[ https://issues.apache.org/jira/browse/SPARK-16510?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15376010#comment-15376010 ]
Shivaram Venkataraman commented on SPARK-16510: ----------------------------------------------- For (a) the assumption is that we will be building the JAR during the Spark build process using Maven. It need not be run during the R package installation process and can be just used as a reference mechanism for generating a JAR. Also since we dont expect the file to change, we can upload the JAR built as a part of a release to some known location ? > Move SparkR test JAR into Spark, include its source code > -------------------------------------------------------- > > Key: SPARK-16510 > URL: https://issues.apache.org/jira/browse/SPARK-16510 > Project: Spark > Issue Type: Sub-task > Components: SparkR > Reporter: Shivaram Venkataraman > > One of the `NOTE`s in the R CMD check is that we currently include a test JAR > file in SparkR which is a binary only artifact. I think we can take two steps > to address this > (a) I think we should include the source code for this in say core/src/test/ > or something like that. As far as I know the JAR file just needs to have a > single method. > (b) We should move the JAR file out of the SparkR test support and into some > other location in Spark. The trouble is that its tricky to run the test with > CRAN mode then. We could either disable the test for CRAN or download the JAR > from an external URL -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org