef fdef);
  ~GraphFunction() override;

//       if (d.IsBad()) {
//         LOG(ERROR) << "Bad Data";
//         return;
//       }
//       PushGoodData(d);
//     }
//   }
//
// You can use Cleanup<F> directly, instead of using MakeCleanup and auto,
// but there's rarely a reason to do that.
//
// You can call 'release()' on a Cleanup object to cancel the cleanup.

#ifndef TENSORFLOW_C_EXPERIMENTAL_FILESYSTEM_PLUGINS_GCS_CLEANUP_H_

                                           GradientFunction, TapeTensor> {
 public:
  using GradientTape<AbstractTensorHandle, GradientFunction,
                     TapeTensor>::GradientTape;
  // Returns whether the tape is persistent, i.e., whether the tape will hold
  // onto its internal state after a call to `ComputeGradient`.
  using GradientTape<AbstractTensorHandle, GradientFunction,
                     TapeTensor>::IsPersistent;

namespace internal {
struct AbstractOperationDeleter {
  void operator()(AbstractOperation* p) const {
    if (p != nullptr) {
      p->Release();
    }
  }
};
}  // namespace internal

using AbstractOperationPtr =
    std::unique_ptr<AbstractOperation, internal::AbstractOperationDeleter>;

}  // namespace tensorflow