-
-
Notifications
You must be signed in to change notification settings - Fork 25.9k
Scikit-learn 0.21.1 strange failure using nosetests with import train_test_split #13943
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
When you say "only" which versions are you excluding? Is there anyway to do
a git bisect or do you think it's a packaging issue
|
Unfortunately the discussion is quite brief so I don't know what the context of its removal was |
Whoops. Thanks for pointing it out. I made a PR. |
NP. I just migrated my package to pytest today as a result of this. Nose, I guess, was just picking up on the word test in the function name, maybe? So for me, the issue is gone, but I assume lots of packages rely on test_train_split in tests and may still use nose. Could be a bunch of failures popping up around the ecosystem. I guess there needs to be a point where old frameworks need to retire though. Could be the push we all need? |
Description
Hey there, I run a simple test suite using nose over my package gplearn and the CRON scheduled tests came up with a failure today. I narrowed the test scopes and removed the test that used that function in a dummy PR on my package and it appears simply importing test_train_split in the test code causes a failure. This only occurs with nosetests, pytest shows no failures locally, while pytest passes cleanly locally. My local configuration is linux, built from source. My travis config is all on pre-built conda packages.
This issue does not occur when running things as code, only through nosetest that I can tell. Tests using test_train_split pass just fine, but then a new "test" appears automagically that fails right after the test file has completed all other tests in the file.
The reason this is happening is a bit of a mystery to me, I don't think I'm doing anything unusual in my test configuration. #13483 appears to be the only recent PR to touch this code. I don't see any reason why that should produce these results though.
See here for the failures which only occur on 0.21.1: https://travis-ci.org/trevorstephens/gplearn/builds/537063072 which is running trevorstephens/gplearn#167
Steps/Code to Reproduce
nosetest -v -s test.py
Expected Results
No error is thrown due to simply importing test_train_split
Actual Results
Versions
Seems to only affect 0.21.1 on both python 3.6, 3.7
The text was updated successfully, but these errors were encountered: