[
https://issues.apache.org/jira/browse/ARROW-12861?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17350633#comment-17350633
]
Eduardo Ponce edited comment on ARROW-12861 at 5/24/21, 11:47 PM:
------------------------------------------------------------------
IEEE 754 defines rules for +0.0 and -0.0, thus *sign(+0.0) = 1* and *sign(-0.0)
= -1*. On the other hand, integral zero results in *sign(0) = 0*. Is this
behavior reasonable?
This approach will result in different behavior for floating-point and integral
operands.
# Floating-point *sign(X) =*
*-1 if X <= -0.0*
*1 if X >= 0.0*
# Integral *sign(X) =*
*-1 if X < 0*
*0 if X = 0*
*1 if X > 0*
was (Author: edponce):
IEEE 754 defines rules for +0.0 and -0.0, thus *sign(+0.0) = 1* and *sign(-0.0)
= -1*. On the other hand, integral zero results in *sign(0) = 0*. Is this
behavior reasonable?
> [C++][Compute] Add sign function kernels
> ----------------------------------------
>
> Key: ARROW-12861
> URL: https://issues.apache.org/jira/browse/ARROW-12861
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: Eduardo Ponce
> Assignee: Eduardo Ponce
> Priority: Major
> Fix For: 5.0.0
>
>
> Add the sign function to the compute kernels.
> sign(X) =
> * 1 if X > 0
> * 0 if X = 0
> * -1 if X < 0
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
