Install support APK together with test APK

build/android/pylib/instrumentation/test_package.py

Index: build/android/pylib/instrumentation/test_package.py
diff --git a/build/android/pylib/instrumentation/test_package.py b/build/android/pylib/instrumentation/test_package.py
index 52e2e0995f9a581cd4f3ea959bddc36c4146be7c..4d6ba27ac532b37306bf4a09d34ea8f70a95c671 100644
--- a/build/android/pylib/instrumentation/test_package.py
+++ b/build/android/pylib/instrumentation/test_package.py
@@ -19,6 +19,9 @@ class TestPackage(test_jar.TestJar):
     self._apk_path = apk_path
     self._apk_name = os.path.splitext(os.path.basename(apk_path))[0]
     self._package_name = apk_helper.GetPackageName(self._apk_path)
+    self._support_apk_name = "%sSupport%s" % (

[jbudorick, 2014/06/17 20:17:08]

I don't think hardcoding this is a good idea. Can we specify the support apk
name in another way?

[aberent, 2014/06/17 20:39:42]

We could add it as an extra instrumentation option to test_runner.py , but I
don't like this because it makes the command more complicated and hence easier
to get wrong (particularly since only some groups of instrumentation tests need
this). I am not sure what other options there are.

+        os.path.splitext(self._apk_path)[0],
+        os.path.splitext(self._apk_path)[1])
 
   def GetApkPath(self):
     """Returns the absolute path to the APK."""
@@ -35,4 +38,6 @@ class TestPackage(test_jar.TestJar):
   # Override.
   def Install(self, device):
     device.Install(self.GetApkPath())
+    if os.path.exists(self._support_apk_name):
+      device.Install(self._support_apk_name)