Currently, a week ago, on December 23, 2020, the Jupyter Notebook was updated to version 6.1.6, but the extension nbextensions (0.5.1) is no longer available. It seems that. Furthermore, if you simply drop the version of Jupyter Notebook to 6.1.5, you will be able to use nbextensions, but you will not be able to use the autocomplete extension Hinterland. I'll leave a solution for this in my case.
| Remarks | ||
|---|---|---|
| OS | Windows10 | |
| conda | 4.9.2 | With Anaconda Promptconda -V |
| Python | 3.9.1 |
Create a new environment nbex_test with Anaconda Prompt as follows and install jupyter and nbextensions.
(base) $ conda create -n nbex_test
(base) $ conda activate nbex_test
(nbex_test) $ conda install jupyter
(nbex_test) $ conda install -c conda-forge jupyter_contrib_nbextensions
In this state, a shortcut for jupyter notebook (nbex_test) will be created in C: \ Users \ Username \ AppData \ Roaming \ Microsoft \ Windows \ Start Menu \ Programs \ Anaconda3 (64-bit) in the Start Menu. .. However, when I launch jupyter notebook from this shortcut, the Nbextensions tab does not appear as shown in the figure below.

I lowered the version of Jupyter Notebook to 6.1.5 and it worked. You can lower the version of Jupyter Notebook to 6.1.5 by re-installing with Anaconda Prompt by specifying the version as shown below.
(nbex_test) $ conda install notebook==6.1.5
Safely, the nbextensions tab appears as shown in the figure below. Now you can use extensions other than autocomplete.

However, with this alone, you cannot use only the autocomplete extension. If you turn on Hinterland from the nbextensions tab, it will suggest import when you enter imp as shown in the figure below. Not only the original syntax of Python, but also the auto-complete function works according to the module installed later. But this cannot be used.

It worked fine when I lowered the version of the autocomplete Python module jedi that comes with Jupyter. The currently installed version of jedi was 0.18.0, so I lowered it to 0.17.2.
(nbex_test ) $ conda install jedi==0.17.2
I knew that if I dropped the version, it would be usable, but I'm in trouble because I don't know which module should be dropped to which version ... Or rather, Anaconda is the one that solves the dependency well. I hope that no more people will suffer from this anyway. (I was eaten for half a day)
Recommended Posts