HADOOP-8918. test-patch.sh is parsing modified files wrong. Contributed by Raja Aluri.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1397411 13f79535-47bb-0310-9956-ffa450edef68

dev-support/test-patch.sh

@@ -763,7 +763,7 @@ findModule (){
 findModules () {
   # Come up with a list of changed files into $TMP
   TMP=/tmp/tmp.paths.$$
-  $GREP '^+++\|^---' $PATCH_DIR/patch | cut -c '5-' | $GREP -v /dev/null | sort | uniq > $TMP
+  $GREP '^+++ \|^--- ' $PATCH_DIR/patch | cut -c '5-' | $GREP -v /dev/null | sort | uniq > $TMP
   
   # if all of the lines start with a/ or b/, then this is a git patch that
   # was generated without --no-prefix

hadoop-common-project/hadoop-common/CHANGES.txt

@@ -259,6 +259,9 @@ Trunk (Unreleased)
     HADOOP-8839. test-patch's -1 on @author tag presence doesn't cause
     a -1 to the overall result (harsh)
 
+    HADOOP-8918. test-patch.sh is parsing modified files wrong.
+    (Raja Aluri via suresh)
+
   OPTIMIZATIONS
 
     HADOOP-7761. Improve the performance of raw comparisons. (todd)