Fresh install of jupyterlab from conda-forge is version

JupyterLab
1

hi all
I am new to jupyterlab and installed minconda3 using

./Miniconda3-latest-Linux-x86_64.sh

then I ran

(test_cond) dave@deepthought:~$ conda install -c conda-forge jupyterlab

which installed the below which doesn’t seem to be the latest version ??? In my version Version 3.6.5 I can’t see the debugger

jupyter_client            7.4.9           py311h06a4308_0  
jupyter_core              5.3.0           py311h06a4308_0  
jupyter_events            0.6.3           py311h06a4308_0  
jupyter_server            1.23.4          py311h06a4308_0  
jupyter_server_fileid     0.9.0           py311h06a4308_0  
jupyter_server_ydoc       0.8.0           py311h06a4308_1  
jupyter_ydoc              0.2.4           py311h06a4308_0  
jupyterlab                3.6.5              pyhd8ed1ab_0    conda-forge
jupyterlab_pygments       0.1.2                      py_0  
jupyterlab_server         2.22.0          py311h06a4308_0

can someone help a noob please???

2

That looks like it’s trying to pull a lot of stuff from defaults (on anaconda.com), which is the (ahem) default behavior of Miniconda. Those packages are often out-of-date, and you it may other dependencies that are forcing something it into a local minimum. Also, for many use cases, using packages from there violates the Anaconda, Inc terms of service.

Further, doing interactive conda install is kinda peril-fraught.

I’d recommend starting with an my-jupyterlab-environment.yml that encapsulates your whole solve.

name: my-jupyterlab
channels:
  - conda-forge
  - nodefaults
dependencies:
  - jupyterlab ==4.*
  - python ==3.11.*

And then:

conda env update --file my-jupyterlab-environment.yml
conda activate my-jupyterlab
conda list

If that’s not working, one can try a clean slate Mambaforge (a spin of GitHub - conda-forge/miniforge: A conda-forge distribution.), which will already use anaconda.org/conda-forge as it’s default channel.

2 Likes
3

thank you so much for your advice and speedy reply. I’m certainly going to look at your yml solution BUT I think I’m headed down the route of purging the conda approach completely and just setting up a python environment and pip installing from there. To be honest I am happy to manage my own packages. I was just trying Miniconda because I followed someone else’s advice.

the example you kindly wrote is VERY helpful. Thank you for taking the time