[jira] [Commented] (ARROW-12959) [C++][R] Option for is_null(NaN) to evaluate to true

["Joris Van den Bossche (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ARROW-12959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17357226#comment-17357226 ] 

Joris Van den Bossche commented on ARROW-12959:
-----------------------------------------------

Also for Python land (where NaN is often used as missing value indicator), this will probably be useful. So agreed that doing this on the C++ side with an option would be good.

> [C++][R] Option for is_null(NaN) to evaluate to true
> ----------------------------------------------------
>
>                 Key: ARROW-12959
>                 URL: https://issues.apache.org/jira/browse/ARROW-12959
>             Project: Apache Arrow
>          Issue Type: Improvement
>          Components: C++, R
>            Reporter: Ian Cook
>            Priority: Major
>
> (This is the flip side of ARROW-12960.)
> Currently the Arrow compute kernel {{is_null}} always treats {{NaN}} as a non-missing value, returning {{false}} at positions of the input datum with value {{NaN}}.
> It would be helpful to be able to control this behavior with an option. The option could be named {{nan_is_null}} or something similar.  It would default to {{false}}, consistent with current behavior. When set to {{true}}, it should check if the input datum has a floating point data type, and if so, return {{true}} at positions where the input is {{NaN}}. If the input datum has some other type, the option should be silently ignored.
> Among other things, this would enable the {{arrow}} R package to evaluate {{is.na()}} consistently with the way base R does. In base R, {{is.na()}} returns {{TRUE}} on {{NaN}}. But in the {{arrow}} R package, it returns {{FALSE}}:
> {code:r}
> is.na(c(3.14, NA, NaN))
> ## [1] FALSE TRUE TRUE
> as.vector(is.na(Array$create(c(3.14, NA, NaN))))
> ## [1] FALSE TRUE FALSE{code}
> I think solving this with an option in the C++ kernel is the best solution, because I suspect there are other cases in which users might want to treat {{NaN}} as a missing value. However, it would also be possible to solve this just in the R package, by defining a mapping of {{is.na}} in the R package that checks if the input {{x}} has a floating point data type, and if so, evaluates {{is.na\(x\) | is.nan\(x\)}}. If we choose to go that route, we should change this Jira issue summary to "[R] Make is.na(NaN) consistent with base R".



--
This message was sent by Atlassian Jira
(v8.3.4#803005)