Rich-T-kid opened a new pull request, #12306:
URL: https://github.com/apache/arrow-rs/pull/12306

   Closes #11940.
   
   ## What changed
   
   - Override `InProgressDictionaryArray::copy_rows_by_indices` for byte 
dictionaries.
   - Gather sparse keys and validity directly while retaining the dictionary 
values arrays.
   - Keep the existing `take` path for selections above the measured sparse 
crossover.
   - Add one focused test covering sliced nullable keys, repeated/out-of-order 
indices, and values reuse.
   
   This does not add new `BatchCoalescer` state or dispatch plumbing.
   
   ## Local benchmarks
   
   Measured on an Apple M4 Max with isolated base and candidate binaries. For 
shared dictionary values:
   
   - 0.1% selected: 53% faster without nulls and 33-35% faster with nulls.
   - 1% selected: 30-32% faster without nulls and about 15% faster with nulls.
   - 10% and 80% continue through the existing `take` path; paired controls 
showed no repeatable material change.
   
   ## Validation
   
   - `cargo test -p arrow-select --lib` (527 passed)
   - `cargo test -p arrow-select --doc` (20 passed)
   - `cargo clippy -p arrow-select --all-targets --all-features -- -D warnings`
   - `cargo +stable fmt --all -- --check`
   - Repository pre-push formatting, Clippy, and typo checks
   


-- 
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]

Reply via email to