fabianmurariu commented on issue #5173: URL: https://github.com/apache/arrow-rs/issues/5173#issuecomment-4682978815
@alamb We have found quite a lot of use cases where NullArray would benefit from type. 1. we need to track global schema separate from the arrow files if we write some of the chunk with null arrays. 2. In our use case the memory savings of using NullArray are quite significant but handling types is extra work 3. We have to write our own recursive null supporting concat 4. Schema merge doesn't work with null datatype. 5. We lost data_type on null arrays after migrating to arrow-rs from [polars-arrow](https://docs.rs/polars-arrow/latest/polars_arrow/array/struct.NullArray.html) I think if NullArray would function as an identity element if some of the some operations like, concat, or schema merge I think that would be quite useful even without it having a designated data_type. -- 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]
