[GitHub] [arrow] jorgecarleitao edited a comment on pull request #8740: ARROW-10679: [Rust] [DataFusion] Implement CASE WHEN physical expression

[GitBox <gi...@apache.org>]

jorgecarleitao edited a comment on pull request #8740:
URL: https://github.com/apache/arrow/pull/8740#issuecomment-731845713


   Cool!
   
   IMO we want to "take" from different arrays depending on a set of boolean conditions. My recommendation is to use  `MutableDataArray` API that we also use in the Join and that was though through exactly for this use-case.
   
   Something like 
   
   ```rust
   let boolean_conditions: Vec<BooleanArray> = <args>
   let first_non_null: Vec<usize> = first_true(boolean_conditions)  // index from 0 to boolean_conditions.len() (inclusive, last represents the "else")
   
   let capacity = boolean_conditions[0].len();
   let arrays_data = arrays.iter().map(|array| array.data_ref()).collect::<Vec<_>>();
   let mutable = MutableDataArray::new(&arrays_data, capacity);
   
   for row in boolean_conditions[0].len() {
       for index in first_non_null {
           if index == boolean_conditions.len() {
               // the else case
               mutable.extend_null(index); // this function is still not available in master, but I have a branch with it
           } else {
               // take the slice `[row,row+1[` from array "index".
               mutable.extend_from_slice(index, row, row + 1);
           }
       }
   }
   
   return make_array(Arc::new(mutable.freeze()));
   ```
   
   the advantage of this is that it works for all types except struct.
   


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org