does anyone know how to cache or keep in memory an io manager handled output?

class IOBigQToDF(IOManager):

	def handle_output(self, context: OutputContext, obj: str) -> None:
		df: pl.DataFrame = context.resources.rsrc_bigquery_runner(obj).bigquery_to_df()
		filter_func = context.config["filter_func"]
		if filter_func is None:
			get_dagster_logger().info(
				f'DATAFRAME WAS NOT ADJUSTED, PER OP: {str(df.shape)} columns: {str(df.shape[1])}, rows: {str(df.shape[0])}')
			self.result = df
		else:
			filtered_df = filter_func(df)
			get_dagster_logger().info(
				f'ADJUSTED DATAFRAME SHAPE: {str(filtered_df.shape)} columns: {str(filtered_df.shape[1])}, rows: {str(filtered_df.shape[0])}')
			self.result = filtered_df

	def load_input(self, context: InputContext) -> pl.DataFrame:
		return self.result

i thought that i could keep it in the

self.result

attribute but i get an error saying

AttributeError: 'IOBigQToDF' object has no attribute 'result'

when i try to use the io manager as an input. when it's used as an output handler, the logger prints out the line above, and that's how i know it works. but i guess after that, the instance is no longer used?? this doesn't work:

@op#(ins={"bq_to_df_io": In(input_manager_key="io_bigq_to_df")})
def some_op(bq_to_df_io):
   get_dagster_logger().info(bq_to_df_io)