Rich-T-kid opened a new issue, #12128: URL: https://github.com/apache/arrow-rs/issues/12128
Scalar predicates on dictionary arrays evaluate once per dictionary value and then use checked `take` to expand the Boolean results through already-validated keys. With DataFusion RLE dictionary reads enabled, this affects queries such as: - ClickBench Q12-Q14: `SearchPhrase <> ''` - ClickBench Q20, Q21, and Q23: `URL LIKE '%google%'` - ClickBench Q22: `Title LIKE '%Google%'`, `URL NOT LIKE '%.google.%'`, and `SearchPhrase <> ''` - ClickBench Q36: `URL <> ''` Avoiding redundant key validation for dictionaries without null keys should reduce the cost of these filters while retaining the checked path for nullable keys. -- 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]
