Jefffrey commented on PR #18999: URL: https://github.com/apache/datafusion/pull/18999#issuecomment-3618902161
In the original PR that kicked off this effort (#17023) it converts the decimal128 to the native i128 representation before doing an integer log, as converting to f64 apparently causes some precision loss. I think we should follow suit as otherwise there is little difference than previous behaviour of casting to float64 first before doing the log 🤔 It also makes me wonder if we should handle negative scale by just casting to float to do the log so we don't lose functionality. Thanks for looking into the other solutions from IBM and Intel; I think we can avoid porting/copying their code unless there is a strong need for what they bring to the table for us. -- 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]
