[rllib] How to correctly set up a dev environment

RLlib
1

Hi I want to set up a development environment to potentially contribute to rllib, however the docs for contributing point to a setup-dev script which does some things I really don’t understand and puts my development environment in a state that is impossible to make a pull request from.

The “things I dont understand” comprise of deleting most/all of the python files in the repository.

Is there a way I can simply do something similar to python setup.py develop so I can make changes/test them/use my current working rllib branch in other projects?

Thanks

1 Like
2

Yeah, you could alternatively just run pip install -e . --verbose in ray/python to get the same effect!

RE your initial point: With setup-dev.py, you just need to be careful to never pip install --upgrade ray. But I understand if you don’t really want to deal with that!

1 Like
3

I tried pip intall -e . but ran into [Bazel] Build error from source · Issue #14117 · ray-project/ray · GitHub

However I;m going to apply that fix and make a PR because it works and there’s no PR just yet :stuck_out_tongue:

4
5

@rliaw any ideas why the dashboard build isn’t automated as part of the main build?

I have to run cd python/ray/new_dashboard/client && npm install && npm ci && npm run build separately (after errors trying to get ray.tune to run)

6

I’m not sure; I guess technically the dashboard is optional.