Should env.step() return done and info?

[mctigger]

autonomous-learning-library/all/environments/abstract.py

Lines 42 to 51 in 68d355a

	Returns
	-------
	State
	The state of the environment after the action is applied
	float
	The reward achieved by the previous action
	done
	True if the environment has entered a terminal state and should be reset
	info
	Diagnostic information useful for debugging

In the abstract Environment class it says that the step() methods should return the state, the reward, an indicator whether the episode ends and some information.

However, in the GymEnvironment only the reward and the state are returned:

autonomous-learning-library/all/environments/gym.py

Line 44 in 68d355a

return self._state, self._reward

Which usage should be the correct one?

[cpnota]

Ah, looks like the documentation is out of date. The State objective includes done and info now.