SPARK-6433 hive tests to import spark-sql test JAR for QueryTest access

1. Test JARs are built & published
1. log4j.resources is explicitly excluded. Without this, downstream test run logging depends on the order the JARs are listed/loaded
1. sql/hive pulls in spark-sql &...spark-catalyst for its test runs
1. The copied in test classes were rm'd, and a test edited to remove its now duplicate assert method
1. Spark streaming is now build with the same plugin/phase as the rest, but its shade plugin declaration is kept in (so different from the rest of the test plugins). Due to (#2), this means the test JAR no longer includes its log4j file.

Outstanding issues:
* should the JARs be shaded? `spark-streaming-test.jar` does, but given these are test jars for developers only, especially in the same spark source tree, it's hard to justify.
* `maven-jar-plugin` v 2.6 was explicitly selected; without this the apache-1.4 parent template JAR version (2.4) chosen.
* Are there any other resources to exclude?

Author: Steve Loughran <stevel@hortonworks.com>

Closes apache#5119 from steveloughran/stevel/patches/SPARK-6433-test-jars and squashes the following commits:

81ceb01 [Steve Loughran] SPARK-6433 add a clearer comment explaining what the plugin is doing & why
a6dca33 [Steve Loughran] SPARK-6433 : pull configuration section form archive plugin
c2b5f89 [Steve Loughran] SPARK-6433 omit "jar" goal from jar plugin
fdac51b [Steve Loughran] SPARK-6433 -002; indentation & delegate plugin version to parent
650f442 [Steve Loughran] SPARK-6433 patch 001: test JARs are built; sql/hive pulls in spark-sql & spark-catalyst for its test runs

Author: steveloughran

pom.xml

@@ -1265,6 +1265,7 @@
               <id>create-source-jar</id>
               <goals>
                 <goal>jar-no-fork</goal>
+                <goal>test-jar-no-fork</goal>
               </goals>
             </execution>
           </executions>
@@ -1473,6 +1474,25 @@
         <groupId>org.scalatest</groupId>
         <artifactId>scalatest-maven-plugin</artifactId>
       </plugin>
+      <!-- Build test-jar's for all projects, since some projects depend on tests from others -->
+      <plugin>
+        <groupId>org.apache.maven.plugins</groupId>
+        <artifactId>maven-jar-plugin</artifactId>
+        <executions>
+          <execution>
+            <id>prepare-test-jar</id>
+            <phase>prepare-package</phase>
+            <goals>
+              <goal>test-jar</goal>
+            </goals>
+            <configuration>
+              <excludes>
+                <exclude>log4j.properties</exclude>
+              </excludes>
+            </configuration>
+          </execution>
+        </executions>
+      </plugin>
     </plugins>
   </build>
 

sql/hive/pom.xml

@@ -89,6 +89,20 @@
       <artifactId>junit</artifactId>
       <scope>test</scope>
     </dependency>
+    <dependency>
+      <groupId>org.apache.spark</groupId>
+      <artifactId>spark-sql_${scala.binary.version}</artifactId>
+      <type>test-jar</type>
+      <version>${project.version}</version>
+      <scope>test</scope>
+    </dependency>
+    <dependency>
+      <groupId>org.apache.spark</groupId>
+      <artifactId>spark-catalyst_${scala.binary.version}</artifactId>
+      <type>test-jar</type>
+      <version>${project.version}</version>
+      <scope>test</scope>
+    </dependency>
   </dependencies>
   <profiles>
     <profile>

sql/hive/src/test/scala/org/apache/spark/sql/QueryTest.scala

@@ -24,21 +24,6 @@ import org.apache.spark.sql.{AnalysisException, DataFrame, QueryTest}
 import org.apache.spark.storage.RDDBlockId
 
 class CachedTableSuite extends QueryTest {
-  /**
-   * Throws a test failed exception when the number of cached tables differs from the expected
-   * number.
-   */
-  def assertCached(query: DataFrame, numCachedTables: Int = 1): Unit = {
-    val planWithCaching = query.queryExecution.withCachedData
-    val cachedData = planWithCaching collect {
-      case cached: InMemoryRelation => cached
-    }
-
-    assert(
-      cachedData.size == numCachedTables,
-      s"Expected query to contain $numCachedTables, but it actually had ${cachedData.size}\n" +
-        planWithCaching)
-  }
 
   def rddIdOf(tableName: String): Int = {
     val executedPlan = table(tableName).queryExecution.executedPlan

sql/hive/src/test/scala/org/apache/spark/sql/catalyst/plans/PlanTest.scala

@@ -97,34 +97,6 @@
     <outputDirectory>target/scala-${scala.binary.version}/classes</outputDirectory>
     <testOutputDirectory>target/scala-${scala.binary.version}/test-classes</testOutputDirectory>
     <plugins>
-      <!--
-           This plugin forces the generation of jar containing streaming test classes,
-           so that the tests classes of external modules can use them. The two execution profiles
-           are necessary - first one for 'mvn package', second one for 'mvn test-compile'. Ideally,
-           'mvn compile' should not compile test classes and therefore should not need this.
-           However, an open Maven bug (http://jira.codehaus.org/browse/MNG-3559)
-           causes the compilation to fail if streaming test-jar is not generated. Hence, the
-           second execution profile for 'mvn test-compile'.
-      -->
-      <plugin>
-        <groupId>org.apache.maven.plugins</groupId>
-        <artifactId>maven-jar-plugin</artifactId>
-        <executions>
-          <execution>
-            <goals>
-              <goal>test-jar</goal>
-            </goals>
-          </execution>
-          <execution>
-            <id>test-jar-on-test-compile</id>
-            <phase>test-compile</phase>
-            <goals>
-              <goal>test-jar</goal>
-            </goals>
-          </execution>
-        </executions>
-      </plugin>
-
       <plugin>
         <groupId>org.apache.maven.plugins</groupId>
         <artifactId>maven-shade-plugin</artifactId>