Closed orboan closed 2 years ago
By the way I need to confirm that the installation of scijava is working perfectly fine in a conda single user installation, that I had before, in the same machine (Centos 7), where in $HOME/anaconda3/envs/scijava/share/jupyter/kernels/ it showed the python3 and also the scijava folders.
All the scijava kernel jars are located in: /opt/cloudera/parcels/Anaconda/envs/scijava/opt/scijava-jupyter-kernel/ and they look that they are properly referenced in /opt/cloudera/parcels/Anaconda/envs/scijava/conda-meta/scijava-jupyter-kernel-0.6.1-0.json
Found that as a root I get:
# jupyter kernelspec list
Available kernels:
scijava /root/.local/share/jupyter/kernels/scijava
python3 /opt/cloudera/parcels/Anaconda/envs/scijava/share/jupyter/kernels/python3
but as a regular user (in the scijava env):
$ jupyter kernelspec list
Available kernels:
python3 /opt/cloudera/parcels/Anaconda/envs/scijava/share/jupyter/kernels/python3
This looks like the scijava kernel cannot be installed in multiuser mode (conda installation mode that uses the root user), and instead of being installed in multiuser mode in the /opt directory, it installs as single user mode taking root as the user, which is very bad.
@orboan thanks for reporting!
Note that the ImageJ tutorial notebooks use a native beakerx
Groovy kernel since a few years. I am not sure how well this component here is maintained still...
@orboan can beakerx
be an alternative for you?
I have conda installed in multiuser mode in the /opt directory, as a parcel installation done with the cloudera manager (CDH 6.3). Running 'conda install -c conda-forge scijava-jupyter-kernel' shows no error and it should be fine. But when I start a jupyter or jupyterlab, the scijava kernel is not shown. In /opt/cloudera/parcels/Anaconda/envs/scijava/share/jupyter/kernels/ there is only the python3 kernel, thus the scijava kernel directory with its json file is missing. Found this json in another path: /opt/cloudera/parcels/Anaconda/envs/scijava/conda-meta/scijava-jupyter-kernel-0.6.1-0.json but making a symlink or a copy from this to the previous directory (where the python3 kernel json is located) makes the jupyter launcher tab appear totally blank and the regular jupyter notebook shows a "kernel error". So what is the problem here? Kind regards