mzabaluev opened a new pull request, #11999: URL: https://github.com/apache/arrow-rs/pull/11999
# Which issue does this PR close? - Closes #11983 # Rationale for this change `FilterPredicate` kept the filter mask alongside every iteration strategy, including materialized slices or indices and the all / none strategies, so code building a predicate had to supply both representations. # What changes are included in this PR? The mask now lives in the lazy `SlicesIterator` / `IndexIterator` strategies, and the materialized `Slices` / `Indices` keep it only when `filter_bits` compresses with it. Kernels needing the mask words otherwise synthesize them lazily from the slices or indices. `filter_run_end_array` counts the selected rows of each run with popcount over the mask words rather than bit by bit, and the `MutableArrayData` fallback merges materialized indices into runs. # Are these changes tested? Added tests exercising state in the built `FilterPredicate` cases and verifying that mask application works as before. # Are there any significant user-facing changes? No public API changes, this is an internal optimization. -- 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]
