Closed fabboe closed 7 years ago
Is ipykernel
package installed in the py35data environment?
It is, as
ipykernel 4.6.1 py35_0
According to my tests that env wouldn't even be available as a kernel if ipykernel wouldn't be there. Actually I figured this issue out during those tests already. I had an env created without ipykernel and nb_conda didn't pick it up. Adding ipykernel to the env afterwards doesn't help (same issue as described above I guess).
I just confirmed with fiona but maybe found a hint to a different problem. There seems to be confusion of root and env conda manager. Is this in the end a problem of conda / conda-env?
Note that it's not pointing to the env in the print statements.
➜ ~ sudo miniconda2/envs/py35data/bin/conda install fiona
Fetching package metadata .........
Solving package specifications: .
Package plan for installation in environment /Users/fabo/miniconda2:
The following NEW packages will be INSTALLED:
~ sudo miniconda2/envs/py35data/bin/conda list
# packages in environment at /Users/fabo/miniconda2:
#
...
fiona 1.7.0 py27_0
And indeed I can find fiona files only in the root env, not in the py35data env.
Do you confirm that's conda / conda-env problem? Maybe you still have a hint where I should go to debug further. We can close this one probably.
Thanks!
Closing this. Reproduced in an env with no relation to nb_conda whatsoever. Issue with conda/$PATH it seems.
Conda expects a -n or -p option to specify the install env. It's not based on the location of the conda executable you run.
That's it, thanks! I was sure that I had used the other behaviour in past versions successfully before.
Hi all,
nb_conda is a great idea, thanks for that. I'm having a strange issue on a very fresh system.
After
The respective kernel of nb_conda does not have the package available. It also does not show it in the list of the conda tab. Same holds after restarting even the whole notebook server.
Whereas after
I'm able to import the package even in the running kernel of py35data. Though the nb_conda tab still doesn't show it.
The kernel definition (which nb_conda generated?) in miniconda2/envs/py35data/share/jupyter/kernels/python3/kernel.json seems pretty generic:
Am I expecting too much here from nb_conda? Why does it pick it up from pip and not from conda?
Thanks for reading, Fabian