What problem are you trying to solve?
DAC has no choropleth / geographic-map widget, and there's no escape hatch to add one in user code.
- JSX loader (
pkg/dashboard/jsloader.go:654) passes unknown tags through, but BruinWidgetFrame.tsx only renders metric / chart / table / text / divider / image — unknown types render blank.
- TSX dashboards aren't a normal Vite/React build, so
react-leaflet / maplibre-gl / deck.gl can't be imported.
<Image> is <img src> only; <Text> strips raw HTML.
Today we either ship non-interactive PNG choropleths via <Image>, build a parallel Streamlit dashboard, or fork DAC.
Proposed solution
First-class <Map> widget:
<Map
geoUrl="https://gisco-services.ec.europa.eu/.../NUTS_RG_03M_2024_4326_LEVL_3.json"
regionColumn="nuts_id"
valueColumn="exceedances"
colorScale="viridis"
sql="SELECT nuts_id, exceedances FROM ..."
/>
Hover tooltip, zoom/pan, <Filter> can drive valueColumn.
Alternatives considered
- 2D
chart: heatmap on binned lat/lon — pixel grid, not a map.
- Pre-rendered PNGs in
<Image> — no interactivity.
- Streamlit alongside DAC — splits the dashboard layer.
- Fork DAC with a custom
MapWidget — works, but per-user install + upstream drift.
Additional context
Geographic views are high-leverage for any regional dataset (EU NUTS, US counties, country comparisons). Current PFAS investigation has 1,165 NUTS3 with metrics where the regional pattern is the story — bar charts hide it.
What problem are you trying to solve?
DAC has no choropleth / geographic-map widget, and there's no escape hatch to add one in user code.
pkg/dashboard/jsloader.go:654) passes unknown tags through, butBruinWidgetFrame.tsxonly rendersmetric / chart / table / text / divider / image— unknown types render blank.react-leaflet/maplibre-gl/deck.glcan't be imported.<Image>is<img src>only;<Text>strips raw HTML.Today we either ship non-interactive PNG choropleths via
<Image>, build a parallel Streamlit dashboard, or fork DAC.Proposed solution
First-class
<Map>widget:Hover tooltip, zoom/pan,
<Filter>can drivevalueColumn.Alternatives considered
chart: heatmapon binned lat/lon — pixel grid, not a map.<Image>— no interactivity.MapWidget— works, but per-user install + upstream drift.Additional context
Geographic views are high-leverage for any regional dataset (EU NUTS, US counties, country comparisons). Current PFAS investigation has 1,165 NUTS3 with metrics where the regional pattern is the story — bar charts hide it.