Would it perhaps make sense to define the total order for non-numbers (NaN,
Inf, -Inf) globally (i.e., in the spec or in Arrow itself) so that the
behavior is the same across all languages?
On Fri, Aug 28, 2020 at 7:42 PM Andy Grove wrote:
> Hi Jörn,
>
> I agree with your concerns about NaN.
Hi Jörn,
I agree with your concerns about NaN. There was a discussion about this in
https://github.com/apache/arrow/pull/7193
I will try and make time this weekend to look at the current implementation
and your suggestions around DictionaryArray.
Hopefully, other contributors that are more
I ran into a few issues with the rust sort kernels and would like to
discuss possible solutions.
1. When sorting by multiple columns (lexsort_to_indices) the Float32
and Float64 data types are not supported because the implementation
relies on the OrdArray trait. This trait is not implemented
