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
Thanks for reporting this. From some quick digging, it looks like the issue here is that numpy's datetime object upcasts to an array and attempts to use the __array_ufunc__ overload when the datetime is the LHS. That is why we get different results here depending on the argument ordering. We can avoid the confusing ordering behavior by wrapping the date objects in np.array, i.e.
cudf.Series([dt1]) > np.array([dt2]) # Also fails
cudf.Series([dt2]) > np.array([dt1]) # Also fails
We'll have to do a bit more digging to determine a fix. I expect that we need to add some extra preprocessing to handle the array case instead of the scalar case.
Describe the bug
Comparison of datetime objects are sensitive to run order when dtype differs.
Steps/Code to reproduce bug
Expected behavior
A clear and concise description of what you expected to happen.
Environment overview (please complete the following information)
Environment details
Please run and paste the output of the
cudf/print_env.sh
script here, to gather any other relevant environment detailsAdditional context
Add any other context about the problem here.
The text was updated successfully, but these errors were encountered: