From 4106558435889261243d186f5f0b51c5f9e98d56 Mon Sep 17 00:00:00 2001
From: Aaron Davidson <aaron@databricks.com>
Date: Sun, 6 Apr 2014 17:48:41 -0700
Subject: SPARK-1314: Use SPARK_HIVE to determine if we include Hive in
 packaging

Previously, we based our decision regarding including datanucleus jars based on the existence of a spark-hive-assembly jar, which was incidentally built whenever "sbt assembly" is run. This means that a typical and previously supported pathway would start using hive jars.

This patch has the following features/bug fixes:

- Use of SPARK_HIVE (default false) to determine if we should include Hive in the assembly jar.
- Analagous feature in Maven with -Phive (previously, there was no support for adding Hive to any of our jars produced by Maven)
- assemble-deps fixed since we no longer use a different ASSEMBLY_DIR
- avoid adding log message in compute-classpath.sh to the classpath :)

Still TODO before mergeable:
- We need to download the datanucleus jars outside of sbt. Perhaps we can have spark-class download them if SPARK_HIVE is set similar to how sbt downloads itself.
- Spark SQL documentation updates.

Author: Aaron Davidson <aaron@databricks.com>

Closes #237 from aarondav/master and squashes the following commits:

5dc4329 [Aaron Davidson] Typo fixes
dd4f298 [Aaron Davidson] Doc update
dd1a365 [Aaron Davidson] Eliminate need for SPARK_HIVE at runtime by d/ling datanucleus from Maven
a9269b5 [Aaron Davidson] [WIP] Use SPARK_HIVE to determine if we include Hive in packaging
---
 bin/compute-classpath.sh | 35 +++++++++++++++++++----------------
 1 file changed, 19 insertions(+), 16 deletions(-)

(limited to 'bin/compute-classpath.sh')

diff --git a/bin/compute-classpath.sh b/bin/compute-classpath.sh
index bef42df71c..be37102dc0 100755
--- a/bin/compute-classpath.sh
+++ b/bin/compute-classpath.sh
@@ -30,21 +30,7 @@ FWDIR="$(cd `dirname $0`/..; pwd)"
 # Build up classpath
 CLASSPATH="$SPARK_CLASSPATH:$FWDIR/conf"
 
-# Support for interacting with Hive.  Since hive pulls in a lot of dependencies that might break
-# existing Spark applications, it is not included in the standard spark assembly.  Instead, we only
-# include it in the classpath if the user has explicitly requested it by running "sbt hive/assembly"
-# Hopefully we will find a way to avoid uber-jars entirely and deploy only the needed packages in
-# the future.
-if [ -f "$FWDIR"/sql/hive/target/scala-$SCALA_VERSION/spark-hive-assembly-*.jar ]; then
-
-  # Datanucleus jars do not work if only included in the uberjar as plugin.xml metadata is lost.
-  DATANUCLEUSJARS=$(JARS=("$FWDIR/lib_managed/jars"/datanucleus-*.jar); IFS=:; echo "${JARS[*]}")
-  CLASSPATH=$CLASSPATH:$DATANUCLEUSJARS
-
-  ASSEMBLY_DIR="$FWDIR/sql/hive/target/scala-$SCALA_VERSION/"
-else
-  ASSEMBLY_DIR="$FWDIR/assembly/target/scala-$SCALA_VERSION/"
-fi
+ASSEMBLY_DIR="$FWDIR/assembly/target/scala-$SCALA_VERSION"
 
 # First check if we have a dependencies jar. If so, include binary classes with the deps jar
 if [ -f "$ASSEMBLY_DIR"/spark-assembly*hadoop*-deps.jar ]; then
@@ -59,7 +45,7 @@ if [ -f "$ASSEMBLY_DIR"/spark-assembly*hadoop*-deps.jar ]; then
   CLASSPATH="$CLASSPATH:$FWDIR/sql/core/target/scala-$SCALA_VERSION/classes"
   CLASSPATH="$CLASSPATH:$FWDIR/sql/hive/target/scala-$SCALA_VERSION/classes"
 
-  DEPS_ASSEMBLY_JAR=`ls "$ASSEMBLY_DIR"/spark*-assembly*hadoop*-deps.jar`
+  DEPS_ASSEMBLY_JAR=`ls "$ASSEMBLY_DIR"/spark-assembly*hadoop*-deps.jar`
   CLASSPATH="$CLASSPATH:$DEPS_ASSEMBLY_JAR"
 else
   # Else use spark-assembly jar from either RELEASE or assembly directory
@@ -71,6 +57,23 @@ else
   CLASSPATH="$CLASSPATH:$ASSEMBLY_JAR"
 fi
 
+# When Hive support is needed, Datanucleus jars must be included on the classpath.
+# Datanucleus jars do not work if only included in the  uber jar as plugin.xml metadata is lost.
+# Both sbt and maven will populate "lib_managed/jars/" with the datanucleus jars when Spark is
+# built with Hive, so first check if the datanucleus jars exist, and then ensure the current Spark
+# assembly is built for Hive, before actually populating the CLASSPATH with the jars.
+# Note that this check order is faster (by up to half a second) in the case where Hive is not used.
+num_datanucleus_jars=$(ls "$FWDIR"/lib_managed/jars/ | grep "datanucleus-.*\\.jar" | wc -l)
+if [ $num_datanucleus_jars -gt 0 ]; then
+  AN_ASSEMBLY_JAR=${ASSEMBLY_JAR:-$DEPS_ASSEMBLY_JAR}
+  num_hive_files=$(jar tvf "$AN_ASSEMBLY_JAR" org/apache/hadoop/hive/ql/exec 2>/dev/null | wc -l)
+  if [ $num_hive_files -gt 0 ]; then
+    echo "Spark assembly has been built with Hive, including Datanucleus jars on classpath" 1>&2
+    DATANUCLEUSJARS=$(echo "$FWDIR/lib_managed/jars"/datanucleus-*.jar | tr " " :)
+    CLASSPATH=$CLASSPATH:$DATANUCLEUSJARS
+  fi
+fi
+
 # Add test classes if we're running from SBT or Maven with SPARK_TESTING set to 1
 if [[ $SPARK_TESTING == 1 ]]; then
   CLASSPATH="$CLASSPATH:$FWDIR/core/target/scala-$SCALA_VERSION/test-classes"
-- 
cgit v1.2.3