ipympl conda install deletes env's default labextensions state (but works with pip)

Describe the issue

Installing ipympl with conda overwrites the content of .../conda/envs/env_name/etc/jupyter/labconfig/page_config.json resulting in enabling of all labextensions for all users (which is a big problem on a JupyterHub).

Install command:

conda install -c conda-forge ipympl

.../conda/envs/env_name/etc/jupyter/labconfig/page_config.json before ipympl install:

{
  "disabledExtensions": {
    "@jupyter/nbgrader:formgrader": true,
    "@jupyter/nbgrader:course-list": true,
    "@jupyter/nbgrader:create-assignment": true,
    "kore-extension": true
  },
  "lockedExtensions": {
    "@jupyter/nbgrader:formgrader": true,
    "@jupyter/nbgrader:course-list": true,
    "@jupyter/nbgrader:create-assignment": true,
    "kore-extension": true
  }
}

and after:

{
  "lockedExtensions": {}
}

This is not caused by a dependency: installing all dependencies first shows that install of ipympl overwrites the file.

Using pip install ipympl everything works as expected (page_config.json remains untouched).

Versions

• ipympl 0.9.4
• jupyterlab 4.2.4
• jupyterhub 5.1.0

# without output from jupyter nbextension list (not installed)

 3.12.5 | packaged by conda-forge | (main, Aug  8 2024, 18:36:51) [GCC 12.4.0]
ipympl version: 0.9.4
Selected Jupyter core packages...
IPython          : 8.26.0
ipykernel        : 6.29.5
ipywidgets       : 8.1.5
jupyter_client   : 8.6.2
jupyter_core     : 5.7.2
jupyter_server   : 2.14.2
jupyterlab       : 4.2.4
nbclient         : 0.10.0
nbconvert        : 7.16.4
nbformat         : 5.10.4
notebook         : 7.2.1
qtconsole        : not installed
traitlets        : 5.14.3
JupyterLab v4.2.4
/opt/conda/envs/jhub/share/jupyter/labextensions
        jupyterlab-execute-time v3.1.2 enabled OK (python, jupyterlab_execute_time)
        jupyterlab_pygments v0.3.0 enabled OK (python, jupyterlab_pygments)
        jupyter-matplotlib v0.11.4 enabled OK
        kore-extension v0.1.0 enabled OK (python, kore_extension)
        @jupyter-notebook/lab-extension v7.2.1 enabled OK
        @jupyter-widgets/jupyterlab-manager v5.0.13 enabled OK (python, jupyterlab_widgets)
        @jupyter/nbgrader v0.9.3 enabled OK

Reproduisez le problème avec conda install -c conda-forge ipympl et comparez .../conda/envs/env_name/etc/jupyter/labconfig/page_config.json avant et après l’installation. Utilisez le comportement signalé de pip comme comparaison ; le travail est terminé lorsque les paramètres existants disabledExtensions et lockedExtensions restent intacts après l’installation de conda.