PyData Global 2021 presentation

[jtpio]

@madhur-tandon @martinRenou @bollwyvl

Here is an initial draft for the presentation: https://github.com/jtpio/pydata-global-2021/blob/main/content/presentation.ipynb

We can iterate here in this repo to add / update content.

We could then use the RetroLab interface during the presentation: https://jtpio.github.io/pydata-global-2021/retro/notebooks/?path=presentation.ipynb

Then we can either:

1. do the presentation from this repo
2. upload the main presentation.ipynb notebook to https://jupyterlite.github.io/demo/ and use the examples from the already existing demo deployment

Feel free to add / edit content.

[jtpio]

2. upload the main presentation.ipynb notebook to https://jupyterlite.github.io/demo/ and use the examples from the already existing demo deployment

Thinking about a bit more, this would make more sense than having it under my github account.

[jtpio]

Or we can also move this pydata-global-2021 repo to the jupyterlite organization.

[martinRenou]

Maybe it makes sense to make it part of the jupyterlite main demo? This way, we use it as an opportunity to fix https://github.com/jupyterlite/jupyterlite/issues/168. And it's also a way to make the demo folder more appealing/more complete.

[jtpio]

Yeah I like the idea of being able to improve what we already have.

The main concern was to not generate too much noise as we iterate on the content of the presentation.

[martinRenou]

Maybe we can iterate here separately from jupyterlite core, then we push a single PR updating everything at once?

[jtpio]

Sounds good :+1:

[bollwyvl]

On the wrong box to PR, but some thoughts:

• lead with the picture, quick walk through it up-front

  • may try and do an update, couldn't edit it last time i tried...

• Then do a quick thing about "the clients"...

## JupyterLite and RetroLite

- full (Jupyter|Retro)Lab with some overloaded/disabled parts
- compatible with _most_ pre-built extensions
- future work to include "single cell" and/or "single document" apps

• The server...

## The Jupyter(Lite) Server(Lite)

- provides in-browser equivalents of much of `jupyter(lab)_server`
- plugin-based Lumino application, _a la_ JupyterLab
- customizable with pre-built _server extensions_
  - new kernels
  - different contents managers
  - remote kernels

• ...Use Cases

## Companion/Power User Mode

- next to an API, serve a computable Lite site
- offer power users a "compute this data"
- self-contained deployment of known extensions, libraries without CDN
- interactions between the "Lite" site and "parent" application
  - pre-fill/wrap API artifacts as contents
  - inject generated code snippets into notebooks
  - expose the "parent" app within kernels

[DerThorsten]

xeus_lite_figures.zip here is a zip with the svg / gifs from the xeus-lite blogpost

[jtpio]

Thanks @DerThorsten. I invited you to the repo, if you want to edit the presentation later.

[DerThorsten]

@jtpio any preferred way of hosting the above images st we can integrate them in the presentation?

[jtpio]

Usually dropping them in a GitHub issue to generate a link works fine.

[jtpio]

Also thinking about the presentation, maybe we should do it in lab instead of retro.

In lab it will be easier to open the example notebooks in the main area. Which would be less disruptive than switching browser tabs. Also since we are more talking about the JupyterLite interface and tools than the actual content of the notebook. For example we can demo running multiple kernels on the same page with the JupyterLab interface.

Retro would have been nice because it's simpler. We could also make use of the simple interface for increased focus if we want.

[jtpio]

Also thinking about the presentation, maybe we should do it in lab instead of retro.

Or we can do the general intro and showcasing the JupyterLite features with the JupyterLab interface.

Those demoing the Python kernel, widgets and xeus-based kernels can choose do it in Retro if they want (for example after preloading the demo notebooks in multiple browser tabs)

[jtpio]

Thanks @bollwyvl for the input :+1: #8 implemented some of the suggestions

btw @bollwyvl do you plan to join the conference? Your invite was still pending, but we can still email the organizers to manually add you if needed.

For now we are planning to be 4 people to present:

• Jeremy: intro, tour of the JupyterLite main features. plugin-based server a la JupyterLab
• Madhur: the Python kernel, IPython, ?
• Martin: Widgets, Interactive Visualization, ?
• Thorsten: Xeus-based kernels

Hopefully switching the screen sharing will be smooth!

[jtpio]

Opened https://github.com/jupyterlite/demo/pull/37 to start updating the demo notebooks.

Maybe we don't have to add the presentation notebook to https://github.com/jupyterlite/demo (so it doesn't get out of date later). Each of us could upload it in their own browser.

[bollwyvl]

I didn't really line up time to do any conference stuff, and am sure you folks will do great! I'm not too worried about the byline. And whew, that's a lot of mic passing... seems like one more could be the last straw! Good luck!

[jtpio]

Everything went well, thanks everyone!!

The notebooks in https://github.com/jupyterlite/demo have been updated. So we can archive this repo. The presentation notebook will still be around but won't be updated. In case someone would like to reuse it for a future talk.

[bollwyvl]

Huzzah! Looking forward to seeing it! Any feedback that we could to hoist to issues/discussions?

[jtpio]

Yes this one when we were doing the dry-run :laughing: https://github.com/jupyterlite/jupyterlite/issues/406