Pandas version checks

  • [x] I have checked that this issue has not already been reported.

  • [x] I have confirmed this bug exists on the latest version of pandas.

  • [ ] I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd
import numpy as np

df = pd.DataFrame(
    {
        "a": [-1., 2., 3., 4., 5.],
        "b": [1., 2., 3., 4., 5.],
    }, dtype=pd.Float64Dtype()
)

df = np.sqrt(df)
# Returns False
print(df.isna().any().any())
# Returns True
print(pd.isna(df.loc[0, "a"]))

Issue Description

Apply a NumPy operation that yields NaN for some value of the dataframe of type Float64Dtype. Then pandas null checking functions (isna, isnull, notna) will not detect the NaN value. However, it is detected if we index the NaN value.

Expected Behavior

Both the above print statements should be True.

Installed Versions

INSTALLED VERSIONS ------------------ commit : 2cc37625532045f4ac55b27176454bbbc9baf213 python : 3.13.5 python-bits : 64 OS : Linux OS-release : 6.15.4-arch2-1 Version : #1 SMP PREEMPT_DYNAMIC Fri, 27 Jun 2025 16:35:07 +0000 machine : x86_64 processor : byteorder : little LC_ALL : None LANG : en_GB.UTF-8 LOCALE : en_GB.UTF-8 pandas : 2.3.0 numpy : 2.1.2 pytz : 2025.2 dateutil : 2.9.0.post0 pip : None Cython : None sphinx : None IPython : 9.3.0 adbc-driver-postgresql: None adbc-driver-sqlite : None bs4 : 4.13.4 blosc : None bottleneck : None dataframe-api-compat : None fastparquet : None fsspec : 2024.6.1 html5lib : None hypothesis : None gcsfs : None jinja2 : 3.1.4 lxml.etree : None matplotlib : 3.10.3 numba : 0.61.2 numexpr : None odfpy : None openpyxl : None pandas_gbq : None psycopg2 : None pymysql : None pyarrow : None pyreadstat : None pytest : 8.4.1 python-calamine : None pyxlsb : None s3fs : None scipy : 1.15.3 sqlalchemy : 2.0.41 tables : None tabulate : None xarray : None xlrd : None xlsxwriter : None zstandard : None tzdata : 2025.2 qtpy : None pyqt5 : None

Comment From: gake572

I confirmed on pandas 2.3.0 + NumPy 2.1.2 that after applying a NumPy ufunc the nullable Float64 dtype mask isn’t catching NaNs:

df_plain = df.astype(float)
print(df_plain.isna().any().any())  # True

This suggests Float64Dtype.isna() isn’t recognizing the NaN created by np.sqrt.

Can someone confirm this is a bug in the nullable array mask logic? Thanks.

Comment From: jbrockmendel

Can someone confirm this is a bug in the nullable array mask logic? Thanks.

It is not a bug but it is a design choice that frequently causes confusion (#60106, #59891, #56451, #53887). The original discussion for how to handle this is in #32265 and more recently in #61618.