This feedstock currently installs both matplotlib and matplotlib-base

[rfox12]

I'm on the latest Miniconda (4.8.2) and notice when I install this in a new environment:
conda create -n tstest fbprophet
That you'll get both matplotlib and matplotlib-base. I see the meta.yaml says "matplotlib-base", but somehow the full install is getting pulled in? I don't understand exactly how conda-smithy works, but isn't the original fbprophet requirement for "matplotlib" supposed to be overwritten with matplotlib-base?

[maresb]

Indeed. On Miniconda 4.8.2, for some reason it installs fbprophet v0.5, which had the faulty dependency on matplotlib instead of matplotlib-base. So the question is, what's preventing it from installing v0.6?

[maresb]

Ok, when I run from the continuumio/miniconda docker image and update Miniconda to 4.8.2, I get the problem described here. However, when I run from continuumio/miniconda3 (the Python 3 version) and update Miniconda to 4.8.2, everything works as expected.

Are you using a conda version based on Python 2? In this case, when I do conda install -c conda-forge fbprophet==0.6 I get

UnsatisfiableError: The following specifications were found
to be incompatible with the existing python installation in your environment:

Specifications:

  - fbprophet==0.6 -> python[version='>=3.6,<3.7.0a0|>=3.7,<3.8.0a0']

Your python: python=2.7

In other words, it seems that v0.6 ends Python 2 support, so it downloads the old version depending on matplotlib. Probably best to upgrade to Python 3 if possible.

[rfox12]

Thanks for taking a look... a few more details. I've got the Python 3.7 install of Miniconda (I think Python 2 is finally dead to me!). I'm on Windows 10... could that be it? Otherwise it's a really clean environment. Channel strict to conda-forge. And I got v0.6 as expected. Everything looks OK just got a bunch of Qt packages and stuff I don't need from the full matplotlib.

[rfox12]

I did poke around a bit.... couldn't really find a nice description of how to use matplotlib-base in conda recipes. I'll try uninstalling both and re-installing just matplotlib-base... that might be the work-around for me.

[maresb]

Very interesting.

For the record, when I run

$ docker run --rm -it continuumio/miniconda3 /bin/bash
(base) root@dc279e4330cb:/# conda config --add channels conda-forge
(base) root@dc279e4330cb:/# conda config --set channel_priority strict
(base) root@dc279e4330cb:/# conda update conda
(base) root@dc279e4330cb:/# conda install fbprophet

I see the following:

  arviz              conda-forge/noarch::arviz-0.6.1-py_0
  binutils_impl_lin~ conda-forge/linux-64::binutils_impl_linux-64-2.33.1-h53a641e_8
  binutils_linux-64  conda-forge/linux-64::binutils_linux-64-2.33.1-h9595d00_16
  bzip2              conda-forge/linux-64::bzip2-1.0.8-h516909a_2
  cftime             conda-forge/linux-64::cftime-1.0.4.2-py37hc1659b7_0
  convertdate        conda-forge/noarch::convertdate-2.1.3-py_1000
  curl               conda-forge/linux-64::curl-7.68.0-hf8cf82a_0
  cycler             conda-forge/noarch::cycler-0.10.0-py_2
  cython             conda-forge/linux-64::cython-0.29.15-py37he1b5a44_0
  ephem              conda-forge/linux-64::ephem-3.7.7.1-py37h516909a_0
  fbprophet          conda-forge/linux-64::fbprophet-0.6-py37he1b5a44_0
  freetype           conda-forge/linux-64::freetype-2.10.0-he983fc9_1
  gcc_impl_linux-64  conda-forge/linux-64::gcc_impl_linux-64-7.3.0-hd420e75_5
  gcc_linux-64       conda-forge/linux-64::gcc_linux-64-7.3.0-h553295d_16
  gxx_impl_linux-64  conda-forge/linux-64::gxx_impl_linux-64-7.3.0-hdf63c60_5
  gxx_linux-64       conda-forge/linux-64::gxx_linux-64-7.3.0-h553295d_16
  hdf4               conda-forge/linux-64::hdf4-4.2.13-hf30be14_1003
  hdf5               conda-forge/linux-64::hdf5-1.10.5-nompi_h3c11f04_1104
  holidays           conda-forge/noarch::holidays-0.10.1-py_0
  icu                conda-forge/linux-64::icu-64.2-he1b5a44_1
  jpeg               conda-forge/linux-64::jpeg-9c-h14c3975_1001
  kiwisolver         conda-forge/linux-64::kiwisolver-1.1.0-py37hc9558a2_0
  krb5               conda-forge/linux-64::krb5-1.16.4-h2fd8d38_0
  libblas            conda-forge/linux-64::libblas-3.8.0-14_openblas
  libcblas           conda-forge/linux-64::libcblas-3.8.0-14_openblas
  libcurl            conda-forge/linux-64::libcurl-7.68.0-hda55be3_0
  libedit            conda-forge/linux-64::libedit-3.1.20170329-hf8c457e_1001
  libgfortran-ng     conda-forge/linux-64::libgfortran-ng-7.3.0-hdf63c60_5
  liblapack          conda-forge/linux-64::liblapack-3.8.0-14_openblas
  libnetcdf          conda-forge/linux-64::libnetcdf-4.7.3-nompi_h9f9fd6a_101
  libopenblas        conda-forge/linux-64::libopenblas-0.3.7-h5ec1e0e_6
  libpng             conda-forge/linux-64::libpng-1.6.37-hed695b0_0
  libssh2            conda-forge/linux-64::libssh2-1.8.2-h22169c7_2
  lunarcalendar      conda-forge/noarch::lunarcalendar-0.0.9-py_0
  matplotlib-base    conda-forge/linux-64::matplotlib-base-3.1.3-py37h250f245_0
  netcdf4            conda-forge/linux-64::netcdf4-1.5.3-nompi_py37hd35fb8e_102
  numpy              conda-forge/linux-64::numpy-1.18.1-py37h95a1406_0
  packaging          conda-forge/noarch::packaging-20.1-py_0
  pandas             conda-forge/linux-64::pandas-1.0.1-py37hb3f55d8_0
  pyparsing          conda-forge/noarch::pyparsing-2.4.6-py_0
  pystan             conda-forge/linux-64::pystan-2.19.1.1-py37hb3f55d8_1
  python-dateutil    conda-forge/noarch::python-dateutil-2.8.1-py_0
  pytz               conda-forge/noarch::pytz-2019.3-py_0
  scipy              conda-forge/linux-64::scipy-1.4.1-py37h921218d_0
  tornado            conda-forge/linux-64::tornado-6.0.3-py37h516909a_4
  xarray             conda-forge/noarch::xarray-0.15.0-py_0

A lot of these are linux-64 architecture, so perhaps it is Windows which is causing the issue. It would be interesting to compare our lists and track down the dependency. Unfortunately, reverse dependency search in conda is broken, so there doesn't seem to be a good systematic way to do this.

[rfox12]

Below is my dump... lines up pretty good... The main thing that concerns me here is the older version of pystan

And I was able to use conda-tree to figure out that pystan is in fact the guilty party:

>conda install conda-tree
...
>conda-tree leaves
['conda-tree', 'fbprophet']
>conda-tree whoneeds matplotlib
['pystan']

Full output of conda list for fbprophet on Windows 10 w/Python 3.7.6 and Conda 4.8.2:

ca-certificates           2019.11.28           hecc5488_0    conda-forge
certifi                   2019.11.28               py37_0    conda-forge
convertdate               2.1.3                   py_1000    conda-forge
cycler                    0.10.0                     py_2    conda-forge
cython                    0.29.15          py37h6538335_0    conda-forge
ephem                     3.7.7.1          py37hfa6e2cd_0    conda-forge
fbprophet                 0.6              py37h6538335_0    conda-forge
freetype                  2.10.0               h563cfd7_1    conda-forge
holidays                  0.10.1                     py_0    conda-forge
icu                       64.2                 he025d50_1    conda-forge
intel-openmp              2020.0                      166
jpeg                      9c                hfa6e2cd_1001    conda-forge
kiwisolver                1.1.0            py37he980bc4_0    conda-forge
libblas                   3.8.0                    15_mkl    conda-forge
libcblas                  3.8.0                    15_mkl    conda-forge
libclang                  9.0.1           default_hf44288c_0    conda-forge
liblapack                 3.8.0                    15_mkl    conda-forge
libpng                    1.6.37               h7602738_0    conda-forge
lunarcalendar             0.0.9                      py_0    conda-forge
m2-base                   1.0.0                         3
m2-bash                   4.3.042                       5
m2-bash-completion        2.3                           2
m2-catgets                1.1                           3
m2-coreutils              8.25                        102
m2-dash                   0.5.8                         2
m2-diffutils              3.3                           4
m2-file                   5.25                          2
m2-filesystem             2016.04                       4
m2-findutils              4.6.0                         2
m2-gawk                   4.1.3                         2
m2-gcc-libs               5.3.0                         4
m2-gettext                0.19.7                        4
m2-gmp                    6.1.0                         3
m2-grep                   2.22                          4
m2-gzip                   1.7                           2
m2-inetutils              1.9.2                         2
m2-info                   6.0                           2
m2-less                   481                           2
m2-libasprintf            0.19.7                        4
m2-libbz2                 1.0.6                         3
m2-libcatgets             1.1                           3
m2-libcrypt               1.3                           2
m2-libgettextpo           0.19.7                        4
m2-libiconv               1.14                          3
m2-libintl                0.19.7                        4
m2-liblzma                5.2.2                         2
m2-libpcre                8.38                          2
m2-libreadline            6.3.008                       8
m2-libutil-linux          2.26.2                        2
m2-libxml2                2.9.2                         3
m2-make                   4.1                           5
m2-mintty                 1!2.2.3                       2
m2-mpfr                   3.1.4                         2
m2-msys2-launcher-git     0.3.28.860c495                2
m2-msys2-runtime          2.5.0.17080.65c939c               3
m2-ncurses                6.0.20160220                  2
m2-sed                    4.2.2                         3
m2-tar                    1.28                          4
m2-tftp-hpa               5.2                           2
m2-time                   1.7                           2
m2-ttyrec                 1.0.8                         2
m2-tzcode                 2015.e                        2
m2-unzip                  6.0                           3
m2-util-linux             2.26.2                        2
m2-which                  2.21                          3
m2-zip                    3.0                           2
m2-zlib                   1.2.8                         4
m2w64-binutils            2.25.1                        5
m2w64-bzip2               1.0.6                         6
m2w64-crt-git             5.0.0.4636.2595836               2
m2w64-gcc                 5.3.0                         6
m2w64-gcc-ada             5.3.0                         6
m2w64-gcc-fortran         5.3.0                         6
m2w64-gcc-libgfortran     5.3.0                         6
m2w64-gcc-libs            5.3.0                         7
m2w64-gcc-libs-core       5.3.0                         7
m2w64-gcc-objc            5.3.0                         6
m2w64-gmp                 6.1.0                         2
m2w64-headers-git         5.0.0.4636.c0ad18a               2
m2w64-isl                 0.16.1                        2
m2w64-libiconv            1.14                          6
m2w64-libmangle-git       5.0.0.4509.2e5a9a2               2
m2w64-libwinpthread-git   5.0.0.4634.697f757               2
m2w64-make                4.1.2351.a80a8b8               2
m2w64-mpc                 1.0.3                         3
m2w64-mpfr                3.1.4                         4
m2w64-pkg-config          0.29.1                        2
m2w64-toolchain           5.3.0                         7
m2w64-toolchain_win-64    2.4.0                         0    conda-forge
m2w64-tools-git           5.0.0.4592.90b8472               2
m2w64-windows-default-manifest 6.4                           3
m2w64-winpthreads-git     5.0.0.4634.697f757               2
m2w64-zlib                1.2.8                        10
matplotlib                3.1.3                    py37_0    conda-forge
matplotlib-base           3.1.3            py37h2981e6d_0    conda-forge
mkl                       2020.0                      166
msys2-conda-epoch         20160418                      1
numpy                     1.18.1           py37hc71023c_0    conda-forge
openssl                   1.1.1d               hfa6e2cd_0    conda-forge
pandas                    1.0.1            py37he350917_0    conda-forge
pip                       20.0.2                     py_2    conda-forge
posix                     1.0.0                         2
pyparsing                 2.4.6                      py_0    conda-forge
pyqt                      5.12.3           py37h6538335_1    conda-forge
pyqt5-sip                 4.19.18                  pypi_0    pypi
pyqtwebengine             5.12.1                   pypi_0    pypi
pystan                    2.17.1.0        py37hb8a7f35_1004    conda-forge
python                    3.7.6                h5b45d93_2    conda-forge
python-dateutil           2.8.1                      py_0    conda-forge
pytz                      2019.3                     py_0    conda-forge
qt                        5.12.5               h7ef1ec2_0    conda-forge
setuptools                45.2.0                   py37_0    conda-forge
six                       1.14.0                   py37_0    conda-forge
sqlite                    3.30.1               hfa6e2cd_0    conda-forge
tornado                   6.0.3            py37hfa6e2cd_4    conda-forge
vc                        14.1                 h0510ff6_4
vs2015_runtime            14.16.27012          hf0eaf9b_1
wheel                     0.34.2                     py_1    conda-forge
wincertstore              0.2                   py37_1003    conda-forge
zlib                      1.2.11            h2fa13f4_1006    conda-forge```

[rfox12]

One interesting/disturbing finding... I am on a 64bit machine (who isn't these days?) and I went over to anaconda and see that indeed there is a new version of PyStan for win-64: https://anaconda.org/conda-forge/pystan/files

My conda info reports platform : win-64 ... so I'm scratching my head wondering if I somehow got the win-32 distribution? Not sure yet.

[rfox12]

Ok well you can close this issue if you want. I found a solution... but I it would sit better with me if I fully understood it!

If I do this:
conda install fbprophet
Conda will list the packages it's going to install, and among them conda-forge/win-64::pystan-2.17.1.0-py37hb8a7f35_1004 and it will install both matplotlib and matplotlib-base

However if I do this:
conda install pystan fbprophet
I see it will install conda-forge/win-64::pystan-2.19.1.1-py37h12199cf_1 (the newest version) AND I will only get matplotlib-base!

So there must be a difference between peer packages vs dependent packages? Perhaps something specific to the way that conda handles matplotlib-base?

[maresb]

This goes beyond my understanding of conda. Perhaps it makes sense to close this issue and open a new one on pystan-feedstock pointing to our discussion here? Also, maybe you'll get someone more knowledgeable over there. ;)

[rmax]

Is this causing a problem to you?

Latest build shows that installing both matplotlib and matplotlib-base are part of the expected behaviour, even though matplotlib is not declared as explicit dependency. See https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=123562&view=logs&jobId=5be07ae1-d8ba-5406-47b6-8e3a3a12f825&j=517fe804-fa30-5dc2-1413-330699242c05&t=c7d0fcb4-3a10-5b26-dfea-e879a93fc2ab

[maresb]

Hello @rmax, thanks for getting back to us! I would not expect matplotlib to install with fbprophet since AFAICT installing pyqt (the additional package that gets installed) serves no purpose and wastes a significant amount of space.

I got pystan to start using matplotlib-base with conda-forge/pystan-feedstock@72d8a88. The Azure pipeline which you linked installs version 2.17.1.0-py37hf2d7682_1004 of pystan. So I'm wondering why 2.19.1.1-py37hb3f55d8_1 doesn't get installed for @rfox12 or the Azure pipeline, even though it installs for me in the continuumio/miniconda3 Docker container?