// need for `Env::Default()`. At that time, we might decide to remove the need
// 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

    // If this file does not exist on server, we will need to sync it.
    bool sync_need = (status_code == TF_NOT_FOUND);
    file->plugin_file = new tf_writable_file::GCSFile(
        {std::move(bucket), std::move(object), &gcs_file->gcs_client,
         TempFile(temp_file_name, std::ios::binary | std::ios::app), sync_need,
         -1});
  } else {
    // If compose is true, we do not download anything.

        *   The random-number-generating ops in the `tf.random` module when the
            global random seed has not yet been set (via `tf.random.set_seed`).
            Throws `RuntimeError` from Python or `InvalidArgument` from C++
        *   `tf.compat.v1.get_seed` if the global random seed has not yet been
            set (via `tf.random.set_seed`). Throws `RuntimeError` from Python or
            `InvalidArgument` from C++

## Security

# limitations under the License.
# ==============================================================================
#
# Usage: rename_and_verify_wheels.sh
# This script is aware of TFCI_ variables, so it doesn't need any arguments.
# Puts new wheel through auditwheel to rename and verify it, deletes the old
# one, checks the filesize, and then ensures the new wheel is installable.
set -euxo pipefail

cd "$TFCI_OUTPUT_DIR"

computation
[**graphs**](https://developers.google.com/machine-learning/glossary/#graph).
Since models are practically programs that TensorFlow executes, using untrusted
models or graphs is equivalent to running untrusted code.

If you need to run untrusted models, execute them inside a
[**sandbox**](https://developers.google.com/code-sandboxing). Memory corruptions
in TensorFlow ops can be recognized as security issues only if they are

  // CustomDeviceOpHandler.
  virtual Status RegisterCustomDevice(const string& name,
                                      std::unique_ptr<CustomDevice> device) = 0;

  // Return FunctionLibraryDefinition. Transformations need to use it to use it
  // to invoke MLIR compiler passes.
  virtual FunctionLibraryDefinition* FuncLibDef() = 0;

  // Resets the global rendezvous used for functions.
  virtual void ResetGlobalRendezvousForFunction() = 0;

 private:
  friend class GraphContext;  // For access to op_.
  TF_Graph* g_;
  std::unique_ptr<TF_OperationDescription> op_;
  // Hold `op_type` and `op_name` till both are available since we need both
  // to build a graph operation.
  string op_type_;
  const char* op_name_ = nullptr;
  // TODO(srbs): Use this.
  string device_name_;
};

    return nullptr;
  }
  std::vector<int64_t> dimvec(num_dims);
  for (int i = 0; i < num_dims; ++i) {
    dimvec[i] = static_cast<int64_t>(dims[i]);
  }

  // TODO(apassos) do we need to wrap the deallocator here to make sure to sync
  // the device?
  TF_ManagedBuffer* buf =
      new TF_ManagedBuffer(data, len, deallocator, deallocator_arg,
                           /*owns_memory=*/false);

    visibility = ["//visibility:public"],
)

# DO NOT ADD ANY NEW EXCEPTIONS TO THIS LIST!
# Instead, please use public APIs or public build rules TF provides.
# If you need functionality that is not exposed, we will work with you to expand our public APIs.
# TODO(b/173549186): Move Google-internal TF code out of learning/brain
package_group(
    name = "internal",
    packages = [

    environ_cp['TF_NEED_CUDA'] = '0'
  else:
    environ_cp['TF_NEED_CUDA'] = str(
        int(get_var(environ_cp, 'TF_NEED_CUDA', 'CUDA', False)))
  if (environ_cp.get('TF_NEED_CUDA') == '1' and
      'TF_CUDA_CONFIG_REPO' not in environ_cp):

    set_action_env_var(
        environ_cp,
        'TF_NEED_TENSORRT',
        'TensorRT',
        False,
        bazel_config_name='tensorrt')