Commits on Aug 14, 2023

1. Makes copy of input ddf to work around dropped column names (#3776) …

   When creating multiple graphs with the same dask_cudf dataframe, there is a metadata mismatch occurring when one or more partitions are empty. In fact, during the second graph creation with the dask_cudf dataframe that was used/modified earlier, the metadata are not conserved for partitions with empty empty dataframes. This is due to the fact a _reference_ to the input dataframe partly destroyed (modfied) during the first graph creation is reused in the second graph creation.
   
   This PR makes a copy of the input dataframe right after the repartition call to avoid that alteration.
   
   Authors:
      - jnke2016 (jnke@gmail.com)
   
   Approvers:
      - Vibhu Jawa (https://github.com/VibhuJawa)
      - Alex Barghi (https://github.com/alexbarghi-nv)
      - Rick Ratzel (https://github.com/rlratzel)

   

   jnke2016 authored Aug 14, 2023
   Configuration menu

   • View commit details
   • Copy full SHA for 20dca85
   • Browse repository at this point

   Copy the full SHA

   20dca85 View commit details

   Browse the repository at this point in the history