Python Core Dumped

[StylianosBakas]

I am trying to run the commands in python as described in the paper.
env = football_env.create_environment(env_name='11_vs_11_stochastic',render=True)
env.reset()
done = False
while not done:
action = env.action_space.sample()
observation, reward, done, info = env.step(action)
But every time I get the same error

python3: /tmp/pip-build-8hrcw4hm/gfootball/third_party/gfootball_engine/src/base/utils.cpp:244: blunted::Vector3 blunted::GetVectorFromString(const string&): Assertion `tokenizedString.size() <= 3' failed.
Aborted (core dumped)

[cyfra]

Thanks for the bug report - could you also mention how you've installed the environment ? (docker, from pypi ? from source?)

[qstanczyk]

Also, are you sure you don't have any local changes made to the environment?

[StylianosBakas]

Yes I did not change anything. I just ran the recommended code from the paper.

import gfootball.env as football_env

env = football_env.create_environment(env_name='11_vs_11_stochastic',render=True)
env.reset()
done = False
while not done:
action = env.action_space.sample()
observation, reward, done, info = env.step(action)

I installed using pypi. I am running i on Ubuntu 18.04 LTS. With python 3.6.8.

[StylianosBakas]

Update: the Error is caused by the env.reset() command

[qstanczyk]

We are not able to reproduce this problem. Could you make a code change inside utils.cpp:GetVectorFromString function in order to log the input vecString being passed to this method upon hitting a problem? Would be great to also get a stack trace from there, you can replace assert with
if (tokenizedString.size() > 3) {
Log(e_FatalError...
}

Thanks a lot.

[shi-yan]

I saw the same problem on ubuntu 18.04 with nvidia 1080 ti driver 415, python 3.6, cuda 10.0

the core dump doesn't provide meaningful callstack

git clone https://github.com/google-research/football.git
cd football
pip3 install tensorflow-gpu 
pip3 install .[tf_gpu] --process-dependency-links
python3 -m gfootball.examples.run_ppo2 --dump_full_episodes=True --render=True

I didn't use docker, this was on native.

[qstanczyk]

Are you guys certain there is no second instance of the environment still running in the background (unclean termination of the previous run)?
We have seen similar reports (failures with random stack traces) in #11

[shi-yan]

the issue appears to be related to the following line:
from baselines.ppo2 import ppo2

I wasn't sure if baselines was installed, so I tried
pip3 install baselines,

then I got

    You appear to be missing MuJoCo.  We expected to find the file here: /home/xxx/.mujoco/mujoco200
    
    This package only provides python bindings, the library must be installed separately.
    
    Please follow the instructions on the README to install MuJoCo
    
        https://github.com/openai/mujoco-py#install-mujoco
    
    Which can be downloaded from the website
    
        https://www.roboti.us/index.html

it appears MujoCo is a dependency, but it is proprietary

[cyfra]

@shi-yan - you must install the baselines version from the github (rather than pypi). See #1 for details.

[shi-yan]

@cyfra you were right, after applying the workaround in #1 , things worked. Thank you!

[kkurach]

Seems the problem was resolved, so closing the issue.