I have tried working from #501 but can't make any progress.
I have Hydrogen 1.18.0 and want to be able to specify the conda environment that in which I launch my jupyter notebook.
From the command line, I'd do
source activate thisenv
jupyter notebook
How do I achieve the same thing from Hydrogen?
Many thanks
Guessing from #501 this should do the job:
source activate thisenv
python -m ipykernel install --user --name thisenv
It may also work if you start Atom from the command line after running source activate thisenv but then you can't switch between different envs in Atom.
Thank you. But where do I put this? It doesn't match the JSON which seems to be required for Kernel specs
It doesn't match the JSON which seems to be required for Kernel specs
As long as you have jupyter installed you don't need to manually add kernelspecs inside the Hydrogen settings, we automatically detect installed kernels. You can savely remove any JSON you added to the Hydrogen kernel specs settings.
You can run "Hydrogen: Update Kernels" or restart Atom for the changes to take effect. Next time you want to run Python code you'll be prompted to select the environment you just created.
One thing to add: You'll only need to run
source activate thisenv
python -m ipykernel install --user --name thisenv
once in the command line.
thank you very much! all works.
Awesome
I'm terribly sorry. Turns out I haven't installed ipykernel yet. 'pip install ipykernel' solves the problem and for anyone using anaconda with tensorflow in a separate env and just couldn't live without atom + hydrogen, be sure to check the above wonderful solution by @lgeiger
Thx for the answer but the second line wouldn't work, it says No module named ipykernel
My kernelspec is
{
"kernelspecs": {
"python2": {
"resource_dir": "/Users/spoc/anaconda2/share/jupyter/kernels/python2",
"spec": {
"language": "python",
"display_name": "Python 2",
"env": {},
"argv": [
"/Users/spoc/anaconda2/bin/python",
"-m",
"ipykernel_launcher",
"-f",
"{connection_file}"
]
}
}
}
}
Most helpful comment
Guessing from #501 this should do the job: