wesm / feather

Feather: fast, interoperable binary data frame storage for Python, R, and more powered by Apache Arrow
Apache License 2.0
2.74k stars 168 forks source link

Does feather work with Python 3.7? #361

Closed Darkonaut closed 5 years ago

Darkonaut commented 5 years ago

I'm trying to upgrade from Python 3.6 to Python 3.7 with conda, but feather seems to prevent this:

$conda install python=3.7 anaconda=custom
Solving environment: failed

UnsatisfiableError: The following specifications were found to be in conflict:
  - feather-format -> python=3.6
  - python=3.7
Use "conda info <package>" to see the dependencies for each package.

https://pypi.org/project/feather-format/ lists support only until 3.5, it still worked with 3.6 for me, though. Is this package not longer maintained?

I would like to use feather-format because it allows reading in only a subset of the columns within a feather-file. I'm also wondering why pandas.read_feather doesn't provide this feature.

wesm commented 5 years ago

It should work fine on Python 3.7. Might want to open an issue with the Anaconda developers. I tried this with conda-forge and no problems

$ conda create -n feather-test python=3.7 feather-format
Solving environment: done

==> WARNING: A newer version of conda exists. <==
  current version: 4.5.4
  latest version: 4.5.11

Please update conda by running

    $ conda update -n base conda

## Package Plan ##

  environment location: /home/wesm/miniconda/envs/feather-test

  added / updated specs: 
    - feather-format
    - python=3.7

The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    numpy-1.15.0               |py37_blas_openblashd3ea46f_200         9.3 MB  conda-forge
    pip-18.1                   |        py37_1000         1.7 MB  conda-forge
    wheel-0.32.3               |           py37_0          34 KB  conda-forge
    setuptools-40.6.2          |           py37_0         606 KB  conda-forge
    arrow-cpp-0.11.1           |   py37h3bd774a_0         6.1 MB  conda-forge
    pandas-0.23.4              |   py37hf8a1672_0        27.5 MB  conda-forge
    six-1.11.0                 |        py37_1001          21 KB  conda-forge
    feather-format-0.4.0       |          py_1003           5 KB  conda-forge
    thrift-cpp-0.11.0          |       he11dc3d_3         2.9 MB  conda-forge
    pyarrow-0.11.1             |   py37hfc679d8_0         1.9 MB  conda-forge
    python-3.7.1               |       h5001a0f_0        26.8 MB  conda-forge
    certifi-2018.11.29         |        py37_1000         145 KB  conda-forge
    ------------------------------------------------------------
                                           Total:        77.1 MB

I suggest also raising an issue with pandas about adding the feature you're describing to pandas.read_feather