*   TF Core:
    *   `tf.constant` always creates CPU tensors irrespective of the current
        device context.
    *   Eager `TensorHandles` maintain a list of mirrors for any copies to local
        or remote devices. This avoids any redundant copies due to op execution.
    *   For `tf.Tensor` & `tf.Variable`, `.experimental_ref()` is no longer
        experimental and is available as simply `.ref()`.

#
# Repacks libtensorflow tarballs into $DIR with provided $TARBALL_SUFFIX,
# and also repacks libtensorflow-src.jar into a standardized format.

# Helper function to copy a srcjar after moving any source files
# directly under the root to the "maven-style" src/main/java layout
#
# Source files generated by annotation processors appear directly
# under the root of srcjars jars created by bazel, rather than under

// in tests only.)
TF_CAPI_EXPORT int TF_PickUnusedPortOrDie(void);

// Fast path method that makes constructing a single scalar tensor require less
// overhead and copies.
TF_CAPI_EXPORT extern TFE_TensorHandle* TFE_NewTensorHandleFromScalar(
    TF_DataType data_type, void* data, size_t len, TF_Status* status);

// Specify the server_def that enables collective ops.

  tfrun python3 tensorflow/tools/ci_build/update_version.py --nightly
fi
source ci/official/utilities/get_versions.sh

# Note on gsutil commands:
# "gsutil cp" always "copies into". It cannot act on the contents of a directory
# and it does not seem possible to e.g. copy "gs://foo/bar" as anything other than
# "/path/bar". This script uses "gsutil rsync" instead, which acts on directory

# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
# any.sh has two run modes.
#
# 1. RUN, TEST, OR BUILD BAZEL TARGET(S) WITHIN A TFCI ENVIRONMENT
#    To use:
#       export TFCI=ci/official/envs/env_goes_here
#       export TF_ANY_TARGETS="quoted list of targets, like on the command line"

      std::move(maybe_parallel_results.value()));
  std::vector<MaybeParallelTensorOwned> result_content;
  result_content.reserve(parallel_results.size());
  for (std::unique_ptr<ParallelTensor>& parallel_result : parallel_results) {
    result_content.push_back(
        MaybeParallelTensorOwned(std::move(parallel_result)));
  }
  result.emplace(std::move(result_content));
  return result;
}

      : device_(device),
        tensors_(std::move(tensors)),
        shape_(std::vector<int64_t>(shape.begin(), shape.end())),
        dtype_(dtype) {}
  ParallelTensor(const ParallelDevice& device,
                 std::vector<TensorHandlePtr> tensors, const TF_DataType dtype)
      : device_(device),
        tensors_(std::move(tensors)),
        shape_(absl::nullopt),
        dtype_(dtype) {}

        return;
      } else if (execution_state_ == ExecutionState::kReadyToExecute) {
        // op_outputs_ may have been std::moved
        op_outputs_ = std::vector<TensorHandlePtr>();
        Execute(context_, operation_name_, std::move(op_inputs_), attributes_,
                expected_max_outputs_, &op_outputs_, status_.get());
        execution_state_ = ExecutionState::kHasResult;
      }
    }