return {};
  }
  const auto& fdef_lib = gdef.library();
  if (fdef_lib.gradient_size() > 0) {
    status->status = tensorflow::errors::Internal(
        "GradientDef is not supported in reading Dataset related functions: ",
        text_proto);
    return {};
  }
  std::vector<UniqueFuncPtr> ret;
  for (const FunctionDef& fdef : fdef_lib.function()) {
    // Make a copy so that we can mutate it.

// Performs specific fusion for GPU targets.
std::unique_ptr<OperationPass<func::FuncOp>> CreateGpuOpFusionPass();

// Creates a pass that decomposes to be compiled ReduceDataset ops into a while
// loop that iterates the dataset and calls the reduction function.
std::unique_ptr<OperationPass<func::FuncOp>> CreateDecomposeReduceDatasetPass();

// Create a pass that convert ops that copy tensors between devices, e.g.
// tf.Identity.

 *  Fix: Recover gracefully when a cache entry's certificate is corrupted.

 *  Fix: Fail permanently when there's a failure loading the bundled public suffix database.
    This is the dataset that powers `HttpUrl.topPrivateDomain()`.

 *  Fix: Immediately update the connection's flow control window instead of waiting for the
    receiving stream to process it.

    RankedTensorType index_ty =
        mlir::dyn_cast<RankedTensorType>(index.getType());
    RankedTensorType data_ty = mlir::dyn_cast<RankedTensorType>(data.getType());
    if (!index_ty || !data_ty) continue;

    int64_t index_rank = index_ty.getRank();
    ArrayRef<int64_t> data_shape = data_ty.getShape();
    ArrayRef<int64_t> index_shape = index_ty.getShape();
    if (failed(mlir::verifyCompatibleShape(index_shape,

  }
  return false;
}

// Helper structure to capture shapes & types for Dataset input.
struct DatasetInput {
  explicit operator bool() const { return shapes && types; }

  ArrayAttr shapes;
  ArrayAttr types;
};

// Returns the input elements shapes and types for Dataset ops.
DatasetInput GetDatasetInput(Value value) {

      }
    ```
  }];
}

def DecomposeReduceDatasetPass : Pass<"tf-decompose-reduce-dataset", "mlir::func::FuncOp"> {
  let summary = "Decomposes ReduceDataset op into dataset operations.";

  let description = [{
    Decomposes ReduceDataset op into a while loop that iterates the dataset and calls
    into the reduction function.  This decomposition is only done if the

       ]>,
       SingleBlockImplicitTerminator<"YieldOp">,
       TF_GeneratorOpSideEffect,
      ]> {
  let summary = "Regional version of GeneratorDataset";

  let description = [{
Creates a dataset that invokes its 'next' region to generate elements. Conceptually,
within MLIR, we treat this op as if it fills a buffer with all the results right away,
and those results are then passed (through the variant tensor result) to

  // pass to merge these together into the earliest copy.
  llvm::SetVector<int64_t> forward_variable_inputs;

  // Validate that the only resource inputs that are read by ops in
  // forward_pass_ops are dataset and variable ops.
  int64_t resource_count = 0;
  for (auto argument : loop_body_func.getArguments()) {
    // Check that all resource arguments are either fed to iterator get next

      "properties": {
        "datasetName": {
          "description": "Name of the dataset stored as metadata -> name on the dataset for Flocker should be considered as deprecated",
          "type": "string"
        },
        "datasetUUID": {
          "description": "UUID of the dataset. This is unique identifier of a Flocker dataset",
          "type": "string"
        }
      },
      "type": "object"
    },

message FlockerVolumeSource {
  // datasetName is Name of the dataset stored as metadata -> name on the dataset for Flocker
  // should be considered as deprecated
  // +optional
  optional string datasetName = 1;

  // datasetUUID is the UUID of the dataset. This is unique identifier of a Flocker dataset
  // +optional
  optional string datasetUUID = 2;
}