TheNeuralBit commented on issue #29365: URL: https://github.com/apache/beam/issues/29365#issuecomment-1848006497
It's really surprising that not many people have run into this! I noticed the [dask bag docs](https://docs.dask.org/en/stable/bag.html#shuffle) do nudge users away from groupby: > These shuffle operations are expensive and better handled by projects like dask.dataframe. It is best to use dask.bag to clean and process data, then transform it into an array or DataFrame before embarking on the more complex operations that require shuffle steps. So I guess it is conceivable that no one is relying on this. It occurred to me that we are also partitioning with a hash in Beam, for the DataFrame API. I refreshed my memory on this and found it's not using `pd.util.hash_array` instead of `hash`: https://github.com/apache/beam/blob/98a26906bfa62708bd796f4de781346e7e019e40/sdks/python/apache_beam/dataframe/partitionings.py#L117 Presumably Dask DataFrames are doing something similar and that's why we don't see this issue there. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
