cloud-fan commented on issue #25347: [SPARK-28610][SQL] Allow having a decimal buffer for long sum URL: https://github.com/apache/spark/pull/25347#issuecomment-525763443 Now I'm wondering if it's worthwhile to sacrifice 10x perf to handle intermedia overflow. It's not about correctness but usability: without this change, queries are more likely to fail with overflow when summing up long values. So it's a tradeoff between usability and performance. Given it's 10x slowdown, I don't think any users would turn it on by default. If they hit overflow exception later, I think they will just cast the data to decimal before sum. That said, I think it's not very useful to have this feature. Anyway, many thanks to @mgaido91 for investigating and providing the perf numbers!
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
