Closed joequant closed 4 years ago
Hey @joequant
The main hypothesis is that you are using/must use conda. So to achieve this, you need to install the base conda environment at the system level. And in that environment you will install jupyterlab or notebook, nb-conda-kernels and this extension.
The tricky part is to enable the base conda environment for all users in their terminal. For that you could take a look at the miniconda dockerfile.
The trick lies in https://github.com/ContinuumIO/docker-images/blob/master/miniconda3/debian/Dockerfile#L14
where you link the conda init script to /etc/profile.d
so it gets loaded for all users:
ln -s /opt/conda/etc/profile.d/conda.sh /etc/profile.d/conda.sh
The last trick you will need is to activate the base environment for each user (conda activate base
).
Note: nb_conda_kernels is only available as conda package. So
pip
will never find it.
What i would like to do is to install jupyter_conda at the system level to allow users to set up their own local conda environments. however when I try to install conda with pip i get
[joe@mcdull extensions]$ pip3 install nb-conda-kernels ERROR: Could not find a version that satisfies the requirement nb-conda-kernels (from versions: none) ERROR: No matching distribution found for nb-conda-kernels
Any pointers for allowing me to do something like this?