Uh oh!
There was an error while loading. Please reload this page.
Making unit tests green - #748
Conversation
Codecov Report
@@ Coverage Diff @@## develop #748 +/- ##
===========================================
- Coverage 88.06% 87.79% -0.28%
===========================================
Files 36 36 Lines 4090 4096 +6 ===========================================
- Hits 3602 3596 -6 - Misses 488 500 +12
Continue to review full report at Codecov.
|
Uh oh!
There was an error while loading. Please reload this page.
Neeratyoy
commented
Aug 1, 2019
@mfeurer, @PGijsbers It looks like the appveyor build failed to install it. How can I fix that? |
mfeurer
commented
Aug 2, 2019
Appveyor doesn't use |
PGijsbers
commented
Aug 5, 2019
I hope you can figure something out. But to answer the original question, if you want to add test dependencies, it's best to add it in |
Neeratyoy
commented
Aug 5, 2019
@mfeurer |
mfeurer
commented
Aug 6, 2019
That's unfortunate. Maybe we should execute that test on the live server? |
Neeratyoy
commented
Aug 6, 2019
I guess you mean a new instance of an OpenML server every test run? |
mfeurer
commented
Aug 6, 2019
Sorry, I don't get that. What I meant is like https://github.com/openml/openml-python/blob/develop/tests/test_datasets/test_dataset_functions.py#L213 |
Neeratyoy
commented
Aug 6, 2019
aah you are okay with the dataset list being fetched from the prod server |
PGijsbers
commented
Aug 6, 2019
We do it at several points in our unit tests. But your general intuition is right, only do it if you have to. In particular never alter the production server and think twice if you really need the production server if it's a demanding query. In this case for the reasons you pointed out it makes sense. |
Neeratyoy
commented
Aug 7, 2019
@mfeurer |
Neeratyoy
commented
Aug 7, 2019
Thanks! |
mfeurer
left a comment
There was a problem hiding this comment.
I like this, this is much simpler now. I left a few comments on how I think this could be improved.
| logger = logging.getLogger("unit_tests_publish") | ||
| logger.setLevel(logging.DEBUG) | ||
| fh = logging.FileHandler('UploadedFiles.log') | ||
| # fh.setLevel(logging.DEBUG) |
There was a problem hiding this comment.
Is it actually necessary to introduce a logger which logs to a file? Also, could you please remove that comment?
| from openml.testing import TestBase | ||
| # creating logger for unit test file deletion status | ||
| logger = logging.getLogger("unit_tests") |
There was a problem hiding this comment.
This looks a lot like a copy of the logger above. Also, I assume that the logger names should be swapped.
| # creating logger for unit test file deletion status | ||
| logger = logging.getLogger("unit_tests") | ||
| logger.setLevel(logging.DEBUG) | ||
| fh = logging.FileHandler('UnitTestDeletion.log') |
There was a problem hiding this comment.
Same as above, do we really need to save this to disk?
There was a problem hiding this comment.
I agree, no.
I was tailing during the testing and this helped.
Will have them removed.
Uh oh!
There was an error while loading. Please reload this page.
| openml.config.server = TestBase.test_server | ||
| openml.config.apikey = TestBase.apikey | ||
| # legal_entities defined in openml.utils._delete_entity() - {'user'} |
There was a problem hiding this comment.
Are these comments still needed?
What does this PR implement/fix? Explain your changes.
Owing to the concurrent runs, combined with deletion of files at the end of unit test modules, some tests may fail. Rerunning them should ensure they pass.