The author would have a point in case of a bug, but this is actually a feature request. Open source or not, customers are not entitled to demand new features.

The author would have a point in case of a bug, but this is actually a feature request. Open source or not, customers are not entitled to demand new features. > But last month, I filed a Beam bug report for an issue in Beam’s BigQuery integration (which, as far as I can tell, is officially maintained by Google). The gist of it is that when you’re using the native Python Beam implementation, you can’t upload data to BigQuery in large batches — you can only stream it, which is significantly slower than batch uploading. While it’s still mostly usable (streaming the data into BigQuery instead of uploading it in one big batch works well enough), the issue makes uploading some large datasets prohibitively slow.