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
{{ message }}
Repository navigation
RFC: Should we remove pyarrow feature from datafusion core #14197
Is your feature request related to a problem or challenge?
Currently the only place Python interface code exists in the core DataFusion repo is in datafusion/common/src/pyarrow.rs. This contains methods for converting to/from ScalarValue and pyarrow objects. Since datafusion-python repository was split out from the core repo, it may make sense to remove this module. Rust users who want to use these conversions can instead use the datafusion-python, which arguably has far more support for datafusion-python interactions.
Describe the solution you'd like
Remove the pyarrow module and associated features. This would be a breaking change, but likely has extremely little impact on downstream users.
Describe alternatives you've considered
Leave as is, but this prevents things like updating pyo3 downstream. It must stay pinned at the same version as datafusion core.
Additional context
No response
Reacted by Kyle Barron, Piotr Findeisen and Heran Lin
Activity
[-]RFC: Should we remove pyarrow from datafusion[/-][+]RFC: Should we remove pyarrow feature from datafusion core[/+]andygrove commentedon Jan 19, 2025
This seems logical to me
Omega359 commentedon Jan 21, 2025
agreed, remove.
robtandy commentedon Jan 24, 2025
+1 remove
timsaucer commentedon Jan 25, 2025
Great, once I have
datafusion-pythonupdated, I'll put up this PRchore: remove `pyarrow` feature (#18528)