address comments

Author: gengliangwang

sql/hive/src/test/scala/org/apache/spark/sql/sources/HadoopFsRelationTest.scala

@@ -118,21 +118,21 @@ abstract class HadoopFsRelationTest extends QueryTest with SQLTestUtils with Tes
     new UDT.MyDenseVectorUDT()
   ).filter(supportsDataType)
 
-  private val parquetDictionaryEncodingEnabledConfs = if (isParquetDataSource) {
-    // Run with/without Parquet dictionary encoding enabled for Parquet data source.
-    Seq(true, false)
-  } else {
-    Seq(false)
-  }
-
-  for (dataType <- supportedDataTypes) {
-    for (parquetDictionaryEncodingEnabled <- parquetDictionaryEncodingEnabledConfs) {
-      val extraMessage = if (isParquetDataSource) {
-        s" with parquet.enable.dictionary = $parquetDictionaryEncodingEnabled"
-      } else {
-        ""
-      }
-      test(s"test all data types - $dataType$extraMessage") {
+  test(s"test all data types") {
+    val parquetDictionaryEncodingEnabledConfs = if (isParquetDataSource) {
+      // Run with/without Parquet dictionary encoding enabled for Parquet data source.
+      Seq(true, false)
+    } else {
+      Seq(false)
+    }
+    for (dataType <- supportedDataTypes) {
+      for (parquetDictionaryEncodingEnabled <- parquetDictionaryEncodingEnabledConfs) {
+        val extraMessage = if (isParquetDataSource) {
+          s" with parquet.enable.dictionary = $parquetDictionaryEncodingEnabled"
+        } else {
+          ""
+        }
+        logInfo(s"Testing $dataType data type$extraMessage")
 
         val extraOptions = Map[String, String](
           "parquet.enable.dictionary" -> parquetDictionaryEncodingEnabled.toString
@@ -769,37 +769,6 @@ abstract class HadoopFsRelationTest extends QueryTest with SQLTestUtils with Tes
     }
   }
 
-  // NOTE: This test suite is not super deterministic.  On nodes with only relatively few cores
-  // (4 or even 1), it's hard to reproduce the data loss issue.  But on nodes with for example 8 or
-  // more cores, the issue can be reproduced steadily.  Fortunately our Jenkins builder meets this
-  // requirement.  We probably want to move this test case to spark-integration-tests or spark-perf
-  // later.
-  test("SPARK-8406: Avoids name collision while writing files") {
-    // The following test is slow. As now all the file format data source are using common code
-    // for creating result files, we can test one data source(Parquet) only to reduce test time.
-    if (isParquetDataSource) {
-      withTempPath { dir =>
-        val path = dir.getCanonicalPath
-        spark
-          .range(10000)
-          .repartition(250)
-          .write
-          .mode(SaveMode.Overwrite)
-          .format(dataSourceName)
-          .save(path)
-
-        assertResult(10000) {
-          spark
-            .read
-            .format(dataSourceName)
-            .option("dataSchema", StructType(StructField("id", LongType) :: Nil).json)
-            .load(path)
-            .count()
-        }
-      }
-    }
-  }
-
   test("SPARK-8887: Explicitly define which data types can be used as dynamic partition columns") {
     val df = Seq(
       (1, "v1", Array(1, 2, 3), Map("k1" -> "v1"), Tuple2(1, "4")),

sql/hive/src/test/scala/org/apache/spark/sql/sources/ParquetHadoopFsRelationSuite.scala

@@ -232,4 +232,33 @@ class ParquetHadoopFsRelationSuite extends HadoopFsRelationTest {
       }
     }
   }
+
+  // NOTE: This test suite is not super deterministic.  On nodes with only relatively few cores
+  // (4 or even 1), it's hard to reproduce the data loss issue.  But on nodes with for example 8 or
+  // more cores, the issue can be reproduced steadily.  Fortunately our Jenkins builder meets this
+  // requirement.  We probably want to move this test case to spark-integration-tests or spark-perf
+  // later.
+  // Also, this test is slow. As now all the file format data source are using common code
+  // for creating result files, we can test Parquet only to reduce test time.
+  test("SPARK-8406: Avoids name collision while writing files") {
+    withTempPath { dir =>
+      val path = dir.getCanonicalPath
+      spark
+        .range(10000)
+        .repartition(250)
+        .write
+        .mode(SaveMode.Overwrite)
+        .format(dataSourceName)
+        .save(path)
+
+      assertResult(10000) {
+        spark
+          .read
+          .format(dataSourceName)
+          .option("dataSchema", StructType(StructField("id", LongType) :: Nil).json)
+          .load(path)
+          .count()
+      }
+    }
+  }
 }