Ted-Jiang commented on issue #3708: URL: https://github.com/apache/arrow-rs/issues/3708#issuecomment-1429017073
> Could you provide a link to the benchmark you were running, when this was previously proposed the numbers were not nearly so flattering - [#3134 (comment)](https://github.com/apache/arrow-rs/pull/3134#issuecomment-1328988681) > > I'm mainly apprehensive as we already have issues with the amount of codegen for the arithmetic kernels 😅 I used `unary_mut ` in datafusion like ``` rust pub(crate) fn add_scalar_cow<T>( array: PrimitiveArray<T>, scalar: T::Native, ) -> Result<PrimitiveArray<T>, ArrowError> where T: ArrowNumericType, T::Native: ArrowNativeTypeOp, { match unary_mut(array, |value| value.add_wrapping(scalar)) { Ok(array) => Ok(array), // Fall back need copy array Err(array) => add_scalar(&array, scalar), } } ``` But there a lot code change during that time, like support dictionaryArray.... So i try to modify less api in datafusion, support `unary_dyn_mut ` -- 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]
