Adding a server side, non-blocking subprocess mechanism.

testing/legion/examples/hello_world/host_test.py

Index: testing/legion/examples/hello_world/host_test.py
diff --git a/testing/legion/examples/hello_world/host_test.py b/testing/legion/examples/hello_world/host_test.py
index 0094d9bd5611c2b7aefb872be3b4b367e9c3a5a5..317cc27d27c01019b87aed00d13f77aaa512e657 100755
--- a/testing/legion/examples/hello_world/host_test.py
+++ b/testing/legion/examples/hello_world/host_test.py
@@ -60,13 +60,16 @@ class ExampleController(host_controller.HostController):
   def CallEcho(self, client):
     """Call rpc.Echo on a client."""
     logging.info('Calling Echo on %s', client.name)
-    logging.info(self.client1.rpc.Echo(client.name))
+    logging.info(client.rpc.Echo(client.name))
 
   def CallClientTest(self, client):
     """Call client_test.py name on a client."""
     logging.info('Calling Subprocess to run "./client_test.py %s"', client.name)
-    retcode, stdout, stderr = client.rpc.Subprocess(
-        ['./client_test.py', client.name])
+    proc = client.rpc.subprocess.Popen(['./client_test.py', client.name])
+    client.rpc.subprocess.Wait(proc)
+    retcode = client.rpc.subprocess.GetReturncode(proc)
+    stdout = client.rpc.subprocess.GetStdout(proc)
+    stderr = client.rpc.subprocess.GetStderr(proc)
     logging.info('retcode: %s, stdout: %s, stderr: %s', retcode, stdout, stderr)