[
https://issues.apache.org/jira/browse/ARROW-2856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16546813#comment-16546813
]
Antoine Pitrou commented on ARROW-2856:
---
Yes, we raise on NaN as well:
{code:python
[
https://issues.apache.org/jira/browse/ARROW-2856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16545473#comment-16545473
]
Wes McKinney commented on ARROW-2856:
-
NumPy _does_ raise with NaN, which aligns with
[
https://issues.apache.org/jira/browse/ARROW-2856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16545428#comment-16545428
]
Wes McKinney commented on ARROW-2856:
-
I'm inclined to agree with [~pitrou] on this.
[
https://issues.apache.org/jira/browse/ARROW-2856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16545383#comment-16545383
]
Antoine Pitrou commented on ARROW-2856:
---
This is consistent with Numpy behaviour. I