Preserve the camera, map source and layer selection across engine sessions. Restore the previous engine if startup fails and retain local history navigation.
The HUD builds place context for its AI summary whenever the camera
settles, and each build paid for up to four reverse geocodes plus a Nearby
Places search, even when no OpenAI key was configured to use them.
- Ask GET /api/openai/hud-summary/status once; with no summary configured,
or after an unconfigured reply, make no place lookups for the page.
- Reverse-geocode at most two points: the view center and one viewport
point.
- Reuse a lookup until the view center moves a quarter of the visible
width, the camera's distance to it doubles or halves, or the camera
crosses a lookup height. The reused points are asked again, so only a
timeout, failure or empty answer spends.
- Repaint the last AI reply when a small pan leaves the context unchanged.
Cached-only hosts and voice context lookups are unchanged.
Main gained three imagery providers (Google Street View, KartaView,
Panoramax), shared line-coverage, cone and photo-viewer modules, one source
lit at a time and the translated interface. Their shared modules still
carried what this branch removes, so the merge ports them onto it:
- lineCoverage.js is this branch's coverage.js made provider-neutral
(draped lines near the ground only, the selection drawn over them, key
and rate-limit blocking); sequenceCones.js is sequences.js the same way.
Mapillary, KartaView and Panoramax all build on them; Mapillary moves
back to providers/mapillary/ beside the others.
- No ground casting, mesh sampling, surface modes, FOLLOW, legend, prewarm,
orbit overview points (Panoramax's density dots included), stale-tile
swaps or water-filled point budget come back.
- Kept from main because the new providers need them: requiresMapStack and
attachMapStackController (Street View on Google 3D only), ground clicks,
openNearest with an optional provider nearestImage, openImage {frame},
reportError, the view-cone marker (now an SVG glyph), filters /
coverageLabel / sinceBasis, OPEN STREET VIEW, keyAlternatives and i18n.
- photoViewer.js treats the viewer as expanded when its wrap is full screen
or the iPhone popover, as EXPAND now is.
- The fixture browser gate keeps its 8 steps and adds one each for
KartaView and Panoramax; catalogs, docs, boundaries and format scope
follow.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The flights records and the mesh-floor sampler kept identical copies of the equirectangular distance; both now import src/data/approxDistance.js. The changelog entry moves to changelog.d (#787), and CURRENT-STATE notes the antimeridian behaviour.
Review of Task 6: from 15 km with cold tiles Navara answers from 3.5-4 km
tiles (13.9 km at AUS), below the 4 km floor limit, 6-16 m off; the mesh
floor samples up to 25 km, so those would latch. MAX_FLOOR_TILE_SIZE_M is
now 1 km: from 800 m the answering tiles are 112-314 m. The 1.8 km tile
that read the Austin ground right under a 600 m camera is now a miss too,
which only leaves the DEM prior in place.
qa-surface-heights runs the 15 km pass first in a fresh page (cold), then
the 800 m pass in another, and keeps every answer the floor rule would
take while the tiles stream. Measured: 800 m 19/20 within 2 m of Cesium;
15 km cold 7/20 (the plan's gate of 15 assumed warm tiles - lod 'finest'
can only use what Navara has loaded, and at 15 km it loads 3.5-14 km
tiles), and the floor rule takes 0/20 of those; while streaming it took
answers more than 2 m off at 3 places: the Empire State spire (39 m, also
the settled answer) and two street points 2.2-2.3 m off.
engine.surface.sampleHeightDetailed(place, { exclude, lod }) answers
{ height, source: 'tiles' | 'terrain', tileSizeM? } (src/globe/surfaceDetail.js).
On Navara tileSizeM is the extent of the tile that answered and
lod 'finest' takes the finest loaded tile, drawn or not; Cesium answers
tiles without a size, else the globe's terrain. acceptedSurfaceHeight()
turns an answer into a height or a miss: tiles larger than
MAX_FLOOR_TILE_SIZE_M (4 km) are too coarse, terrain is not the mesh.
The mesh floor sampler and the ground snap are engine-only now (a viewer
or scene reaches its engine through engineFromViewer, which also registers
the scene) and treat a coarse answer as a miss, as they treat undefined.
Measured on Navara at Austin: while the tiles stream, a 27.9 km tile reads
143.2 m where the ground is 122.5 m (a 1.8 km tile).
`exclude` takes the engine's own mover sets (MoverSet.id): Cesium hides a
set's primitive for the synchronous sample (Cesium matches excluded objects
per billboard, not per collection); Navara's ray only meets tiles.
"sampleHeight undefined in the app on Navara" did not reproduce: the
startup Austin sample is 121.9 m (tile 889 m) on Navara and 122.4 m on
Cesium; one ModelMesh class is patched.
S2 QA (scripts/qa-surface-heights.mjs, 20 places at JFK, SFO, AUS, roofs
and streets): Navara within 2 m of Cesium at 19/20 from 800 m and 18/20
from 15 km with lod 'finest' (gates 18 and 15); offsets per region
-0.21 / +0.4 / +0.1 m; the Empire State roof differs by 39 m (spire
geometry). Navara tile sizes at 800 m: 112-314 m.
The flights ground-floor pipeline subtracted longitudes without wrapping.
approxDistanceKm, and its copy in the mesh-floor sampler, measured two
points 0.2° apart either side of 180° (21 km at Fiji) as about 38,000 km,
so low and grounded aircraft just across the line from the viewer were
never floored or mesh-sampled. corridorFloorCells walked a segment across
180° the long way round, filling its cell budget with ground behind the
contact and never requesting the cells it was about to cross, and
stickyFloorCell dropped its hysteresis for a contact on the seam. Each now
takes the short way round, and corridor points are normalised back into
[-180, 180].
SEARCH NEARBY SITES centred its Google Places search on the mean of west
and east, which for a view across 180° (west > east) is longitude 0, and
sized the radius from the view's height alone. It now measures the view's
width through 180° and searches its real centre.
groundSnap.heightFor and meshFloorSampler.sampleMeshFloorCells take the
globe engine: engine.surface.sampleHeight for the probe,
engine.surface.tilesLoaded (and, for mesh cells, maps.photorealActive)
for readiness, camera.height for the mesh sampler's altitude gate. Callers
that still hold a viewer or scene (the layers until phase 1c) keep the
unchanged Cesium path.
createCesiumEngine wraps an existing viewer without changing it: camera, picker, projector, frames, render (through the render governor), surface, input, credits and maps, with geo values out and Cesium values in at the boundary. Map and surface code is shared with src/maps and services/groundSnap instead of copied. scripts/qa-globe-contract.mjs runs the contract suite against the engine in headless Chrome; the camera-events check now waits for the previous move to end, as Cesium's moveStart marks a move from rest.
A warm entry could be evicted by the same call's later fetches and then
reported unresolved. The call now remembers warm hits as it does fetched ones.
Moves the batch LRU test beside the existing cache tests and folds the docs
note into the section's eviction paragraph.
Move the Cartesian3/Cartographic/Matrix/Quaternion/Transforms/Math calls of
the first batch (src/data motion, anchor, tracked-camera and icon helpers;
the root camera, locations, HUD, share-link, world-focus, ground-guard,
orbit and cloud modules; services/groundSnap) to src/geo. Behaviour is
unchanged: same argument order, same `result` semantics.
Kept on Cesium: out-of-scope classes (JulianDate, Axis, HeadingPitchRange,
BoundingSphere, Rectangle, EasingFunction, EllipsoidalOccluder with its
WGS84 and seed vector, SceneTransforms) and Matrix4.IDENTITY wherever it is
handed straight to a Cesium camera call (locations.test pins its identity).
synthesizeForwardKinematicsFix now returns a plain position, so the two
snapshot renderers clone it with the static Cesium.Cartesian3.clone instead
of the instance method. The package-boundary groups that reach these
modules now own the src/geo files.
FOLLOW is gone: no setFollow, follow state, map-stack availability or
handoff listener. Opening a photo still frames it with one flight through
the application's deferred navigation (begin/reassert), so a slow photo
never overrides newer navigation and tracking is released as before.
Ground casting is gone: groundCast.js, meshSampler.js, the surface mode,
the draped-to-cast swap and remesh. Coverage lines are always draped
GroundPolylinePrimitives (ClassificationType.BOTH); cones and the marker
clamp to the ground and keep the horizon cull. The terrain/meshFloor/ground
services and meshFloor.visibleTilesetLoaded are no longer needed.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Photo framing and FOLLOW claim the camera through the application's
navigation, injected with attachNavigation({ run, subscribeHandoff }):
aircraft and satellite tracking is released first, the cockpit refuses,
and FOLLOW stops when another feature takes the camera. No UI imports
in the layer.
- The mesh sampler follows the mesh floor's rules: it waits for the
visible tileset to finish streaming (visibleTilesetLoaded, now exposed
by the mesh floor service), needs a real bare-earth prior and keeps to
the shared meshFloorSampleWithinPrior window, which cameraFollow now
uses too. A cell is probed again once the camera is half as far from
it (down to 40 m), so a coarse early sample is replaced and redrawn.
- The MCP globe panel's CSP allows the Mapillary Graph API and its image
CDN (https://*.xx.fbcdn.net).
- The tile proxy reads bodies with the shared readResponseBytesCapped.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Traffic roads come from OpenFreeMap vector tiles, or TomTom flow tiles with a
key (Hybrid by default). Military areas use OpenFreeMap polygons with a bundled
worldwide name index; wide views show named points. ALPR reads an hourly
OpenStreetMap extract. Cockpit place context and country, state and US county
outlines use bundled Natural Earth and US Census data. Overpass is used only
when an operator configures their own endpoint.
Also: camera arrivals no longer land below ground, keyless terrain retries on
429, bounded relation geometry for configured Overpass, one shared
OpenStreetMap credit, and separate tile and API transports for embedders.
Co-authored-by: theGaryLarson <[email protected]>
Co-authored-by: Milan Khanal <[email protected]>
The in-memory cache in the terrain-height resolver had no eviction, so a
session retained every coordinate it ever resolved for the lifetime of the
page. The 5-decimal rounding documented in the file header makes repeated
visits to one place share a key, but it does not bound how many distinct
places a session visits.
Cap it at 20 000 entries, overridable per instance so the bound is testable
without minting twenty thousand coordinates. Eviction is least-recently-used
rather than insertion-ordered because revisits here are spatial: the cell the
camera is parked on must not be dropped for having been resolved early.
The bound sits far above a city-scale session at the ~111 m floor grid, so
normal use never evicts and issues no additional proxy requests. Batch results
are assembled from what the call resolved, so eviction during a large batch
cannot report a just-resolved point as unresolved.