Open jlumpe opened 2 months ago
Thank you for reporting this @jlumpe .
@LiamConnors FYI.
couldnt we turn this into 'make plotly installation more foolproof'? I'm fairly competent with py environments and still struggle.
There's tons of stack overflow (TONS!) advice that people run into (e.g. offline mode). I'm running freshly upgraded versions , installed plugin 'serverside' etc, try out all kinds of modes, it's just not straightforward right now (not venting, sharing experiences with the current chosen path)
[EDIT] just seeing this issue is targeting v3
The "Getting Started" page has a "JupyterLab Support" support section which recommends using pip or conda to install the
jupyterlab
andipywidgets
packages only. Further down it mentions that for Jupyter Lab 2.x only you need to install some extensions manually usingjupyter labextension install
.However, information in the Troubleshooting in Python page is somewhat contradictory and shows that you do in fact need the
jupyterlab-plotly
extension to be installed in version 3 (and presumably 4):It seems to me that many people are installing the
jupyterlab-plotly
extension more or less accidentally because the kernel they use theplotly
package with happens to be running in the same environment that the JupyterLab server/application is. However there are several issues filed here where people who are using separate environments are running into problems and it's not completely clear what the resolution is. In many cases they are trying to resolve them withjupyter labextension install ...
, which is deprecated and often doesn't work properly in v3.I think the documentation on the main page could be clearer that you do in fact need the extension with JupyterLab v3, and that you should make sure to install the
plotly
package in the same environment you're running JupyterLab in even if you're using a different environment for the kernel.