lengths[index] = s.size();
    if ((p + s.size()) > (static_cast<char*>(storage) + storage_size)) {
      tsl::Set_TF_Status_from_Status(
          status,
          absl::InvalidArgumentError(
              "Not enough storage to hold the requested list of nodes"));
      return;
    }
    memcpy(values[index], s.data(), s.size());
    p += s.size();
    index++;
  }
}

[here](https://github.com/tensorflow/tensorflow/commit/052445e04ce20fd747657e0198a1bcec2b6dff5b),
for an example.

See
[this commit](https://github.com/tensorflow/tensorflow/commit/5f7f05a80aac9b01325a78ec3fcff0dbedb1cc23)
as a rough example of the steps below.

All the files referenced below are located in the same directory as this README,
unless indicated otherwise.

1) Add the new version to the `VERSIONS` variable inside

                                               TF_Status* status,
                                               void* device_info);

  // Method to execute an operation.
  //
  // Arguments provide enough information to reconstruct the original `TFE_Op`,
  // or construct a transformed version, by inspecting the passed `op`.
  //
  // TFE_OpGetDevice(op) records the original placement of the operation. It may

### Saved graphs and checkpoints

When loading untrusted serialized computation graphs (in form of a `GraphDef`,
`SavedModel`, or equivalent on-disk format), the set of computation primitives
available to TensorFlow is powerful enough that you should assume that the
TensorFlow process effectively executes arbitrary code.

The risk of loading untrusted checkpoints depends on the code or graph that you

//   * allows having newer plugins than the current core TensorFlow: the
//     additional entries in the plugin's table are just discarded;
//   * allows having older plugins than the current core TensorFlow (though
//     we are still warning users): the entries that core TensorFlow expects
//     but plugins didn't provide will be set to `nullptr` values and core
//     TensorFlow will know to not call these on behalf of users;

        TensorFlow.

*   `tf.experimental.dtensor`: Added DTensor, an extension to TensorFlow for
    large-scale modeling with minimal changes to user code. You are welcome to
    try it out, though be aware that the DTensor API is experimental and up-to
    backward-incompatible changes. DTensor and Keras integration is published
    under `tf.keras.dtensor` in this release (refer to the `tf.keras` entry).

  status = env_->NewRandomAccessFile(filepath, &read_file);
  if (!status.ok())
    GTEST_SKIP() << "NewRandomAccessFile() not supported: " << status;

  char scratch[64 /* big enough to accommodate test_data */] = {0};
  StringPiece result;
  status = read_file->Read(0, test_data.size(), &result, scratch);
  EXPECT_PRED2(UnimplementedOrReturnsCode, status, Code::OK);
  EXPECT_EQ(test_data, result);

  delete[] dims;

  TF_TString* data = static_cast<TF_TString*>(TF_TensorData(t));
  for (int i = 0; i < batch_size; ++i) {
    TF_TString_Init(&data[i]);
    // The following input string length is large enough to make sure that
    // copy to tstring in large mode.
    std::string source =
        "This is the " + std::to_string(i + 1) + "th. data element\n";
    TF_TString_Copy(&data[i], source.c_str(), source.length());

    const string& s = attr->list().s(i);
    values[i] = p;
    lengths[i] = s.size();
    if ((p + s.size()) > (static_cast<char*>(storage) + storage_size)) {
      status->status = InvalidArgument(
          "Not enough storage to hold the requested list of strings");
      return;
    }
    memcpy(values[i], s.data(), s.size());
    p += s.size();
  }
}

TensorFlow was originally developed by researchers and engineers working within
the Machine Intelligence team at Google Brain to conduct research in machine
learning and neural networks. However, the framework is versatile enough to be
used in other areas as well.

TensorFlow provides stable [Python](https://www.tensorflow.org/api_docs/python)
and [C++](https://www.tensorflow.org/api_docs/cc) APIs, as well as a