Updating "JupyterLab Support" Docs + Prebuilt JupyterLab 3.0 extension

[layne-sadler]

Greetings. Is this section of the documentation still up-to-date?
https://github.com/plotly/plotly.py#jupyterlab-support

As you probably know, JupyterLab 3.0 allows for prebuilt extensions that can be pip installed. This removes the user-facing dependency on Node.js and skips the lengthy extension/ lab build process, which results in a much smoother experience. Additionally, you could specify ipywidgets as a pypi dependency of the extension's setup.py install_requires=[]. Making the installation process a one-liner.

Thank you.

[nicolaskruchten]

Yes, this section is up to date. We are aware of prebuilt extensions and will include them in our build process in the future, we just haven't gotten around to it yet. See #3036

[nicolaskruchten]

OK, if someone wants to kick the tires on it, pip install plotly==5.0.0rc1 should "just work" and install the prebuilt extension and it'll all just work in JLab3 regardless if you have our other extensions already installed.

[aiqc]

@nicolaskruchten Success! Thank you!

Upgrade workflow:

• Uninstalled plotly-charts or whatever the extension was called.
• Rebuilt jupyterlab env.
• As expected, plots failed to display.
• Installed plotly==5.0.0rc1.
• Plots failed to load.
• Restarted kernel.
• Plots failed to load.
• Restarted jupyterlab server. 👈
  • Do you know of a workaround without shutting down server?
• Plots succeeded.

[aiqc]

@nicolaskruchten would you guys be interested in demoing it at next month's Jupyter community call?

[nicolaskruchten]

Glad it's working well! I'd be happy to chat about our stuff any time but really in this particular case it's all @fcollonval's work here so maybe ask them? This was the (epic) PR that resolved a whole bunch of open issues in our extension all at once: #3142

[aiqc]

Ohhhh @fcollonval now I see what you were saying by "you did it." Ha.

[nicolaskruchten]

This is now released as part of v5.0.