Merge branch 'integration' into normalized_gym_env

rllab/algos/batch_polopt.py

@@ -107,6 +107,8 @@ def __init__(self,
 
     def start_worker(self):
         self.sampler.start_worker()
+        if self.plot:
+            plotter.init_plot(self.env, self.policy)
 
     def shutdown_worker(self):
         self.sampler.shutdown_worker()
@@ -130,12 +132,7 @@ def train(self):
                 logger.log("saved")
                 logger.dump_tabular(with_prefix=False)
                 if self.plot:
-                    rollout(
-                        self.env,
-                        self.policy,
-                        max_path_length=self.max_path_length,
-                        animated=True,
-                        speedup=5)
+                    self.update_plot()
                     if self.pause_for_plot:
                         input("Plotting evaluation run: Press Enter to "
                               "continue...")

rllab/plotter/plotter.py

@@ -1,8 +1,11 @@
 import atexit
-from queue import Empty
 from multiprocessing import Process, Queue
-from rllab.sampler.utils import rollout
 import numpy as np
+import platform
+from threading import Thread
+from queue import Empty
+
+from rllab.sampler.utils import rollout
 
 __all__ = ['init_worker', 'init_plot', 'update_plot']
 
@@ -60,12 +63,22 @@ def _shutdown_worker():
 def init_worker():
     global process, queue
     queue = Queue()
-    process = Process(target=_worker_start)
+    process = Thread(target=_worker_start) if (
+        'Darwin' in platform.platform()) else Process(target=_worker_start)
+    process.daemon = True
     process.start()
     atexit.register(_shutdown_worker)
 
 
 def init_plot(env, policy):
+    global process, queue
+    if not (process and queue):
+        init_worker()
+
+    # Needed in order to draw glfw window on the main thread
+    if ('Darwin' in platform.platform()):
+        rollout(env, policy, max_path_length=np.inf, animated=True, speedup=5)
+
     queue.put(['update', env, policy])