Cartopy: Conda installation requires DOWNgrading of dependencies
Hi,
I just installed carroty using condo. When doing so, condo forces me to actually DOWNgrade some of the dependencies as follows:
The following packages will be DOWNGRADED due to dependency conflicts:
matplotlib: 2.0.0-np111py27_0 --> 1.5.1-np110py27_0
numpy: 1.11.3-py27_0 --> 1.10.2-py27_0
pandas: 0.19.2-np111py27_1 --> 0.18.1-np110py27_0
pyqt: 5.6.0-py27_2 --> 4.11.4-py27_4
qt: 5.6.2-0 --> 4.8.7-4
qtconsole: 4.2.1-py27_1 --> 4.2.1-py27_0
scipy: 0.18.1-np111py27_1 --> 0.16.0-np110py27_1
Is this really required or is it just that the condo package build recipe is not specific enough regarding the dependencies?
The carroty installation documentation does not say that there is a need for a specific version of e.g. spicy. It just says spicy > X.XX is required.
However, it seems that only v0.16 is currently supported for scipy.
It would be helpful to check if carroty can also work with the newer versions of these dependencies and if so, change the condo packaging accordingly.
Thanks!
bulli92
All 10 comments
That's a weird set of downgrades. Can you paste in the actual line you were using to install/upgrade cartopy? Also, can you post the output of conda config --show?
dopplershift
on 6 Feb 2017
It's the same for me on Python 3.5:
The following packages will be DOWNGRADED due to dependency conflicts:
matplotlib: 2.0.0-np111py35_0 --> 1.5.1-np110py35_0
mkl: 2017.0.1-0 --> 11.3.3-0
numpy: 1.11.3-py35_0 --> 1.10.4-py35_2
pandas: 0.19.2-np111py35_1 --> 0.18.1-np110py35_0
pyqt: 5.6.0-py35_2 --> 4.11.4-py35_4
qt: 5.6.2-3 --> 4.8.7-4
qtconsole: 4.2.1-py35_1 --> 4.2.1-py35_0
scipy: 0.18.1-np111py35_1 --> 0.17.1-np110py35_1
I use the conda install -c scitools cartopy command to install cartopy.
The output of conda config --show is
add_anaconda_token: True
add_pip_as_python_dependency: True
allow_softlinks: True
always_copy: False
always_softlink: False
always_yes: False
auto_update_conda: True
binstar_upload: None
changeps1: True
channel_alias: https://conda.anaconda.org
channel_priority: True
channels:
- defaults
client_ssl_cert:
client_ssl_cert_key:
create_default_packages: []
debug: False
default_channels:
- https://repo.continuum.io/pkgs/free
- https://repo.continuum.io/pkgs/r
- https://repo.continuum.io/pkgs/pro
disallow: []
envs_dirs:
- /home/elrond16/miniconda3/envs
json: False
offline: False
proxy_servers: {}
quiet: False
shortcuts: True
show_channel_urls: None
ssl_verify: True
track_features: []
update_dependencies: True
use_pip: True
verbosity: 0
agolovanov
on 10 Feb 2017
You might consider adding the conda-forge channel as a place to get more packages.
dopplershift
on 10 Feb 2017
I have a similar problem, with a different set of requested downgrades, when trying to install into uvcdat 2.8.0
I have created a _uvcdat-2.8.0_ environment using the following steps: https://wiki.lsce.ipsl.fr/pmip3/doku.php/other:uvcdat:cdat_conda#installation_with_miniconda2
And then I have cloned this into a _cdatm14_ environment using: https://wiki.lsce.ipsl.fr/pmip3/doku.php/other:uvcdat:cdat_conda#cloning_uvcdat_to_add_specific_packages_for_lsce
If I then try to install cartopy with
conda install -n cdatm14 -c conda-forge -c uvcdat -c scitools cartopy
I get the following set of packages and downgrades
The following NEW packages will be INSTALLED:
cartopy: 0.15.0-np111py27_0 conda-forge
libxslt: 1.1.29-3 conda-forge
lxml: 3.7.2-py27_1 conda-forge
olefile: 0.44-py27_0 conda-forge
owslib: 0.14.0-py27_0 conda-forge
pillow: 4.0.0-py27_2 conda-forge
pyepsg: 0.3.1-py27_0 conda-forge
shapely: 1.5.17-np111py27_2 conda-forge
The following packages will be DOWNGRADED due to dependency conflicts:
geos: 3.6.0-0 conda-forge --> 3.5.1-1 conda-forge
matplotlib: 2.0.0-np111py27_2 conda-forge --> 1.5.3-np111py27_7 conda-forge
It gets even worse if I add the scitools channel
conda install -n cdatm14 -c scitools -c conda-forge -c uvcdat cartopy
The following NEW packages will be INSTALLED:
cartopy: 0.15.0-np110py27_0 scitools
funcsigs: 1.0.2-py27_0 conda-forge
libxslt: 1.1.29-3 conda-forge
lxml: 3.7.2-py27_1 conda-forge
mock: 2.0.0-py27_0 conda-forge
nose: 1.3.7-py27_2 conda-forge
olefile: 0.42.1-py27_1 scitools
openblas: 0.2.14-4
owslib: 0.10.3-py27_0 scitools
pbr: 1.10.0-py27_0 conda-forge
pillow: 4.0.0-py27_1 conda-forge
proj.4: 4.9.1-1 scitools
pyepsg: 0.2.0-py27_0 scitools
shapely: 1.5.13-np110py27_1 scitools
The following packages will be UPDATED:
asciidata: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
basemap: 1.0.8.dev0-np111py27_1 conda-forge --> 1.0.8.dev0-np110py27_1 conda-forge
binaryio: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
cdms2: 2.8.1-np111py27_0 uvcdat --> 2.8.1-np110py27_0 uvcdat
cdtime: 2.8-np111py27_0 conda-forge --> 2.8-np110py27_2 uvcdat
cssgrid: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
dsgrid: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
esmf: ESMF_6_3_0rp1_ESMP_01-np111py27_2 uvcdat --> ESMF_6_3_0rp1_ESMP_01-np110py27_2 uvcdat
genutil: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
hdf5tools: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
lmoments: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
natgrid: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
ort: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
regridpack: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
shgrid: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
trends: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
unidata: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
uvcmetrics: 2.2-np111py27_2 uvcdat --> 2.2-np110py27_2 uvcdat
zonalmeans: 2.8-np111py27_2 uvcdat --> 2.8-np110py27_2 uvcdat
The following packages will be DOWNGRADED due to dependency conflicts:
fontconfig: 2.12.1-4 conda-forge --> 2.12.1-1 conda-forge
freetype: 2.7-1 conda-forge --> 2.6.3-1 conda-forge
matplotlib: 2.0.0-np111py27_2 conda-forge --> 1.5.3-np110py27_5 conda-forge
numpy: 1.11.3-py27_0 --> 1.10.2-py27_0
scikit-learn: 0.18.1-np111py27_1 --> 0.17-np110py27_1
scipy: 0.18.1-np111py27_1 --> 0.16.0-np110py27_1
It's also strange that in the second case (with _-c scitools_ at the front of the _-c_ options) I'm told that proj.4 4.9.1-1 will be installed, when I already have proj4 4.9.3 installed
I confess I'm quite new to using conda
It seems I could avoid the downgrades by using --no-update-dependencies, but this would get me an older version of cartopy (0.14 instead of 0.15) as well as extra packages (funcsigs, mock, nose, pbr)...
(uvcdat-2.8.0) bash-4.1$ conda install -n cdatm14 --no-update-dependencies -c conda-forge -c uvcdat cartopy
Fetching package metadata .............
Solving package specifications: .
Package plan for installation in environment /home/share/unix_files/cdat/miniconda2/envs/cdatm14:
The following NEW packages will be INSTALLED:
cartopy: 0.14.3-np111py27_3 conda-forge
funcsigs: 1.0.2-py27_0 conda-forge
libxslt: 1.1.29-3 conda-forge
lxml: 3.7.2-py27_1 conda-forge
mock: 2.0.0-py27_0 conda-forge
nose: 1.3.7-py27_2 conda-forge
olefile: 0.44-py27_0 conda-forge
owslib: 0.14.0-py27_0 conda-forge
pbr: 1.10.0-py27_0 conda-forge
pillow: 4.0.0-py27_2 conda-forge
pyepsg: 0.3.1-py27_0 conda-forge
shapely: 1.5.16-np111py27_0 conda-forge
Proceed ([y]/n)?
jypeter
on 23 Feb 2017
I have had a similar problem with shapely that wanted the same downgrade of _geos_ (https://github.com/Toblerity/Shapely/issues/472). Unfortunately, now that I have installed shapely (and some other dependencies), conda still wants me to downgrade geos
bash-4.1$ conda install -n cdatm14 -c conda-forge cartopy
Fetching package metadata ...........
Solving package specifications: .
Package plan for installation in environment /home/share/unix_files/cdat/miniconda2/envs/cdatm14:
The following NEW packages will be INSTALLED:
cartopy: 0.15.0-np111py27_0 conda-forge
libxslt: 1.1.29-3 conda-forge
lxml: 3.7.2-py27_1 conda-forge
owslib: 0.14.0-py27_0 conda-forge
pyepsg: 0.3.1-py27_0 conda-forge
The following packages will be DOWNGRADED due to dependency conflicts:
geos: 3.6.0-0 conda-forge --> 3.5.1-1 conda-forge
matplotlib: 2.0.0-np111py27_2 conda-forge --> 1.5.3-np111py27_7 conda-forge
Proceed ([y]/n)? n
Both the cartopy and shapely packages on conda-forge are pinned to version 3.5 of geos, so they are saying they won't work with 3.6, which is why conda tries to downgrade.
ajdawson
on 24 Feb 2017
o.k., thanks, this explains the issue.
Thus the question would be more to the cartopy and shapely maintainers if geos3.5.x is a strong constraint or if these packages would also work with higher versions.
bulli92
on 24 Feb 2017
I am getting the same issue. Almost every time I try to install or update anything I get:
The following packages will be DOWNGRADED:
blas: 1.1-openblas conda-forge --> 1.0-mkl
numpy: 1.15.1-py36_blas_openblashd3ea46f_1 conda-forge [blas_openblas] --> 1.14.3-py36h9bb19eb_1
scikit-learn: 0.19.2-py36_blas_openblasha84fab4_201 conda-forge [blas_openblas] --> 0.19.1-py36hffbff8c_0
scipy: 1.1.0-py36_blas_openblash7943236_201 conda-forge [blas_openblas] --> 1.1.0-py36hcaad992_0
Proceed ([y]/n)?
Most recently this occurred during the following commands:
conda update bokehconda install -c conda-forge pscriptconda update statsmodels
Each time I let it downgrade them, then simply update them again:
$ conda update numpy blas scipy scikit-learn
...
The following packages will be UPDATED:
blas: 1.0-mkl --> 1.1-openblas conda-forge
numpy: 1.14.3-py36h9bb19eb_1 --> 1.15.2-py36_blas_openblashd3ea46f_0 conda-forge [blas_openblas]
numpy-base: 1.14.3-py36h479e554_1 --> 1.14.3-py36h7ef55bc_1
scikit-learn: 0.19.1-py36hffbff8c_0 --> 0.19.2-py36_blas_openblasha84fab4_201 conda-forge [blas_openblas]
scipy: 1.1.0-py36hcaad992_0 --> 1.1.0-py36_blas_openblash7943236_201 conda-forge [blas_openblas]
Proceed ([y]/n)?
This doesn't seem to cause any problems (yet) but is it good practice and is this downgrading/re-upgrading really necessary?
B.t.w. I'm in an environment that has Python 3.6 installed:
(py36) MyMacBookPro:~ username$ which python
/Users/billtubbs/anaconda/envs/py36/bin/python
billtubbs
on 25 Sep 2018
That has nothing to do with Cartopy; it is a problem with conda or whoever makes those packages.
QuLogic
on 25 Sep 2018
That has nothing to do with Cartopy; it is a problem with conda or whoever makes those packages.
Sorry everyone, just realised this is a Cartopy issue list. (Thought it was Anaconda). Please ignore, I will repost.
billtubbs
on 25 Sep 2018
Related issues
leecronce
·
8Comments
maxnoe
·
5Comments
choldgraf
·
3Comments
lvphj
·
8Comments
rutgerhofste
·
3Comments
Most helpful comment
Both the cartopy and shapely packages on conda-forge are pinned to version 3.5 of geos, so they are saying they won't work with 3.6, which is why conda tries to downgrade.