[SPARK-10422] [SQL] String column in InMemoryColumnarCache needs to override clone method

https://issues.apache.org/jira/browse/SPARK-10422

Author: Yin Huai <yhuai@databricks.com>

Closes #8578 from yhuai/SPARK-10422.

sql/core/src/main/scala/org/apache/spark/sql/columnar/ColumnType.scala

@@ -339,6 +339,8 @@ private[sql] object STRING extends NativeColumnType(StringType, 7, 8) {
   override def copyField(from: InternalRow, fromOrdinal: Int, to: MutableRow, toOrdinal: Int) {
     setField(to, toOrdinal, getField(from, fromOrdinal))
   }
+
+  override def clone(v: UTF8String): UTF8String = v.clone()
 }
 
 private[sql] object DATE extends NativeColumnType(DateType, 8, 4) {

sql/core/src/test/scala/org/apache/spark/sql/columnar/InMemoryColumnarQuerySuite.scala

@@ -191,4 +191,24 @@ class InMemoryColumnarQuerySuite extends QueryTest with SharedSQLContext {
       ctx.table("InMemoryCache_different_data_types").collect())
     ctx.dropTempTable("InMemoryCache_different_data_types")
   }
+
+  test("SPARK-10422: String column in InMemoryColumnarCache needs to override clone method") {
+    val df =
+      ctx.range(1, 100).selectExpr("id % 10 as id").rdd.map(id => Tuple1(s"str_$id")).toDF("i")
+    val cached = df.cache()
+    // count triggers the caching action. It should not throw.
+    cached.count()
+
+    // Make sure, the DataFrame is indeed cached.
+    assert(sqlContext.cacheManager.lookupCachedData(cached).nonEmpty)
+
+    // Check result.
+    checkAnswer(
+      cached,
+      ctx.range(1, 100).selectExpr("id % 10 as id").rdd.map(id => Tuple1(s"str_$id")).toDF("i")
+    )
+
+    // Drop the cache.
+    cached.unpersist()
+  }
 }