stable-retro
stable-retro copied to clipboard
Add basic testing for different OSes / CPUs
As part of the effort to comply with the Farama fondation standards: details
I would love to take this one. I think that
- First I will double check current tests and discuss which ones to keep and which ones to add
- Test github actions for a stable environment
- Cover OSes to test out using gh actions How does that sound?
Sounds good!
Im now with the instructions Zac from discords kindly shared with a working install of 3.10 on 22.04. Also did a 3.8 with openai/distribution install just changing some of the instructions we had. Ran my first run of tests - had to install pytest independently with pip, maybe we will need a dev setup later? OpenAI Retro tests only returns 2 warnings, nothing else runs. Will discard that, No need to keep the reference IMHO Stable-retro tests returns 7 passed, 4 warnings and 11 errors. I will start contributing from there, fix those errors and have a pipeline running with all passing.
On a quick review, the errors came for fixtures that were missing, game
and testenv
specifically.
Noticed from #24 the removal of two lines that might actually include them, adding them as they were given threw 28 failed, 17 passed, 2194 skipped, 4 warnings, 5 errors. Interesting!
If anyone can share the reasoning behind that would be great, but I will try to join the dev meeting to double-check on that.
hi @ricalanis, this code was removed as it wasn't used explicitly later in the tests therefore it appeared like a useless import however, you are probably right that this did something under the hood. In the meantime, you can revert all of the changes that I made in the PR as I noted
I would prefer we not merge this until we have proper testing to check that this all works still.
I can look at the pre-commit code again after you get your PR working