feature.
*   Full new features (e.g., a new op implementing a cutting-edge algorithm)
    typically will live in
    [tensorflow/addons](https://github.com/tensorflow/addons) to get some
    airtime before a decision is made regarding whether they are to be migrated
    to the core.
*   As every PR requires several CPU/GPU hours of CI testing, we discourage

The TensorFlow contrib module will not be included in TensorFlow 2.0.
For more information, please see:
  * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md
  * https://github.com/tensorflow/addons
  * https://github.com/tensorflow/io (for I/O related ops)
If you depend on functionality not listed there, please file an issue.
"""
contrib = _LazyLoader("contrib", globals(), "tensorflow.contrib",

                               TF_Status* status, TF_Output* dy) {
#if defined(IS_MOBILE_PLATFORM) || defined(IS_SLIM_BUILD)
  status->status = tensorflow::errors::Unimplemented(
      "Adding gradients is not supported on mobile. File a bug at "
      "https://github.com/tensorflow/tensorflow/issues if this feature is "
      "important to you");
#else

*   `tf.contrib`:

    *   `tf.contrib` has been deprecated, and functionality has been either
        migrated to the core TensorFlow API, to an ecosystem project such as
        [tensorflow/addons](https://www.github.com/tensorflow/addons) or
        [tensorflow/io](https://www.github.com/tensorflow/io), or removed
        entirely.
    *   Remove `tf.contrib.timeseries` dependency on TF distributions.

  const char* op_name_ = nullptr;
  // TODO(srbs): Use this.
  string device_name_;
};

// GraphContext wraps a TF_Graph modeling a single function and manages the
// "execution" of operation, i.e. adding them to the function.
class GraphContext : public TracingContext {
 public:
  explicit GraphContext(const char* name)
      : TracingContext(kGraph),
        graph_(new TF_Graph(), TF_DeleteGraph),

  TFE_DeleteOp(op);
  if (TF_GetCode(status) != TF_OK) return nullptr;
  CHECK_EQ(0, num_retvals);

  TF_DeleteStatus(status);

  return var_handle;
}

TFE_Op* AddOp(TFE_Context* ctx, TFE_TensorHandle* a, TFE_TensorHandle* b) {
  TF_Status* status = TF_NewStatus();

  TFE_Op* op = TFE_NewOp(ctx, "AddV2", status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

      if (!status->status.ok()) return;
      // TODO(nareshmodi): TFE_OpSetAttrFunction and TFE_OpSetAttrFunctionList
      // require TFE_Op* and just convert it internally a NameAttrValue, so
      // consider adding an overload to the C API to make this case easier.
      TFE_OpSetAttrFunction(op, attr_name, func_op);
      TFE_DeleteOp(func_op);
    } break;
    case tensorflow::AttrValue::kList: {
      // String

// Operations the tape needs to perform on tensors to do backpropagation. Named
// "vspace" because a subset of these are related to a vector space, such as
// 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

    // Make a copy so that we can mutate it.
    FunctionDef fdef_to_load = fdef;
    if (mutate_proto_func) {
      (*mutate_proto_func)(&fdef_to_load);
    }
    VLOG(1) << "Adding func to graph: " << fdef_to_load.DebugString();
    std::vector<char> binary_proto_buf(fdef_to_load.ByteSizeLong());
    fdef_to_load.SerializeToArray(binary_proto_buf.data(),

    question: optional string for how to ask for user input.
    yes_reply: optional string for reply when feature is enabled.
    no_reply: optional string for reply when feature is disabled.
    bazel_config_name: adding config to .bazelrc instead of action_env.
  """
  var = int(
      get_var(environ_cp, var_name, query_item, enabled_by_default, question,
              yes_reply, no_reply))

  if not bazel_config_name: