[
https://issues.apache.org/jira/browse/SPARK-22436?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16238712#comment-16238712
]
Eric Maynard commented on SPARK-22436:
--
[~asmaier] Wouldn't the right way to implement this be to
[
https://issues.apache.org/jira/browse/SPARK-22436?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16237530#comment-16237530
]
Sean Owen commented on SPARK-22436:
---
Yes, but that's true of any Python UDF, and not every one can be
[
https://issues.apache.org/jira/browse/SPARK-22436?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16237446#comment-16237446
]
Andreas Maier commented on SPARK-22436:
---
Python UDFs are very slow, aren't they? I believe a Spark
[
https://issues.apache.org/jira/browse/SPARK-22436?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16237420#comment-16237420
]
Sean Owen commented on SPARK-22436:
---
Why does it need to be in Spark as opposed to a simple UDF?
> New