Closed akrherz closed 5 years ago
I'm on Linux and I just did:
conda create --name TEST python=3.7 numpy=1.16.3 --yes
conda activate TEST
python -c "import numpy as np; print(np.__version__)"
1.16.3
conda install numpy=1.16.4 --yes --quiet
python -c "import numpy as np; print(np.__version__)"
1.16.4
Same conda version and .condarc
as you :-/
Mixing defaults and conda-forge is not supported. (numpy from conda-forge and numpy-base from defaults)
The odd thing is how
numpy 1.16.4 py36h95a1406_0 conda-forge
numpy-base 1.14.3 py36h2b20989_0
happened with strict
enabled! That is a quite complex env with many PyPI packages. I wonder if conda has enough metadata there to do a proper upgrade.
Interesting, I able to remove numpy-base
and it did not seem to take the entire env with it :/
Collecting package metadata: done
Solving environment: done
## Package Plan ##
environment location: /opt/miniconda3/envs/prod
removed specs:
- numpy-base
The following packages will be downloaded:
package | build
---------------------------|-----------------
libopenblas-0.3.6 | h6e990d7_2 7.6 MB conda-forge
------------------------------------------------------------
Total: 7.6 MB
The following packages will be REMOVED:
numpy-base-1.14.3-py36h2b20989_0
The following packages will be UPDATED:
libopenblas pkgs/main::libopenblas-0.2.20-h9ac955~ --> conda-forge::libopenblas-0.3.6-h6e990d7_2
Proceed ([y]/n)? y
Downloading and Extracting Packages
libopenblas-0.3.6 | 7.6 MB | ########################################################################################### | 100%
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
now numpy=1.16.4
is OK. I recall attempting to remove numpy-base
many moons ago and it was not pleasant. Will see how long it can be kept away now. thanks for the help here!
Will see how long it can be kept away now. thanks for the help here!
As long as you start the env with strict
numpy-base
should never be installed unless explicitly requested. (But it will be incompatible with conda-forge
's numpy
.)
Issue: A currently working numpy=1.16.3 install breaks after updating to numpy=1.16.4
Environment (
conda list
):Details about
conda
and system (conda info
):