[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17080599#comment-17080599
]
Wes McKinney commented on ARROW-2714:
-
Seems like we can implement this using {{Array.take}}
>
[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16519174#comment-16519174
]
Wes McKinney commented on ARROW-2714:
-
The Tensor classes are independent from the columnar data
[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16519136#comment-16519136
]
Antoine Pitrou commented on ARROW-2714:
---
Yes... but then our arrays would do zero-copy slices in
[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16519130#comment-16519130
]
Wes McKinney commented on ARROW-2714:
-
Only if they're using NumPy, though. With Python lists it's a
[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16519126#comment-16519126
]
Antoine Pitrou commented on ARROW-2714:
---
Hmm, then I think it can be counter-intuitive for users
[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16519123#comment-16519123
]
Wes McKinney commented on ARROW-2714:
-
The only option would be to produce a copy. That makes this a
[
https://issues.apache.org/jira/browse/ARROW-2714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16518263#comment-16518263
]
Antoine Pitrou commented on ARROW-2714:
---
I'm not sure how that would work. Do you expect a copy? A
