You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Traditionally, `pandas.Index([1, 2]).equals([1, 2])` returned True. In
0.19.0, it returns False, and we have to explicitly upcast the argument
to an Index to get the test we want.
pandas-dev/pandas#14411
Well, we had some discussion about this, see #13986 (comment) (I thought we had a longer discussion somewhere else, but can't find it directly), but I think I lost the discussion for allowing this.
Basically, to be equal to an index, you now have to be an index yourself (so a list can no longer be equal). But I see we didn't really explain this in the whatsnew docs.
I can see the argument for the new behavior, but it was a backwards incompatible change with no deprecation, so wasn't sure what was going on.
I've already worked around it for patsy.
toobaz
added
Index
Related to the Index class or subclasses
Docs
and removed
Indexing
Related to indexing on series/frames, not to indexes themselves
labels
Jun 28, 2019
toobaz
added a commit
to toobaz/pandas
that referenced
this issue
Jun 28, 2019
Traditionally, we have:
But this changed
Looks like it might not be intentional so FYI.
(Noticed because it broke a few tests in the patsy test suite.)
The text was updated successfully, but these errors were encountered: