feather-format

What it is and what it does

Feather-format is a wrapper around Apache Arrow that lets you save pandas DataFrames to disk in the Feather columnar format, which is designed for fast serialization and deserialization. It was created when Feather support was not yet built into pandas itself.

The package has significant limitations: it does not support non-string column names, row indexes, or object-type columns with mixed data types. More importantly, the package is unmaintained—its last release was in April 2020, and its classifiers declare support only for Python 2.7 and 3.4–3.5. Modern pandas and pyarrow have evolved substantially since then, and installation often requires conda-forge rather than pip because of pyarrow's compiled dependencies.

Use it for:

• Serialize pandas DataFrames to a columnar format for faster I/O when working with large datasets that fit the supported schema.
• Archive DataFrames with string column names and homogeneous data types in a format designed for efficient columnar access.
• Migrate legacy code that was written when Feather support was not yet built into pandas itself.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Feather-format provides a Python interface to store and load pandas DataFrames using the Apache Arrow-based Feather file format for efficient disk serialization.

No. The package is unmaintained (last release April 2020) and has high installation friction due to pyarrow's compiled dependencies. Modern pandas includes native Feather support via DataFrame.to_feather() and read_feather(), which is actively maintained and works with current Python versions. Use pandas' built-in Feather methods instead unless you have a specific reason to use this legacy wrapper.

Install

feather-format on PyPI

pip install feather-format

uv add feather-format

poetry add feather-format

Installing feather-format

Before you install

Installation has high friction: the package depends on pyarrow, which may not be available via pip on all platforms and often requires conda-forge instead. The package is aging—last release was 2020-04-27, over five years ago—and classifiers indicate Python 2.7 and 3.4–3.5 support, making it unmaintained for modern Python versions.

License in practice

Licensed under Apache License, Version 2.0 (permissive), which allows commercial and private use with minimal restrictions, though you must include a copy of the license and state significant changes.

Quickstart

pip install feather-format
# or: conda install feather-format -c conda-forge

import feather
feather.write_dataframe(df, 'data.feather')
df_loaded = feather.read_dataframe('data.feather')

pyarrow must be installed and available on your platform; pip installation may fail and conda-forge is often required. Package is unmaintained for modern Python versions.

Verify before relying

• Whether the package works reliably with current pandas and pyarrow versions despite being unmaintained since 2020.
• Whether pandas.DataFrame.to_feather() (built into pandas itself since ~2018) is a better alternative for your use case.
• Exact Python version compatibility beyond the outdated classifiers listed.

Package facts

Evidence: feather-format-0.4.1.tar.gz

Tags

More Information Analysis packages