Use API instead of S3 to get the current dataset .zip

[zygmuntz]

Implementing #36

[xanderdunn]

This is a great start, thanks! Please remove the download_dataset function in 3_util.py since it isn't used anymore.

[xanderdunn]

Also, although it works, it's not a good idea to use the testing_api.py to get the production dataset. That file has special behavior for the unit and integration tests.

Instead, add numerapi as a dependency and import it with from numerapi.numerapi import NumerAPI.

[zygmuntz]

Done that. The CI test is failing, I don't know why. The details link is 404.

[xanderdunn]

Great, thanks! CI doesn't pass only because this is a non-internal branch. It's fine.

Can you also add numerapi to the setup.py dependencies, and remove the zipfile import in s3_util.py since it isn't used anymore. Than it should be good for merge.

[zygmuntz]

Sure thing.

[xanderdunn]

The linter has a nit-pick: Remove the dest_path= form this line:
https://github.com/numerai/submission-criteria/pull/56/files#diff-3481e7e91a5b7dd3985f325aa5f95e87R58

It's a positional argument, so it doesn't want a keyword.

If you have pylint installed, you can see the output of the linter using the command that's in the .circleci/config.yml:

find . -iname "*.py" ! -name "setup.py" ! -name "__init__.py" ! -path "./venv/*" | xargs pylint --rcfile=./.pylintrc

[zygmuntz]

OK