neilconway commented on PR #22516: URL: https://github.com/apache/datafusion/pull/22516#issuecomment-4555336966
> Would we see similar issues for other math functions, meaning if we follow this approach we'd have to change math functions to operate predominantly on f64? We wouldn't. The main thing we need to avoid is using f32 for integer inputs. We can either do that by just removing f32 support, or ensuring that we use the f64 path for integer inputs. This PR initially did the former, but on closer examination, I see that quite a few DataFusion numerical functions have a float32 code path. So it is probably more consistent to switch to the second approach, which I've now done. -- 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]
