Dashboard Compiler & Publisher

[domoritz]

Given a Mosaic YAML/JSON specification (and possibly additional data source information), analyze the spec and produce a standalone dashboard web application. The compiler should analyze the data sources and convert them for rapid loading within a WASM-based DuckDB deployment. The compiler should also raise warnings or exit with an error if the dataset is too big to export to a standalone web-based deployment. The compilation process might also consider data cube indexing, for example by precomputing cube tables and adding them to DuckDB upon dashboard initialization. The compiler might also suggest dashboard modifications when helpful, for example by rewriting a selection to use single resolution if cube computation will be too costly at runtime. A dashboard publishing tool like Observable Framework might be one way to then deploy the results.

[jheer]

I've started some initial work on this, so if others are interested in pitching in please get in touch. Or, just "thumbs up" vote on the entry above if this is something you'd like to see supported.

[nicosuave]

I'm interested in helping

[alexkreidler]

I built a tool to edit and live-preview Mosaic specs in your browser: https://mosaic-playground.netlify.app/. It doesn't generate a static site, but I've been using it to create Mosaic JSON specs that I then put in an Observable Framework markdown file.

I'd love some feedback and would also be open to contribute it to Mosaic: https://github.com/alexkreidler/mosaic-playground

It would be awesome to have a way to pre-bake data for initial views (#506), or an option to render Mosaic to a non-interactive static SVG that can be used in a static site generator like in this Observable Framework example.

Thank you for your awesome work! I'm excited to see what everyone is cooking up for this issue!

[jheer]

Wow, this looks great @alexkreidler! Thanks for developing and sharing this.

I'd definitely be interested in folding some variant of this into the Mosaic project proper (e.g., as a new package within the monorepo). In terms of constructive feedback, it would be nice to include support for YAML and for JSON schema auto-complete. I also notice a rather aggressive stream of red error notifications in the midst of normal typing in the editor.

FWIW, I also plan to further develop both pre-baked data and server-side SVG rendering, but need to clear a number of other obligations off my plate before diving into those. In any case, thanks again for sharing!