vishwanathmuthuraman commented on PR #24402: URL: https://github.com/apache/datafusion/pull/24402#issuecomment-6143073362
Thank you for addressing the fallible-cast case. On e91860399, the 42 CASE tests pass. I also checked the specialized path against an ordinary CASE path using a second column with identical divisor values: mixed nulls, integer overflow, decimal scale, NaN/infinities/signed zero, and empty/all-true/all-false inputs agree in those cases. These are local correctness checks, not performance evidence. Before evaluating the performance tradeoff, could you include reproducible head-versus-base benchmark results across the zero-density cases? The current benchmark constructs a divisor_copy column but does not benchmark the corresponding generic CASE path. A same-input control using that column would also help compare paths on the same checkout. I have not measured a speedup or tested a rebase onto current main. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
