It seems the arrow-dataset api already has the async IO layer. However, I want to use the low-level Parquet api with async IO. That is, the decoded values are consumed by some user-defined function, not converted to arrow table. Something similar to ScanFileContents: https://github.com/apache/arrow/blob/master/cpp/src/parquet/file_reader.cc#L818
The current async io interface inside ParquetFileReader seems to be served for arrow dataset api. I was wondering if there is any code snippet to implement the async version of ScanFileContents? If there is no, one way for me to approach this is to try to use ParquetFileReader::PreBuffer and ParquetFileReader::WhenBuffered and refer to dataset api implementation.
