[SPARK-15942][REPL] Unblock `:reset` command in REPL.

## What changes were proposed in this pull
(Paste from JIRA issue.)
As a follow up for SPARK-15697, I have following semantics for `:reset` command.
On `:reset` we forget all that user has done but not the initialization of spark. To avoid confusion or make it more clear, we show the message `spark` and `sc` are not erased, infact they are in same state as they were left by previous operations done by the user.
While doing above, somewhere I felt that this is not usually what reset means. But an accidental shutdown of a cluster can be very costly, so may be in that sense this is less surprising and still useful.

## How was this patch tested?

Manually, by calling `:reset` command, by both altering the state of SparkContext and creating some local variables.

Author: Prashant Sharma <prashant@apache.org>
Author: Prashant Sharma <prashsh1@in.ibm.com>

Closes #13661 from ScrapCodes/repl-reset-command.

1 files changed, 2 insertions, 1 deletions

diff --git a/repl/scala-2.11/src/test/scala/org/apache/spark/repl/ReplSuite.scala b/repl/scala-2.11/src/test/scala/org/apache/spark/repl/ReplSuite.scala
index 2444e93d9a..c10db947bc 100644
--- a/repl/scala-2.11/src/test/scala/org/apache/spark/repl/ReplSuite.scala
+++ b/repl/scala-2.11/src/test/scala/org/apache/spark/repl/ReplSuite.scala
@@ -49,7 +49,8 @@ class ReplSuite extends SparkFunSuite {
 
     val oldExecutorClasspath = System.getProperty(CONF_EXECUTOR_CLASSPATH)
     System.setProperty(CONF_EXECUTOR_CLASSPATH, classpath)
-
+    Main.sparkContext = null
+    Main.sparkSession = null // causes recreation of SparkContext for each test.
     Main.conf.set("spark.master", master)
     Main.doMain(Array("-classpath", classpath), new SparkILoop(in, new PrintWriter(out)))