4

I ran df.to_feather method in pandas 0.20.3, and got an error message advising me to run:

conda install feather-format -c conda-forge

Running that warns that it will change many python packages, and some appear to be (small) downgrades. Is this expected behavior? Would it be better to wait for the next update of Anaconda (I now have Anaconda 5.0.1 x64), which presumably will have pandas 0.21 in it? Thanks.

Below is the list of changes that would be made if I install feather:

The following NEW packages will be INSTALLED:

    arrow-cpp:                     0.7.1-py36_vc14_2     conda-forge [vc14]
    backports.functools_lru_cache: 1.4-py36_1            conda-forge
    feather-format:                0.4.0-py36_vc14_2     conda-forge [vc14]
    krb5:                          1.14.2-vc14_0         conda-forge [vc14]
    parquet-cpp:                   1.3.0.post-vc14_2     conda-forge [vc14]
    pyarrow:                       0.7.1-py36_vc14_1     conda-forge [vc14]
    snappy:                        1.1.7-vc14_1          conda-forge [vc14]

The following packages will be UPDATED:

    anaconda:                      5.0.1-py36h8316230_2              --> custom-py36h363777c_0
    hdf5:                          1.10.1-vc14hb361328_0             --> 1.10.1-vc14_1         conda-forge [vc14]
    jpeg:                          9b-vc14h4d7706e_1                 --> 9b-vc14_2             conda-forge [vc14]
    libxml2:                       2.9.4-vc14h8fd0f11_5              --> 2.9.5-vc14_1          conda-forge [vc14]
    matplotlib:                    2.1.0-py36h11b4b9c_0              --> 2.1.0-py36_1          conda-forge
    pillow:                        4.2.1-py36hdb25ab2_0              --> 4.3.0-py36_1          conda-forge

The following packages will be SUPERSEDED by a higher-priority channel:

    bzip2:                         1.0.6-vc14hdec8e7a_1              --> 1.0.6-vc14_1          conda-forge [vc14]
    conda:                         4.3.30-py36h7e176b0_0             --> 4.3.29-py36_0         conda-forge
    conda-env:                     2.6.0-h36134e3_1                  --> 2.6.0-0               conda-forge
    curl:                          7.55.1-vc14hdaba4a4_3             --> 7.55.1-vc14_0         conda-forge [vc14]
    freetype:                      2.8-vc14h17c9bdf_0                --> 2.7-vc14_2            conda-forge [vc14]
    icu:                           58.2-vc14hc45fdbb_0               --> 58.2-vc14_0           conda-forge [vc14]
    libiconv:                      1.15-vc14h29686d3_5               --> 1.14-vc14_4           conda-forge [vc14]
    libpng:                        1.6.32-vc14h5163883_3             --> 1.6.28-vc14_2         conda-forge [vc14]
    libssh2:                       1.8.0-vc14hcf584a9_2              --> 1.8.0-vc14_2          conda-forge [vc14]
    libtiff:                       4.0.8-vc14h04e2a1e_10             --> 4.0.7-vc14_1          conda-forge [vc14]
    libxslt:                       1.1.29-vc14hf85b8d4_5             --> 1.1.29-vc14_5         conda-forge [vc14]
    lzo:                           2.10-vc14h0a64fa6_1               --> 2.10-vc14_0           conda-forge [vc14]
    openssl:                       1.0.2l-vc14hcac20b0_2             --> 1.0.2l-vc14_0         conda-forge [vc14]
    qt:                            5.6.2-vc14h6f8c307_12             --> 5.6.2-vc14_1          conda-forge [vc14]
    sqlite:                        3.20.1-vc14h7ce8c62_1             --> 3.19.3-vc14_1         conda-forge [vc14]
    tk:                            8.6.7-vc14hb68737d_1              --> 8.6.6-vc14_5          conda-forge [vc14]
    yaml:                          0.1.7-vc14hb31d195_1              --> 0.1.6-vc14_0          conda-forge [vc14]
    zlib:                          1.2.11-vc14h1cdd9ab_1             --> 1.2.11-vc14_0         conda-forge [vc14]
techvslife
  • 2,273
  • 2
  • 20
  • 26
  • hope you could resolve this meanwhile, if not, install `pyarrow` it contains `feather` and you can import it with `import pyarrow.feather as feather`. – Darkonaut Feb 01 '18 at 16:52
  • Thanks, install pyarrow has the same effect in the new version. It's not a problem, as I keep it in a separate environment, as I don't like it when I get downgrades on installing a package. Hopefully this will be integrated better in future anaconda packages. – techvslife Feb 20 '18 at 04:11

1 Answers1

1

See that you'll be downgrading a bunch of standard packages. You probably can do it without fear.

I had a similar problem, my conda install was complaining of the Python version:

ResolvePackageNotFound:
  - python=3.8.3

My venv was using python 3.8.8.

I don't like to mix pip and conda, but I'm in a hurry and running pip install feather-format solved my problem.

neves
  • 33,186
  • 27
  • 159
  • 192