search

aidudezzz / deepbots

A wrapper framework for Reinforcement Learning in the Webots robot simulator using Python 3.
https://deepbots.readthedocs.io/
GNU General Public License v3.0
236 stars 50 forks source link

Setup.py now uses requirements.txt #111

Closed tsampazk closed 2 years ago

tsampazk commented 2 years ago

Added parsing of requirements.txt in setup.py and removed unused requirements. Now gym gets properly installed via pip.

I tested it by cloning the repo and checking out the fix-requirements branch, creating a venv and running pip install -e . in the root directory of cloned deepbots. Pip successfully installs deepbots as well as gym and its dependencies.

Edit: Also removed the unused version line from the __init__.py file.

KelvinYang0320 commented 2 years ago

@tsampazk Should we specifiy the version of gym? If we install deepbots first and then install SB3, we will need to downgrade gym to 0.21. https://github.com/DLR-RM/stable-baselines3/blob/d64bcb401ad7d45799af1feee5c1058943be23f0/setup.py#L76

tsampazk commented 2 years ago

Hmm this is tricky. SB3 is not a core requirement for deepbots. We could preemptively make sure that users don't have an issue by specifying gym==0.21, but it doesn't feel right, due to future support reasons, etc. Or we can leave it up to the specific example "installation" to downgrade gym if needed. What do you think?

KelvinYang0320 commented 2 years ago

we can leave it up to the specific example "installation" to downgrade gym if needed.

@tsampazk Yeah, I think that is a good idea.

tsampazk commented 2 years ago

Thanks @KelvinYang0320 , just straight away committed your suggestions.

KelvinYang0320 commented 2 years ago

@tsampazk Could you squash to one commit? I think the PR is good to go.

tsampazk commented 2 years ago

Thanks @KelvinYang0320, fixed the version typo and squashed the commits again. Should be good to go now?