TST: fix _most_ remaining xfails in tests/arithmetic #23308
Conversation
|
Hello @jbrockmendel! Thanks for updating the PR.
Comment last updated on October 24, 2018 at 01:34 Hours UTC |
| # In particular non-nanosecond timedelta64 needs to be cast to | ||
| # nanoseconds, or else we get undesired behavior like | ||
| # np.timedelta64(3, 'D') / 2 == np.timedelta64(1, 'D') | ||
| return pd.Timedelta(obj) |
There was a problem hiding this comment.
FIXME: this will screw up on np.timedelta64(“nat”)
|
This should be put on hold until after #23320. |
|
Closing until after #23320. |
Most of the remaining xfails in tests/arithmetic are caused by single-column DataFrame operations not being equivalent to Series/Index operations. This PR fixes those tests by transposing both the DataFrame and the expected result in the appropriate cases.
We end up with 3 xfailed tests instead of 223. Runtime drops from 168.9 seconds to 99.4 seconds.
Some other ancillary fixes along the way: RangeIndex correctly returns NotImplemented when operating with a DataFrame. Series operations with Tick and non-nano timedelta64 are fixed.