GitHub user sohil-chhabra added a comment to the discussion: Idea/Wish: geometric data from Deck.gl readable in a format other than GeoJSON -> XYZ vector tiles
Hi @herve-lg , Thank you for such a detailed and a thoughtful response! Regarding the limitations of `deck.gl Geojson` and how styling needs to work. Here's a proposal for styling needs (we've split it in terms of MVP and stretch/follow-up) : ### 1. Direct Styling in Superset UI (MVP) The idea is to provide native Superset Explore controls (Categorical schemes, Sequential scales, threshold breakpoints) where you pick a property from dropdown and pick a palette. Under the hood, this should compile directly into the MapLibre expression that you mentioned ### 2. In stretch goal we'll allow the user to directly paste the MapLibre expression ### Future goals As a future goal, it'll be awesome to have something like Geostyler like style editor as in [this PR ](https://github.com/apache/superset/pull/39760) by @jansule But I think this would be out of the scope Additionally, for semantic layers, here's an idea: ### 1. For the MVP (Tile-first / Decoupled): We want to allow users to connect directly to tile servers without forcing them to create a dummy SQL dataset in Superset. We’re planning to support auto-discovering layer properties via the tile server’s `TileJSON` endpoint so users can pick properties directly in Explore. Also, add tooltips to display feature properties on hover Although, they could be allowed to add a dataset optionally to populate property dropdowns (can be also a fallback to TileJSON ### 2. Data-Join Mode (The Semantic Layer Bridge): This will be exciting but I think might be a bit too much to initially support: combining static vector tile boundaries with live SQL metrics from the Superset semantic layer. Superset runs a lightweight aggregation query (e.g. `SELECT region_id, SUM(metric) GROUP BY region_id`), and [deck.gl](http://deck.gl/) joins the metrics onto the tile features in-browser on the key (`feature.properties.id === row.region_id`). This preserves full Superset dashboard filtering and cross-filtering while completely bypassing the 50MB+ GeoJSON serialization bottleneck. We would love to know if you have any feeback and again appreciate your insights! We'd love to get your thoughts once our initial PR is ready for testing too :)) GitHub link: https://github.com/apache/superset/discussions/43001#discussioncomment-19089267 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
