Desktop 2.0 Usecase

[vacquah]

With desktop 2.0, What is the intended workflow? I assume the idea is to use it to setup your visualizations and push to a static site for sharing? Coming from another notebook ( jetbrains datalore ), I am missing a few things like:

• scheduling notebook data refresh
• create reports using the visuals created ( no code cells )

Does the observable 2.0 notebook fit with a streaming data use case ? Live/scheduled updates of reports/dashboards ?

[mbostock]

In the future — hopefully later this year — we plan on releasing a web editor for Notebooks 2.0 similar to the Desktop editor, but in the browser. This will allow realtime collaboration on notebooks similar to the current 1.0 editor. It will also allow sharing by building static versions of notebooks, similar to the notebooks build command. The web platform will likely also offer support for live data and automatic refreshing, though that may come later.

In the meantime, there are currently two primary workflows.

(1) If you want to collaborate on (coauthor) notebooks, you check your notebooks into source control (such as git and GitHub), and then have your collaborators clone the repository and commit or propose changes to the notebooks. We recommend using Observable Desktop to edit notebooks, but you can also use your preferred code/text editor and use Notebook Kit for a live preview.

(2) If you want to share (or “publish”) notebooks, you can use Notebook Kit’s notebook build command to build a static site. You can then share that static site on your preferred static hosting service, such as GitHub Pages. This command-line interface is designed and intended to work with continuous integration systems such as GitHub Actions if you want to update data on a schedule. You can see an example of automating builds in this repository’s deploy.yml workflow; we use this technique to update the Notebooks documentation.

By default, any SQL cells and data loaders cells fetch fresh data when building with Notebook Kit in CI. However, you can preserve the .observable/cache directory if you want to cache data across builds. We haven’t documented this yet, but the technique is very similar to caching for Observable Framework. (Though note that Notebook Kit uses .observable/cache rather than .observablehq/cache…)

If you want to hide code cells in your shared notebooks, simply unpin them. If you need to disable the implicit display (such as for SQL cells or data loader cells), you can also mark the cells as hidden; there isn’t a toolbar icon for hiding cells yet, but you can use the Edit menu or the Command-, keyboard shortcut.