Rich-T-kid opened a new issue, #11940: URL: https://github.com/apache/arrow-rs/issues/11940
### Is your feature request related to a problem or challenge? Follow-up to #11743 and #11174. #11743 adds a specialized in-progress coalescer for byte dictionaries, but the indexed-copy path introduced by #11174 will still use the default `InProgressArray::copy_rows_by_indices` implementation. That implementation calls `take` to materialize an intermediate dictionary before copying it into the coalescer. ### Describe the solution you'd like Add `copy_rows_by_indices` support to `InProgressDictionaryArray` for `Dictionary<K, Utf8 | LargeUtf8 | Binary | LargeBinary>`. For a pointer-identical values array, gather the selected primitive keys and validity directly into the in-progress buffers while reusing the values `Arc`. This should avoid the intermediate `take` allocation while preserving repeated and out-of-order indices, null keys, bounds validation, and target-batch chunking. Inputs with a different values mapping must continue to use a correct fallback. Use the dictionary `take` cases in `coalesce_kernels` to compare the specialized path with the existing materialized path for both shared and changing values arrays. -- 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]
