if (h == nullptr) {
    status->status = tensorflow::errors::InvalidArgument("Invalid handle");
    return -1;
  }

  int64_t dim = -1;
  status->status = tensorflow::unwrap(h)->Dim(dim_index, &dim);
  return dim;
}

const char* TFE_TensorHandleDeviceName(TFE_TensorHandle* h, TF_Status* status) {
  if (h == nullptr) {

  ASSERT_EQ(num_dims, -1);
  ASSERT_EQ("Invalid handle", string(TF_Message(status.get())));

  TF_SetStatus(status.get(), TF_OK, "");

  int dim = TFE_TensorHandleDim(h, 0, status.get());
  ASSERT_EQ(TF_INVALID_ARGUMENT, TF_GetCode(status.get()));
  ASSERT_EQ(dim, -1);
  ASSERT_EQ("Invalid handle", string(TF_Message(status.get())));
}

TEST(CAPI, TensorHandleDevices) {

#undef EXPECT_TF_META

TEST(CAPI, TestTensorAligned) {
  int64_t dim = 7;
  size_t tensor_size_bytes = dim * TF_DataTypeSize(TF_FLOAT);
  TF_Tensor* a = TF_AllocateTensor(
      /*dtype=*/TF_FLOAT, /*dims=*/&dim, /*num_dims=*/1,
      /*len=*/tensor_size_bytes);
  float* data = reinterpret_cast<float*>(TF_TensorData(a));
  for (int i = 0; i < dim; ++i) {
    data[i] = 0;
  }
  if (TF_TensorDefaultAlignment() > 0) {

  auto dtype = TFE_TensorHandleDataType(t->tensor);
  TFE_CustomDeviceTensorHandleMethods handle_methods;
  handle_methods.num_dims = &LoggedTensorNumDims;
  handle_methods.dim = &LoggedTensorDim;
  handle_methods.deallocator = &LoggedTensorDeallocator;
  return TFE_NewCustomDeviceTensorHandle(context, logging_device_name.c_str(),
                                         dtype, t.release(), handle_methods,

* `tf.argmax`: `dimension` becomes `axis` * `tf.argmin`: `dimension` becomes
`axis` * `tf.count_nonzero`: `reduction_indices` becomes `axis` *
`tf.expand_dims`: `dim` becomes `axis` * `tf.reduce_all`: `reduction_indices`
becomes `axis` * `tf.reduce_any`: `reduction_indices` becomes `axis` *
`tf.reduce_join`: `reduction_indices` becomes `axis` * `tf.reduce_logsumexp`: