Fix incorrect test discovery due to np.arctan2 being left as a global#12858
Fix incorrect test discovery due to np.arctan2 being left as a global#12858pllim merged 1 commit intoastropy:mainfrom
Conversation
This should fix the CI failure in astropy#12853.
|
Nice sleuthing! I was at first surprised this file gets checked at all, since it is not documented, but only then realized that pytest collection and sphinx documentation are of course completely unrelated. It still seems odd, however, that a doctest would run on something that is not actually defined in the module. But checking, I see that there is just no way to see where the So, am happy to go with this - though do please add a comment! Also, I guess if we will eventually use the new pytest-doctestplus functionality in LTS, this should be backported? p.s. Longer term, it might more sense to move all those additions to a function, as is done for |
This comment was marked as outdated.
This comment was marked as outdated.
|
Oh wait... My PR is not supposed to be merge. Never mind... 🤪 |
…tan2 being left as a global
…858-on-v5.0.x Backport PR #12858 on branch v5.0.x (Fix incorrect test discovery due to np.arctan2 being left as a global)
This should fix the CI failure in #12853.