Old version of jupyter-bokeh will be installed if jupyterlab>=4 is installed

[hoxbro]

It will download jupyter-bokeh 2.0.4. I don't know if it just updating the pins or if more work is needed.

Example:

❯ conda create -n tmp55 jupyterlab=4 jupyter_bokeh -c conda-forge --dry-run | grep jupyter
    - jupyter_bokeh
    - jupyterlab=4
    jupyter_bokeh-2.0.4        |     pyhd8ed1ab_0         1.6 MB  conda-forge
  jupyter-lsp        conda-forge/noarch::jupyter-lsp-2.2.0-pyhd8ed1ab_0 
  jupyter_bokeh      conda-forge/noarch::jupyter_bokeh-2.0.4-pyhd8ed1ab_0 
  jupyter_client     conda-forge/noarch::jupyter_client-8.2.0-pyhd8ed1ab_0 
  jupyter_core       conda-forge/linux-64::jupyter_core-5.3.1-py311h38be061_0 
  jupyter_events     conda-forge/noarch::jupyter_events-0.6.3-pyhd8ed1ab_0 
  jupyter_server     conda-forge/noarch::jupyter_server-2.6.0-pyhd8ed1ab_0 
  jupyter_server_te~ conda-forge/noarch::jupyter_server_terminals-0.4.4-pyhd8ed1ab_1 
  jupyterlab         conda-forge/noarch::jupyterlab-4.0.2-pyhd8ed1ab_0 
  jupyterlab_pygmen~ conda-forge/noarch::jupyterlab_pygments-0.2.2-pyhd8ed1ab_0 
  jupyterlab_server  conda-forge/noarch::jupyterlab_server-2.23.0-pyhd8ed1ab_0 
  jupyterlab_widgets conda-forge/noarch::jupyterlab_widgets-3.0.7-pyhd8ed1ab_1 

DryRunExit: Dry run. Exiting.

[mattpap]

This is because one major version of jupyter_bokeh was supposed to be compatible with one major version of jupyterlab, however, pre jupyter_bokeh 3.x there was no constraint on jypyterlab version. In fact no mention of jupyter at all. I will publish jupyter_bokeh 4.0.0 with updated dependency constraints.

[mattpap]

It doesn't build anymore:

Internal Error: @bokeh/jupyter_bokeh@workspace:.: This package doesn't seem to be present in your lockfile; run "yarn install" to update the lockfile

Probably easy to fix, just annoying.

[wieczyk]

How it is going? I also have jupyterlab 4.0.3 and jupyter tells me that jupyter_bokeh is outdated.

wieczyk@tortilla:~/Workspace/maxws/proj.e90$ jupyter labextension list
JupyterLab v4.0.3
/home/wieczyk/.pyenv/versions/3.10.6/share/jupyter/labextensions
        jupyterlab_pygments v0.2.2 enabled  X (python, jupyterlab_pygments)
        @jupyter-widgets/jupyterlab-manager v5.0.5 enabled  X (python, jupyterlab_widgets)


   The following extensions are outdated:
        jupyterlab_pygments
        @jupyter-widgets/jupyterlab-manager
        
   Consider checking if an update is available for these packages.

[philippjfr]

@mattpap You think we can just publish a version without the pin finally? While it won't build against JLab 4 I think an old build will likely still work with 4.

[mattpap]

I started PR #196 for this. Sadly interactivity doesn't work anymore with 4.x.

[itsgifnotjiff]

Can confirm the interactivity does not work with JupyterLab : 4.0.5 either (through JupyterHub) and hvplot : 0.8.4.