[
https://issues.apache.org/jira/browse/SPARK-14594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15251507#comment-15251507
]
Marco Gaido commented on SPARK-14594:
-------------------------------------
The code is quite simple, what I can't give you is the data since it's on the
cluster of a client and I can't access it:
df <- sql(sqlContext, "select id, data, time from the_table")
rdd<-SparkR:::toRDD(df)
gb<-SparkR:::groupByKey(rdd, 1000)
... //here you can do any action and it will crash with the above error
If you can generate by yourself some fake data I think it will be fine anyway...
> Improve error messages for RDD API
> ----------------------------------
>
> Key: SPARK-14594
> URL: https://issues.apache.org/jira/browse/SPARK-14594
> Project: Spark
> Issue Type: Improvement
> Components: SparkR
> Affects Versions: 1.5.2
> Reporter: Marco Gaido
>
> When you have an error in your R code using the RDD API, you always get as
> error message:
> Error in if (returnStatus != 0) { : argument is of length zero
> This is not very useful and I think it might be better to catch the R
> exception and show it instead.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
