[
https://issues.apache.org/jira/browse/SPARK-15159?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15283321#comment-15283321
]
Felix Cheung edited comment on SPARK-15159 at 5/14/16 12:01 AM:
----------------------------------------------------------------
Right, possibly with SQLContext/HiveContext as wrapper of SparkSession.
If we are to implement SparkSession in R, its builder syntax might not be
natural to replicate. I think we should implement only a small subset of
functions and build from there.
And perhaps to implement SparkSession as a S4 class for better type checking
enforcement - we might also need to do that for function-signature backward
compatibility with functions that takes sqlContext.
It sounds like a lot of work.
was (Author: felixcheung):
Right, possibly with SQLContext/HiveContext as wrapper of SparkSession.
If we are to implement SparkSession in R, its builder syntax might not be
natural to replicate. I think we should implement only a small subset of
functions and build from there.
> SparkSession R API
> ------------------
>
> Key: SPARK-15159
> URL: https://issues.apache.org/jira/browse/SPARK-15159
> Project: Spark
> Issue Type: Sub-task
> Components: SparkR
> Affects Versions: 1.6.1
> Reporter: Sun Rui
>
> HiveContext is to be deprecated in 2.0. Replace them with
> SparkSession.builder.enableHiveSupport in SparkR
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
