jrmccluskey commented on code in PR #40704:
URL: https://github.com/apache/beam/pull/40704#discussion_r4340361521
##########
sdks/python/setup.py:
##########
@@ -156,12 +156,9 @@ def cythonize(*args, **kwargs):
'pyarrow>=14.0.1,<26.0.0',
]
-# Exclude pandas<=1.4.2 since it doesn't work with numpy 1.24.x.
-# Exclude 1.5.0 and 1.5.1 because of
-# https://github.com/pandas-dev/pandas/issues/45725
# must update the below "docs" and "test" for extras_require
dataframe_dependency = [
- 'pandas>=1.4.3,!=1.5.0,!=1.5.1,<2.4',
+ 'pandas>=2.2.2,<2.4',
Review Comment:
The numpy ranges were updated to cover the first release that provides
Python 3.11 wheels all the way up to the latest minor release, which is
compatible with our code.
Pandas, on the other hand, has a quirk where the first release that supports
Python 3.11, 1.5.3, doesn't contain a restriction for `numpy<2` despite
actually being incompatible with numpy 2+. Pandas 2.2.2 is the first version
that supports `numpy>=2` so we avoid the issue of a dependency mismatch that
leads to pandas crashing on import. The upper bound for pandas is functionally
just bounding to the 2.3.3 release, which was followed by 3.0.0. The dataframes
code will need an overhaul to support pandas 3+, so I pinned it here for now.
I can add the context for pandas versions in the code + file a feature
request to support pandas 3 if there isn't one already.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
