[ https://issues.apache.org/jira/browse/KUDU-1676?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16452632#comment-16452632 ]
Grant Henke commented on KUDU-1676: ----------------------------------- Thanks for the comment [~ivan.vergiliev]. I didn't realize there was a Jira for this. This is fixed via [e94556e|https://github.com/apache/kudu/commit/e94556e]. The unit test added in [DefaultSourceTest.scala|https://github.com/apache/kudu/commit/e94556e#diff-e35eedde7d54f5aac66eb41b13e7efee] is a brief example. > Spark DDL needs elegant way to specify range partitioning > --------------------------------------------------------- > > Key: KUDU-1676 > URL: https://issues.apache.org/jira/browse/KUDU-1676 > Project: Kudu > Issue Type: New Feature > Components: spark > Affects Versions: 1.0.0 > Reporter: Mladen Kovacevic > Assignee: Mladen Kovacevic > Priority: Major > Original Estimate: 96h > Remaining Estimate: 96h > > To define partition column splits, you need a PartialRow object. > These are easy to create when you have the Schema object. But since your > table schema in Spark is defined with StructType instead of Schema, then its > cumbersome to define a new Schema object to be the exact duplicate of the > StructType version, only to get the PartialRow, to set what the range > partition will values would be, then set the addSplitRow() function call to > your CreateTableOptions. > We need an elegant way to have the Spark API handle specifying range > partition attributes without having to drop into the Java API in Spark. -- This message was sent by Atlassian JIRA (v7.6.3#76005)