"Jupyter package is not installed" using system Python 3.8 and yes, it is installed
I'm on Arch Linux and using /usr/bin/python as the interpreter for this project. Yes I know most larger projects use virtualenvs, but for many smaller projects the system-installed Python has plenty of packages including all the main math/science Python stuff. For this I don't need more and would just rather use the system packages to avoid duplication.
So what's with this message? I can "import jupyter" in the project and I can run "jupyter notebook" from the command line.
I also tried running a separate Jupyter and configuring PyCharm to connect to it, but that message doesn't go away and code doesn't seem to run. What gives?
Thanks!
Please sign in to leave a comment.
Perhaps jupyter is installed with --user option? We have a known issue with it https://youtrack.jetbrains.com/issue/PY-38761
No, Jupyter is installed via the Arch Linux system package. /usr/bin/jpyter
Same with all other Python modules in use here.
I just tried on my Fedora laptop, also with Fedora Python 3.8 package and the Fedora Jupyter package. Same thing.
On Fedora I did manage to connect to a separately running Jupyter server, but I get a message "Notebook kernel doesn't match project interpreter" which is weird because they should be the same -- system Python in both cases.
Can you say what process PyCharm goes through to look for Jupyter? Why can it not find /usr/bin/jupyter?
>Can you say what process PyCharm goes through to look for Jupyter? Why can it not find /usr/bin/jupyter?
It's internal and I don't know much about it, unfortunately.
Would it be possible to uninstall jupyter and then install it with pip?
That wouldn't be desirable even if it worked. Using pip as root conflicts with the normal way to install system packages, and as you noted above, there's an issue with a --user install.
I think you guys need to take a good look at these related issues. Seems like a real bug. I see no reason why it shouldn't be able to find /usr/bin/jupyter.
I did manage to get it to connect to another running notebook server on both Fedora and Arch, but with both noted warnings on the screen, kind of annoying.
Guess in the meantime I could try with Conda if necessary; I would just rather avoid it since everything I need is already there in system packages.
I agree it's a bug. I suspect it has the same root cause as https://youtrack.jetbrains.com/issue/PY-34592
yeah the same, except I didn't see that particular message. Hopefully it can be fixed soon...
works in Conda, but would still like to see this fixed. Let me know if I can test anything. Thanks.
What is the work around. I can't use the product w/o this! I purchased specifically to work on notebooks
Bruce Bookman
Sorry for the issue. As a workaround, could you please use virtualenv/conda environment?