The flood is two draw sets per layer: the corridor's dashed halo and
glowing core as draped lines (on tiles too under photoreal, on terrain
otherwise), moved in the frame the event moves them, and the surge front
as a clamped point. Only what changed is drawn again. Neither set is
pickable, and height queries see through both, so the shot paths no
longer list them to exclude.
The before/after images are globe-port layers on either side of the
split, over every layer drawn when they are added and draped on the
globe's surface only, as before. The divider moves the globe port's
split. An image that fails to load now takes the comparison (the pair,
the divider and its labels) down and leaves the event on, where a
failed provider used to fail the enable.
Evidence heights come through heightsOf, whose batch can now count an
answer from the bare ellipsoid as a miss, so points still fall back to
their authored elevations without terrain. The camera, canvas and the
event's vector maths go through the camera port, the renderer and
src/geo (lerp joins geo/vector).
The Cesium camera port gains move, look, rotateAround and zoom, and the
browser conformance run now checks them, the lookAt-frame commands, the
changed thresholds and moveEnd (timed by the clock, not a frame count).
Orbit, the Bhote Koshi focus and cinematic, the launch zoom anchor and
replay chase, the camera verbs' motion tick and roll levelling, and
voice zoom write through the port.
About 20 sites called scene.sampleHeight, globe.getHeight,
scene.clampToHeight or the MostDetailed variants directly, each with its own
idea of a miss (undefined, NaN, a throw). They now ask heightsFor(scene) in
src/queries/heights.js:
- now(point, { against, exclude }) answers from what is loaded, or null;
against is 'surface' (what is drawn) or 'terrain' (loaded terrain only).
- clampNow(position) moves a position onto the drawn surface, or null.
- batch() and clampBatch() wait for the most detailed tiles: null per
missed point, a rejection when the whole query fails.
- supported() reports { surface, clamp } without allocating.
Cesium still answers now() synchronously, so nothing changes on screen and
nothing is slower: in the live app 1,000 calls cost the same through the
service as direct (terrain 0.5-0.7 ms, surface 2-6 ms a call, run-to-run
noise larger than the difference). Every site already treated a miss as
"hold the last value or fall back"; a renderer that can only answer from a
cache or asynchronously can now plug in at one place.
- A Bhote Koshi evidence point whose terrain sample misses stays at its
fallback elevation instead of reading the miss as sea level.
- The mesh floor sampler skips a scene without depth-texture support
instead of spending its budget on probes that throw.
- The world annotation probe's trace reads hit/missed.
- A source scan forbids direct height calls outside the service.
The service is exported as ./queries/heights and documented in
docs/APPLICATION.md.
The Bhote Koshi layer keeps its panel controls as plain records and
presents them to a store; the context rail renders the panel as its last
child and marks itself while it is on. Controls are bound to the enable
that published them, and an unchanged frame is not sent again.
createImagerySplit keeps the divider's value, drag, keyboard, CSS position
and scene split, and publishes its line to the imagery split store.
ImagerySplitView, in the scene chrome, teleports each line to the body, so
recent imagery and the Bhote Koshi event no longer build divider DOM.
Retain active-flight preloading while canceling hidden frames and pending Facebook sessions on stop, disable, and replacement. Add lifecycle and pointer-ownership regression coverage. Apply the runtime formatting adopted on main and keep the existing source-shape check insensitive to that formatting.