TimeLimit refactor with Monitor Simplification (openai#482)

* fix double reset, as suggested by @jietang

* better floors and ceilings

* add convenience methods to monitor

* add wrappers to gym namespace

* allow playing Atari games, with potentially more coming in the future

* simplify example in docs

* Move play out of the Env

* fix tests

* no more deprecation warnings

* remove env.monitor

* monitor simplification

* monitor simplifications

* monitor related fixes

* a few changes suggested by linter

* timestep_limit fixes

* keep track of gym env variables for future compatibility

* timestep_limit => max_episode_timesteps

* don't apply TimeLimit wrapper in make for VNC envs

* Respect old timestep_limit argument

* Pass max_episode_seconds through registration

* Don't include deprecation warnings yet

gym/__init__.py

@@ -47,5 +47,6 @@ def sanity_check_dependencies():
 from gym.benchmarks import benchmark_spec
 from gym.envs import make, spec
 from gym.scoreboard.api import upload
+from gym import wrappers
 
-__all__ = ["Env", "Space", "Wrapper", "make", "spec", "upload"]
+__all__ = ["Env", "Space", "Wrapper", "make", "spec", "upload", "wrappers"]

gym/benchmarks/__init__.py

@@ -269,30 +269,44 @@
         {'env_id': 'HalfCheetah-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  -280.0,
+         'reward_ceiling': 4000.0,
         },
         {'env_id': 'Hopper-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  16.0,
+         'reward_ceiling': 4000.0,
         },
         {'env_id': 'InvertedDoublePendulum-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  53.0,
+         'reward_ceiling': 10000.0,
         },
         {'env_id': 'InvertedPendulum-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  5.6,
+         'reward_ceiling': 1000.0,
         },
         {'env_id': 'Reacher-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  -43.0,
+         'reward_ceiling': -0.5,
         },
         {'env_id': 'Swimmer-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  0.23,
+         'reward_ceiling': 500.0,
         },
         {'env_id': 'Walker2d-v1',
          'trials': 3,
          'max_timesteps': 1000000,
+         'reward_floor':  1.6,
+         'reward_ceiling': 5500.0,
         }
     ])
 

gym/configuration.py

@@ -1,8 +1,6 @@
 import logging
 import sys
 
-import gym
-
 logger = logging.getLogger(__name__)
 
 root_logger = logging.getLogger()