nordisk/facett

facett-demo — crate card

Reverse-engineered (from source — verify, don't trust the name)

• One-line pitch: The showcase app — ONE eframe binary (plus a headless DemoState lib) that exercises every facett component with fake-but-realistic data, switchable across all palettes.
• Does: main.rs wraps facett_demo::DemoState in an eframe window with a palette switcher (over Theme::ALL), deck fx (glow toggle + send-raven), a facett-menu command palette (Ctrl+P), and an about/legend panel. The lib wires per-component tabs: graphview/graphpan, the falkor FalkorDB graph-DB browser (a facett_graphview::FalkorBrowser Cypher→laid-out-GraphModel feeding an interactive DecoratedGraphView, driven by a stubbed in-memory CannedRunner — no live redis; robot-tested via kittest: renders nodes+edges, a click selects a node), jobview, timeline/plandag, security, helix, pop (popgraph/pop_model/pop_tab over knut-popsim's Swedish 11M population sim), osm 2D/3D tabs, region picker (in-process osm-katana region clip), velo L1 tab, widget gallery, and the primitives tab (the #22-Part-B shared reducers: a dragdrop cards→columns board, a pan/zoom time_axis scrubber, and the action_bus outbox they dispatch onto — all engine-agnostic facett-core). Native build uses the wgpu backend (installs map/map3d GPU renderers); wasm build (cargo xtask demo-wasm) uses glow/WebGL, --no-default-features, CPU painters, published to Codeberg Pages, and ships the deck MINUS the 7 native-only korp caseworker-flow tabs (see the parity section below). Extra bins: osm2d_demo, osm3d_demo.
• Key deps: the umbrella facett (+map-wgpu/map3d-wgpu native), plus directly facett-map, facett-model3d, facett-app, facett-graphview, facett-graphpan, facett-jobview, facett-timeline, facett-plandag, facett-security, facett-helix, facett-core, knut-popsim; eframe, arrow-array. Native-only: facett-map3d, osm-katana (region-clip), korp-collectors (region-connector). Wasm: wasm-bindgen/web-sys/console_error_panic_hook.
• Shape: both (lib facett_demo + binaries); medium (~18 src files). publish = true (gated on osm-katana 0.2.x landing on crates.io first).
• Observed role in the workspace: The living D4-compliance proof + screenshot gallery; the top-level consumer that pulls the whole facett surface together. Driven by a robot-UI live-drive test matrix.

Native ↔ wasm deck parity (the ONE documented divergence)

The deck (DemoState::new) and the LEGEND catalogue are built identically for native and wasm32 except for seven native-only korp caseworker-flow tabs. These are heavyweight case-management surfaces that belong to the full desktop caseworker app, not the lightweight wasm web demo, so both the deck vec![…] and their LEGEND entries are gated behind #[cfg(not(target_arch = "wasm32"))]:

So the deck is 72 facets natively (65 shared + these 7) and 65 on wasm (72 − 7). This is the only intentional native↔wasm divergence; every other tab is shared. The invariant is pinned by tests/pop_ran_list.rs:

• NATIVE_ONLY_TABS is the explicit allowlist for these seven (with rationale).
• native_wasm_deck_parity asserts the native deck is the canonical 72 AND that the wasm surface is the native deck minus exactly those seven — so any OTHER tab that goes missing from wasm (a divergence NOT on the allowlist) still turns the check RED.

To add a tab: shared → just add it to the deck + LEGEND (it flows to wasm automatically); native-only → gate it in both places in lib.rs AND add it to NATIVE_ONLY_TABS (+ the EXPECTED_NATIVE list in the parity test).

Fill in (one line each — only the human knows these)

• Primary consumer(s):
• Why it exists (vs the obvious alternative):
• Hard constraint (airgap / no-C-deps / no-FFI / pure-Rust / etc.):
• Release cadence & independence (rides a parent? own repo someday?):
• Status (production / prototype / experimental / stubbed):
• Must-NOT-change invariant:
• Biggest known gap or TODO:
• Anything a newcomer would get wrong about it: