Commit Graph
100 Commits
Author SHA1 Message Date
Sameh Khamis bb2d22dbce fix(wildlife): the card test's click carries the exact top hit the layer now reads
4f24b9a06 made a wildlife tap read the click's exact top hit; the card
test still handed only the touch search's id, so its clicks opened no card.
2026-11-14 09:10:09 -08:00
Sameh Khamis e72de07dab fix(rail): the selected route splits at the train and keeps its dark rim, as before
Before, the route was one clamped polyline whose material coloured it
ahead of the train and dimmer behind it, split exactly where the train is,
with a dark rim at its edges. Through the draw port it became per-segment
colours chosen by each segment's midpoint (the split jumped from vertex to
vertex) and lost its rim.

The contract gains a draped line style `split`: `color` from `at` (a
fraction of its length) on, `behind` before it (`reverse`: the other way),
and an optional `rim` (the outer `widthFraction` of its width at `shade` of
each colour). `at`, `reverse` and the colours change in place. The Cesium
renderer draws it with one material; the rail route uses it, measuring its
split in metres on the ground (one per vertex where the scene draws over
the surface). New conformance check draw.lines.split.
2026-11-14 08:33:31 -08:00
Sameh Khamis 96ef7f509e fix(render): Recent Imagery's swipe slots wait for each other's tiles without redrawing
The two slots share one small request budget through one admit hook.
When one slot held it, the other's deferred tiles asked the Cesium globe
for a redraw every 250 ms, about four extra frames a second while tiles
loaded; before, they waited for the holder's tile, whose settling draws
the frame that asks again. Layers given the same admit now share their
waits: with one of their tiles in flight, nothing more is asked for. A
layer deferring while an unrelated layer loads still retries after 250 ms.
2026-11-14 08:03:05 -08:00
Sameh Khamis ecf8b31756 fix(library): Norway's cadastre and sea depths draw to the far edge of a tilted view again
The two dense entries (cadastre from z14, sea depths from z11) were drawn
by Cesium only on globe tiles as fine as their minimum zoom, with their
minimum zoom tiles filling the coarsest of those. Through the globe port
they dropped the minimum level and refused every tile below it, so the
far band of a tilted view, one globe level coarser, was blank.

The imagery contract gains hiddenBelowMinLevel: a layer drawn nowhere the
surface is coarser than its minLevel. Cesium gives such a layer
minimumTerrainLevel as the library set it before (minLevel - 1 for Web
Mercator tiles on the geographic globe); the entries keep their minimum
level and drop the admit hook.
2026-11-14 08:01:36 -08:00
Sameh Khamis 91b080a45a fix(geo-import): KML icons stay on their points when zoomed out, and turn by heading as before
Cesium's KML loader shrank an icon's hot-spot offset with the icon and
turned a heading about the Earth's axis. The drawn pins kept the full
offset at globe distance (up to ~14 px off for the Google Earth pushpin)
and turned about local north, which differs in a tilted view. They now
scale the offset with the icon and turn about the Earth's axis by the
heading.
2026-11-14 02:03:26 -08:00
Sameh Khamis 7ba15deb5e feat(render): an icon's offset may scale with it, and an axis-turned icon turns by its rotation too
The contract gains offsetScalesByDistance: the pixel offset grows and
shrinks with scaleByDistance, as Cesium's pixelOffsetScaleByDistance does
with the same curve. rotationMode 'axis' now also turns by rotation, as a
Cesium billboard with an aligned axis and a rotation does. Cesium draws
both on its billboards; its GPU icons take the rotation and refuse a
scaled offset; the stub picks both. New conformance check
draw.icons.offsetTurn.
2026-11-14 02:03:26 -08:00
Sameh Khamis c99c577b3e fix(trails, data-packs): a scene with no page to take input from draws without a double-click
The double-click follow asked for the scene's input when a set was first
drawn. On a headless scene Cesium's event handler needs a document, so
drawing a trail or a pack threw there. Asking is guarded, as Directions
does: a scene that cannot take input has no double-click.
2026-11-14 01:55:35 -08:00
Sameh Khamis 4f24b9a062 fix(wildlife): a tap selects only a sighting exactly under it again
A click read the scene's exact pick at the tap. Through the input
dispatcher it read the click's id, which on a touch screen comes from the
wider contact search: a tap near a sighting selected it, and one near
another layer's contact kept the selection. It reads the exact top hit
again, as the shipwrecks and Local ADS-B layers do.
2026-11-14 01:51:24 -08:00
Sameh Khamis cbc0c114eb fix(local-adsb): a tap selects only an aircraft exactly under it again
A click read the scene's exact pick and a drill at the tap. Through the
input dispatcher it read the click's id, which on a touch screen comes
from the wider contact search: a tap near a local aircraft selected it,
and one near another layer's contact kept the selection. It reads the
exact top hit again, as the shipwrecks layer does.
2026-11-14 01:49:17 -08:00
Sameh Khamis 805c9b5ccd fix(voice): point-and-ask over a Street Level coverage line names its sequence again
The coverage lines were one primitive per sequence, picked as the
sequence's id. Drawn in parts, a part after the first carries its own
#n id, which point-and-ask passed on. A hit whose item names the id its
parts pick as (meta.pick) resolves as that id, as the line's click
already does.
2026-11-14 01:47:20 -08:00
Sameh Khamis d2e161372c fix(vessels): the ship the Cockpit rides is no object in view again
In-view scanning skipped a vessel whose chevron was hidden and not drawn
as its model, which covered the ship the bridge camera rides. It now
skipped only vessels past the horizon, so the ridden ship was reported,
at the camera's own position. It is skipped again unless its model draws
it.
2026-11-14 01:46:41 -08:00
Sameh Khamis 4a781f0480 fix(cockpit): the orbital view gives the field of view back on exit
Cesium's fov was saved and put back when the orbital view closed, so a
later resize or rotation kept Cesium's own behaviour. The exit held the
saved vertical angle through setFov instead, which the camera port kept
for the rest of the session: rotated to portrait, the view stayed
zoomed in. setFov(null) gives the field of view back: Cesium's fov as it
was before the first hold, the stub's own angle.
2026-11-14 01:45:27 -08:00
Sameh Khamis 03a7d8827a fix(embed): an inline panel's timer frames move the camera, take input and run the clock again
When a host stops animation frames, the panel drew from a timer with the
viewer's render(): the camera's flight and input, the clock's tick, then a
draw if due. Through the frame port it called renderNow(), which on Cesium
only forced a draw, so flights stood still, drags and zooms did nothing
and the clock stopped. renderNow takes {ifDue: true}: one tick of the
loop as its own would run it, drawing only a frame that is due (the
widget's render() on Cesium; the stub's tick without a forced request).
2026-11-14 01:43:13 -08:00
Sameh Khamis f4beb676ae fix(rail): the selected train's line draws over every network loaded before it again
Each route was a primitive added when it was drawn, so it sat over the
networks of feeds loaded until then. Its draw set was made once and kept
the slot of the first route, so a feed's network loaded after that drew
over later routes. Each route is drawn in a set made for it.
2026-11-14 01:40:41 -08:00
Sameh Khamis 0287ddb9be fix(local-adsb): the selected aircraft's trail head ends on its icon every frame again
The head was a polyline entity whose positions were read each frame. Its
draw-port line was redrawn at most every 100 ms while the aircraft flew,
so its end lagged the icon. It is drawn each frame the aircraft moves, as
the Flights trail head is; the trail still skips a head that has not
moved.
2026-11-14 01:38:50 -08:00
Sameh Khamis 6af4b08293 fix(local-adsb): a click or double-click on a UAT ring takes its aircraft again
The 978 MHz ring was a point on the aircraft's own entity, so a pick on
it was a pick on the aircraft. As a separate points set it was drawn
unpickable and a click there fell through to what is under it. The rings
are picked again and read as their aircraft.
2026-11-14 01:37:59 -08:00
Sameh Khamis aebb3e79c3 fix(wildlife): the reveal pan eases quadratic out again
The pan used Cesium's QUADRATIC_OUT; through the camera port it took
cubicOut, the contract's nearest ease. The contract gains quadraticOut,
which Cesium maps to QUADRATIC_OUT and the stub camera computes.
2026-11-14 01:37:24 -08:00
Sameh Khamis 4b665147e1 fix(traffic): road lines stack as before: flow lines jam on top, heat-lines slow over jam
The lines were one ground primitive per bucket, added free, slow, jam for
flow lines and jam, slow for heat-lines, so where roads cross the later
bucket drew on top. Through the draw port they were one set in kept order
(jam first), and its glow lines draw after its solid ones, which turned
both stackings over. Flow lines are now set free, slow, jam; slow
heat-lines go in a set made after the jam lines' set.
2026-11-14 01:35:56 -08:00
Sameh Khamis df0afa82e1 fix(library): a double-click follows a view dot and zooms to a view area or coverage border again
A view-bounded layer's features and a tile entry's coverage outline were
entities, which the default renderer's double-click followed (points) or
zoomed to (lines and areas); a new view replaced them and ended the
follow. As draw-port items they took no part in it, and the outline was
drawn unpickable. It is picked again, as its entities were.
2026-11-14 01:34:13 -08:00
Sameh Khamis 898909c245 fix(cctv): a double-click zooms to a coverage line and follows the monitor plane again
The coverage rays and cap were polyline entities, which the default
renderer's double-click zoomed to, and the monitor plane a plane entity
with a position, which it followed until the plane was removed. As
draw-port items they took no part in it. The camera icons were billboards
before, so they still take none.
2026-11-14 01:32:03 -08:00
Sameh Khamis 0dc3a1fb66 fix(bikeshare): a double-click follows the selected station again
The selected station's highlight was a point entity, which the default
renderer's double-click followed; deselecting removed it and ended the
follow. As a draw-port point it took no part in it. The station dots were
point primitives before, so they still take none.
2026-11-14 01:30:38 -08:00
Sameh Khamis 9c4c831ba2 fix(data-packs): a double-click follows a pack's point again
A pack's points and media marker were point entities, which the default
renderer's double-click followed; removing the pack ended the follow. As
draw-port points they took no part in it. Lines, polygons and images were
primitives before, so they still take none.
2026-11-14 01:29:22 -08:00
Sameh Khamis a9de7b3510 fix(trails): a double-click zooms to a trail line again
A trail was polyline entities, which the default renderer's double-click
zoomed to. As draw-port lines they took no part in it. Every trail on a
viewer shares one double-click handler, let go with the last trail.
2026-11-14 01:28:28 -08:00
Sameh Khamis 322e4a91d3 fix(annotations): a double-click follows a pin and zooms to an area or line again
A mark's ring, marker, areas and lines were entities, which the default
renderer's double-click followed (ring and marker) or zoomed to (the rest)
until the mark was removed. As draw-port items they took no part in it.
2026-11-14 01:25:45 -08:00
Sameh Khamis 94ea9cd01d fix(submarine-cables): a double-click follows a landing point or reference and zooms to a cable again
The cables, landing points and references were entities, which the default
renderer's double-click followed (a point, or a reference at its tip) or
zoomed to (a cable or area) until the sources were replaced or hidden. As
draw-port items they took no part in it.
2026-11-14 01:24:24 -08:00
Sameh Khamis d003d9ee78 fix(news): a double-click follows a marker or zooms to an outline again
The markers and outlines were entities, which the default renderer's
double-click followed (a marker, until the markers were drawn anew) or
zoomed to (an outline or hotspot box). As draw-port items they took no part
in it.
2026-11-14 01:22:44 -08:00
Sameh Khamis 6ac8aed1b5 fix(radio): a double-click follows a station again
Each station, and the selected one's dot, was an entity, which the default
renderer's double-click followed until the stations or the selection were
drawn anew. As draw-port items they took no part in it.
2026-11-14 01:20:12 -08:00
Sameh Khamis c57a8bea83 fix(alpr): a double-click follows a camera with its card again
Each camera was an entity, which the default renderer's double-click
followed, showing its card; a camera no longer shown, or a hidden layer,
took the follow with it. As a draw-port icon it took no part in it.
2026-11-14 01:18:51 -08:00
Sameh Khamis 37a1e1500b fix(air-quality): a double-click follows a station again
The stations were entities, which the default renderer's double-click
followed, framed to fit the disc; one no longer reported took the follow
with it. As draw-port items they took no part in it.
2026-11-14 01:17:37 -08:00
Sameh Khamis fbc40ef3b6 fix(earthquakes, cyclones): an item that goes takes its double-click follow with it
Removing a followed entity ended the default renderer's follow. A disc gone
from the feed, or a storm point gone from the advisory (or a cleared one),
now ends it too.
2026-11-14 01:17:08 -08:00
Sameh Khamis def2e1ecca fix(perimeters, drone-airspace): a double-click zooms to an area again
The areas were polygon entities, which the default renderer's double-click
zoomed to. As draw-port items they took no part in it.
2026-11-14 01:16:20 -08:00
Sameh Khamis 26e8e26ba8 fix(ghostmaps): a double-click follows a pin or zooms to a line or area again
Each edition was drawn as entities, which the default renderer's
double-click followed (a pin) or zoomed to (a line or polygon); an expired
report or a replaced edition took the follow with it. drawnItemTarget reads
that rule off a draw item, for every layer that drew entities before.
2026-11-14 01:15:29 -08:00
Sameh Khamis 766cf68272 fix(local-adsb): a double-click follows a local aircraft with its card again
Each aircraft was an entity, which the default renderer's double-click
followed, showing its tracked card; an aircraft that expired took the
follow with it. As a draw-port icon it took no part in it. A double-click
follow can now carry the holder whose entity record shows the card.
2026-11-14 01:12:14 -08:00
Sameh Khamis f2983e8c1a fix(osm-places): a double-click follows a place again
The places were entities, which the default renderer's double-click
followed; new results replaced them all, which ended the follow. As
draw-port items they took no part in it.
2026-11-14 01:10:46 -08:00
Sameh Khamis fabf4439fe fix(cyclones): a double-click follows a storm's point or zooms to its track or cone again
The advisory was drawn as entities, which the default renderer's
double-click followed (centre and forecast points) or zoomed to (tracks,
cones, outlines). As draw-port items they took no part in it.
2026-11-14 01:10:16 -08:00
Sameh Khamis 50cd49cc31 fix(earthquakes): a double-click follows a disc again
The discs were entities, which the default renderer's double-click followed,
framed to fit the disc. As draw-port items they took no part in it.
2026-11-14 01:09:31 -08:00
Sameh Khamis b9ccdff0a3 fix(flights): a delivery after a pause in polling does not raise the display delay
A hidden tab pauses polls; its first delivery after that measured the pause
as feed lateness and pushed the delay up for minutes.
2026-11-14 01:07:04 -08:00
Sameh Khamis 72cfea35c4 A live motion clock's new delay restyles nothing on Cesium
Aircraft layers move their display delay a little every frame. Every frame
already reads the clock from the set, so a delay-only change now just asks
for a frame instead of restyling every row.
2026-11-14 01:06:21 -08:00
Sameh Khamis e3cfdcf44e Tracked aircraft keep moving when reports give no speed or a stale one
A fix that moves while the reported speed is 0 or stale used to leave the
aircraft frozen, then jump it kilometres at once. The step between fixes now
measures a jump, and the drawn speed bounds the recovery.
2026-11-14 01:06:21 -08:00
Sameh Khamis 22952f8f16 fix(cockpit): double-clicking a ship that the first click moved enters without dropping the follow
Selecting a ship re-centres the camera, so by the second press of a
double-click the ship is drawn elsewhere. The Cockpit remembers where it
was and still enters there. On Cesium it then kept the stock double-click
follow from running; on the input port the hold was taken only for a press
where the ship is drawn now, so the stock follow ran first and could let
go of the ship. A press where the ship was as the gesture began now holds
it off too.
2026-11-14 00:57:31 -08:00
Sameh Khamis b754418706 fix(cockpit): a ship's bridge reads its water from the drawn surface, globe included
The waterline probes read Cesium's scene sample, which takes the globe's
terrain as well as 3D tiles. On the ports they asked for tiles alone,
which hides the globe, so on a globe map a lake or river above sea level
went unread and the eye fell back to the sea. They now ask for the drawn
surface, as before.
2026-11-14 00:56:07 -08:00
Sameh Khamis 5d48503c51 fix(camera): moveEnd fires during a follow once the eye stands still, as Cesium's did
Cesium raised moveEnd 500 ms after the camera last moved, following or
not: a followed target that moves keeps the eye moving, and one that
stands still (a marker, a parked aircraft) lets the move end once
the user stops orbiting it. The port never raised moveEnd while
following, so Weather, Recent Imagery, the share link and every other
moveEnd listener stopped catching up around a still target. The rule is
now the eye's alone, through the shared camera events.
2026-11-14 00:46:51 -08:00
Sameh Khamis 63a6595eaf fix(directions): a double-click follows an A or B marker and zooms to the route again
The markers and the route were entities, which the default renderer's
double-click followed (a marker) or zoomed to (the route). As draw-port
items they took no part in it. Directions now joins the stock double-click
as launches and imported pins do.
2026-11-14 00:44:58 -08:00
Sameh Khamis 4c7deb11dc fix(voice): point-and-ask over an ALPR camera or a local ADS-B aircraft names it again
A pick resolves to its context record through the drawn item's meta. The
ALPR and local ADS-B icons carried none, so point-and-ask got the layer
name and no position instead of the camera or aircraft and where it is.
Each icon now carries its context carrier, as before the draw port, when
the pick was the registered entity itself.
2026-11-14 00:42:34 -08:00
Sameh Khamis ca7a736456 fix(alpr): no height sample is spent without sampling, and no terrain anchor under a hidden globe
On Cesium the overlay sampled the drawn surface only where the scene
supported height sampling (depth textures), and fell back to the globe's
terrain only while the globe was shown. On the ports, a renderer without
sampling spent the paint's sample budget on misses and the refine pass
kept asking every paint, and a hidden globe's stale terrain could anchor a
marker under photoreal tiles. Both rules are back, through the renderer's
sampleHeight capability and the globe's visibility.
2026-11-14 00:35:32 -08:00
Sameh Khamis 1c036957cd fix(alpr): anchors wait for the surface heights answer from, as before
The overlay replaced floor anchors with sampled surface heights only once
the surface had settled. On Cesium that was the globe's terrain while the
globe was shown, otherwise every shown 3D tileset; on the globe port it
became the terrain and the drawn base tileset together. It now asks the
globe port for 'surface', the same rule.
2026-11-14 00:34:18 -08:00
Sameh Khamis 667ca632ed fix(library): view-bounded GeoJSON draws as the GeoJSON loader drew it
On the draw port, a Data Library dataset's polygon and line edges were
drawn as geodesics and every feature took the entry's colour. The loader
drew ground edges along rhumb lines and honoured a feature's own
simplestyle fill, fill-opacity, stroke, stroke-opacity and stroke-width,
with a colorBy fill over them. The library now uses local GeoJSON's
rhumb densifying and simplestyle reading for the same result.
2026-11-14 00:32:22 -08:00
Sameh Khamis 5f98f8bdf3 Aircraft drawn as far behind real time as their feed is late, along one curve their trail follows
Civil and military aircraft were drawn a fixed 30 s (15 s) behind real time
while anonymous feed reports arrive about 90 s old every 30 s, so every
contact dead reckoned about a minute past its newest report and snapped
when the next one landed (several km on a turn), and the tracked trail,
recorded from those drawn positions and patched as reports arrived, kinked.

- The display delay (data/displayDelay.js) is the 90th percentile of the
  newest 256 report lags plus 5 s, between the layer's floor (30 s, 15 s)
  and five minutes, slewed at most 0.2 s/s up and 0.05 s/s down. Icons,
  the renderer's live motion clock, labels, HUD, detection, the tracked
  camera and the trail read the one delay; following the world clock
  draws with none and deliveries leave it alone.
- Between two reports an aircraft flies a biarc (data/fixCurve.js) that
  passes through both and leaves each along its track, sent as contract
  arc segments; the trail is the reports joined along the same curve, so
  the icon sits on it. Nothing drawn between reports is kept.
- A report a coasting contact missed is eased onto from where it would
  have been that frame (at least 3 s, at most 30% over its speed, within
  20 s); over 5 km it jumps. Past its newest report a contact coasts
  straight on: a turn rate averaged over earlier reports outlived the turn.
- Forward synthetic fixes no longer block real reports older than them,
  and a history keeps 24 fixes.
- A feed simulation (testSupport/aircraftFeedSim.mjs) checks both layers
  against a 90 s late feed, a fallback switch, repeated reports, outages
  and the world clock.
2026-11-14 00:31:29 -08:00
Sameh Khamis c57eb74506 fix(radio): a cluster kept from the last pass names the same station first
Points too close together on screen are grouped by
geo/screenCluster.js as the default renderer's entity clustering grouped
them. Its range queries ran on a uniform grid sorted by index, while the
reference queries a KDBush tree, so a cluster kept as the eye came lower
could list its stations in another order, and a click on it played a
different station. The index is now KDBush, built as the reference builds
it, and kdbush is a declared dependency (it already came with the globe).

The package boundaries list kdbush as external wherever screenCluster is
owned, and own src/viewSky.js beside src/viewWeather.js.
2026-11-14 00:29:26 -08:00
Sameh Khamis 14e406002c fix(voice): the in-view scan sees News reports and Wildlife Sightings again
Voice's in-view scan projects each context record's targetPosition. News
and Wildlife Sightings still set only the entity's base position, so both
dropped out of "what am I looking at". They now pass it as targetPosition,
as GeoJSON and FIRMS do.
2026-11-14 00:23:36 -08:00
Sameh Khamis ea3b87805a Raised GeoJSON and cable polygons are lit by the sun again
The loader's entities drew a polygon with per-position heights through a
lit appearance; through the draw port it went flat. Local GeoJSON and
submarine cable fills with heights pass lit: true, as file import does.
2026-11-13 23:59:19 -08:00
Sameh Khamis af3383a5a9 Display panel: a Sky group for the atmosphere, day and night, sun, moon, stars and Milky Way
createViewSky sets globe.sky from the panel's switches and Milky Way
slider and keeps the choice for the next visit; what the user has not set
stays as the renderer draws it. Distance fog stays with the styles that
switch it.
2026-11-13 21:13:35 -08:00
Sameh Khamis 5cd28731fa Renderer conformance: checks that read an exact colour set a bare sky
With atmosphere, fog and lighting off and clear weather a renderer applies
none of its look, so draw.models.tint, draw.models.blend, globe.base.vector
and globe.imagery.video set that sky first and test what they draw, not a
renderer's sky light or tone mapping.
2026-11-13 16:31:49 -08:00
Sameh Khamis 1b226e14f2 Renderer contract: the renderer's look is its own; clouds are observed cover placed by scene time
globe.sky switches parts of the sky on and off; how the sky, atmosphere,
haze, lighting, tone mapping and exposure look is the renderer's own, and
no check compares colours. globe.weather's clouds is the fraction of the
sky covered; clouds are placed by scene time and the wind and shade the
ground under them where the sun lights it. New conformance check
globe.weather.sceneTime.
2026-11-13 16:08:41 -08:00
Sameh Khamis 7a61bf04a0 feat(selection): selecting a contact keeps its size and the camera's view: aircraft, vessels, trains and transit vehicles keep their form and size when selected (colour only); a click follows from the camera's own distance and angle, the contact gliding to the centre over one second (FollowTarget transitionS, shared rule render/camera/followTransition, conformance camera.follow.transition) 2026-11-13 15:02:28 -08:00
Sameh Khamis e3865f462f Merge origin/main into the renderer-contract branch: the procedural fleet models drawn through the renderer's model sets
Aircraft, military, local ADS-B, vessel, transit and rail models come from
the fleet model set, loaded up front and drawn life size through the draw
port's models sets; no layer touches the viewer or Cesium outside
src/render/cesium/. The models capability gains glTF bytes (gltf), a
per-axis stretch (axisScale), a colour for tinted parts (maskColor) and a
cut below the origin (clip: 'belowOrigin'), with conformance checks.

Main's rail layer and Data Library are ported to the ports; the CCTV
footprint probe keeps the globe-terrain clamp through the query port.
2026-11-13 14:14:42 -08:00
Sameh Khamis 7e45d4d989 fix(flights): a trail does not join its track history across a landing or take-off that was not seen 2026-11-13 12:53:18 -08:00
Sameh Khamis 693faae44e fix(flights): an adsb.lol aircraft with no ground track faces its true heading, not north 2026-11-13 11:50:05 -08:00
Sameh Khamis 0b1dc325bc feat(render): a renderer can draw weather: the weather capability adds globe.weather (clouds, fog, haze, rain and snow, 0 to 1, and the wind they drift on); the view shows the conditions observed under the camera, and the Display panel's Weather group sets them by hand 2026-11-13 11:02:56 -08:00
Sameh Khamis 92d34bc4cc Stretch a vessel to its beam only beside a reported length
A vessel drawn at its type's typical length keeps the model's proportions:
a reported beam alone would squeeze a 300 m container ship to that width.
2026-11-13 07:42:27 -08:00
Sameh Khamis 432a29ba59 Add the changelog fragment 2026-11-13 07:34:50 -08:00
Sameh Khamis a9d390c165 Draw a vessel of unknown length at its ship type's typical size
Most live vessel feeds send no length, and the resolver then fell through
every length rule to the range's last model drawn as built, so a cruise
ship or a container ship showed as the smallest model of its range.
types.json gives each passenger, cargo, tanker, fishing and service range
a typical_len (cargo 300 m, tanker 250 m, passenger 140 m, fishing 40 m,
...); a vessel with no usable length is chosen and drawn at it.
2026-11-13 07:34:41 -08:00
Sameh Khamis 71bfc8aba0 docs: changelog entry names #1714 2026-11-13 07:19:21 -08:00
Sameh Khamis f0b0378b11 docs: fleet models on vessels, transit and rail; no live source names monorail, gondola or funicular; model frame axes 2026-11-13 07:16:25 -08:00
Sameh Khamis 0d1e3fb703 fix(models): update the fleet type table (SF34 and SB20 draw the ATR 72; no tower model) 2026-11-13 07:16:25 -08:00
Sameh Khamis a44c42eb94 fix(vessels, transit, rail): fleet models keep their cards and detection, stand on the floor, clip ships at the waterline and follow NVG/FLIR
- a vehicle drawn as a model keeps its selected card and stays in the detection overlay
- transit and rail models drop the 1.5 m floor lift their icons carry
- vessel models are clipped below the waterline
- models switch to an unlit white look under NVG/FLIR and back
- unknown transit modes keep their icon
- aircraft model specs are cached per set; no key-array copies per frame
2026-11-13 07:16:25 -08:00
Sameh Khamis 058ed0ffeb docs: the fleet models, and changelog entry 2026-11-13 07:16:25 -08:00
Sameh Khamis 7b93045895 feat(vessels, transit, rail): draw fleet models under the 3D toggle
Vessels, Transit and Rail turn nearby icons into fleet models at the
aircraft layers' distances. A vessel is drawn at its AIS length and beam
(static dimensions now reach vessel rows), standing at the sea surface; a
transit vehicle or train stands on its road or rail, raised by its type's
elevation, and is painted in its mode or category colour where the model
marks it.
2026-11-13 07:16:25 -08:00
Sameh Khamis fcc44d0eaf feat(flights): aircraft draw life-size fleet models chosen by type
Flights, Military Flights and Local ADS-B draw each aircraft with the fleet
model its ICAO type, class or emitter category selects, scaled to the type's
length, built from the up-front loaded bytes. The eight third-party aircraft
models and their attribution are removed; every shipped model is now the
project's own.
2026-11-13 07:16:25 -08:00
Sameh Khamis f6760ad96c feat(models): the project's procedural fleet models and the resolver that picks one per vehicle
121 GLBs under public/models/fleet (about 2 MB, MIT like the code) with
catalog.json and types.json. src/data/fleetModels.js picks a model by ICAO
type, aircraft class, emitter category, AIS ship type and size, GTFS route
type, transit mode or rail category, and sizes it life size; the set loads
once, up front. Trail anchors are measured from the meshes.
2026-11-13 07:16:25 -08:00
Sameh Khamis ad9bd0f4f9 docs(changelog): 1696-i18n-entry 2026-11-12 22:07:30 -08:00
Sameh Khamis 5f96009b67 feat(i18n): an ./i18n entry for an application with its own language choice
startPageLocale, followPageLocale and switchLocale start, follow and switch
the page's language as the standalone application and the globe picker do,
for an application that offers the language elsewhere and passes
localePicker: false. The standalone application uses them.
2026-11-12 22:07:13 -08:00
Sameh Khamis 2bc3c1bed1 docs(changelog): 1695-vessel-photos 2026-11-12 20:11:20 -08:00
Sameh Khamis d5b030e7fa docs(changelog): 1694-cctv-globe-plane-clamp 2026-11-12 20:11:19 -08:00
Sameh Khamis 23aa1027fc docs(changelog): 1693-cctv-frame-no-flash 2026-11-12 20:11:17 -08:00
Sameh Khamis 347e77d1f6 feat(vessels): a selected vessel shows its Open Vessel Photos photo and MMSI
Selecting a vessel asks Open Vessel Photos (a free, keyless API) for its
photo by MMSI, once per MMSI. The CONTACTS view shows it under the vessel's
name and MMSI with the photographer, the photo's open licence and a link to
the vessel's page; only images on Open Vessel Photos' own host are shown, and
an answer about another MMSI is no photo. The vessel source exposes the lookup
as getEnrichment({kind: 'photo'}) and fetchOpenVesselPhoto for applications
with their own vessel source.
2026-11-12 20:10:30 -08:00
Sameh Khamis 2445af349d fix(cctv): on a globe map the camera plane ends where the view meets the ground
The activation probe raycast only the photoreal tiles. On a globe map the ray
missed, or ran on through the Earth and met a primitive thousands of
kilometres away, and the monitor plane stood at the camera's full range:
about a kilometre wide and hundreds of metres up. A hit at or beyond the range
now counts as none, and on a globe map the plane also clamps short of the
globe's terrain along the axis, or flat ground at the mount's ground while
that terrain is still loading, whichever is nearer.
2026-11-12 20:10:02 -08:00
Sameh Khamis 6e4df4c287 fix(cctv): a new camera's frame shows the moment it is decoded, without a fade
The preview settled when the image had loaded, so the panel's <img> took the
src while it was still decoding and then faded in over 180 ms: every camera
click lingered on an empty box. The preview now settles once the frame is
decoded (a rejected decode still shows the loaded frame, and a decode that
never settles shows it after one second), and the fade is gone.
2026-11-12 20:10:02 -08:00
Sameh Khamis 639e92b4f7 feat(app): the globe opens at planet level
The camera starts with the whole Earth in view, looking straight down,
instead of flying in to Austin. A shared link flies on from there. A
different start, a city for instance, is set with the startView option
of createApplicationControls. The Cesium renderer and the test stub now
start from the same pose, so state that carries over from the first
frames (screen clusters, marker selection, label placement) no longer
depends on which of them drew it.
2026-11-12 19:05:44 -08:00
Sameh Khamis 1d476c87ee fix(location): a POI key held with Cmd, Ctrl or Alt runs no POI
The points-of-interest keys (Q/W/E/R/T while a city is expanded) are a
second bare-key document listener, and it did not look at modifiers
either: Alt+R or a delivered Ctrl+R picked the fourth POI and flew the
camera there. It now leaves Cmd, Ctrl and Alt chords to the browser like
the application shortcuts; Shift still counts as the key. The docs and
the changelog entry say so.
2026-11-12 18:02:41 -08:00
Sameh Khamis d1966b9751 Clear a view-bounded layer's error when the camera returns to the drawn view
A failed request for the view in between left its error set, and the
early return for the already drawn view never cleared it, so the drawn
features showed with an error state.
2026-11-12 17:57:31 -08:00
Sameh Khamis 92d6188045 Stop a moved-on view from cancelling the GeoParquet index the next view needs
The browser shared one partition-index build per entry but ran it under
the first view's signal. A camera move cancels the old view before the
next one starts, so the next view was handed the same build, already
cancelled, and drew nothing (the view layer takes an AbortError for its
own cancellation) until the camera moved again. EPC hit this on its
first view after any quick move. The build now belongs to no view: each
view stops waiting on its own signal, and the build runs on for the next
view and every view after it.
2026-11-12 17:56:17 -08:00
Sameh Khamis 8cca78bcd5 Stop sharing a Data Library entry as on once its failed Remove switched it off
When Remove switched a layer off and then its destroy failed, the
manager published no visibility change, so the selection still listed
the entry as on: the link shared it as on and a reload restored it on
while its row read off. A failed Remove now drops the entry from what is
on and remembered whenever its layer ended up off.
2026-11-12 17:49:38 -08:00
Sameh Khamis 837eff7328 Let a share fit a full Data Library and show entries it cannot serve
A share's entries joined the library after the stored ones, and the
selection kept its first 64, so a link opened into a full library never
showed what it shared and the next hash write dropped its lib=. Now the
oldest entries the share leaves off make room for it, and their layers
are removed with them.

Browse counted an entry as added only while its layer was registered,
but an entry this instance cannot serve (switched off by the operator,
or its server unreachable) stayed in the selection and held one of the
64 places. A library full of them looked empty and refused every Add.
Such entries now show as added, so they can be removed one by one or
with Remove all.
2026-11-12 17:49:01 -08:00
Sameh Khamis 4fe8796e89 Swap refreshed Data Library features in place, on every tick
The live refresh had three faults. Its data age counted from when a
fetch finished, so with the manager ticking every cacheSeconds the next
tick found the data just short of due and data refreshed every other
tick. It switched the layer off and on to redraw, which removed the old
features before the new ones were built, closed an open feature card and,
if the rebuild failed, left the layer on but empty. And the tick the
manager runs right after enable refetched after a failed first load, so
a second failure switched the layer off instead of showing UNAVAILABLE.

The local GeoJSON layer gains replaceFeatures(): it builds the new set
while the old one stays drawn, removes the old source only once the new
one is in the scene, keeps a selected feature that is still in the new
set (same id, or same place and label), and on a failed rebuild puts the
old features, records and selection back and rejects. The library's
refresh uses it, counts age from the request, allows a little slack on
the tick, and does nothing until a first load has succeeded.
2026-11-12 17:47:29 -08:00
Sameh Khamis 1a657e2a64 Keep the library layer namespace in the layer lifecycle package
The core layer lifecycle imported the whole Data Library manifest
contract only to check the lib- prefix, so the layer-lifecycle package
owned a feature module. The namespace rule (prefix, entry id pattern,
the two id helpers) now lives in src/data/libraryLayerIds.js, which the
lifecycle owns; the manifest contract imports and re-exports it, so
library callers are unchanged.
2026-11-12 16:04:55 -08:00
Sameh Khamis 9e57fd84c8 Draw Data Library coverage outlines from the bundled Natural Earth countries
Imagery outlines fetched Natural Earth's 1:50m countries (about 3 MB)
from the natural-earth-vector repository's master branch on GitHub at
run time: a moving, rate-limited source whose failures silently removed
the outlines. Core already ships the pinned 1:10m countries pack, so
the outlines now read that, once, the first time an outline needs it.
2026-11-12 16:02:07 -08:00
Sameh Khamis 8d90cac7dc Keep lib= in the URL before the library starts and for entries it cannot serve
The share encoder wrote only registered entries. Until availability
loaded nothing was registered, so any hash write in the meantime (a
camera move during start-up) deleted lib= from the address bar, and an
entry this instance could not serve (its proxy unreachable) was dropped
for good: copying the link or reloading lost those entries. The URL now
carries what the page opened with until the library starts, and every
entry that is on afterwards, served here or not.
2026-11-12 15:59:14 -08:00
Sameh Khamis 2d5db9f63d Never let a share rewrite the Data Library entries remembered as on
Every restore wrote the selection it applied to browser storage, so
opening a link that switched no library entry on stored every saved
entry as off, and the next bare visit restored them off. Core layer
state never writes a recipient's preferences from a share. The library
now keeps what is remembered as on apart from what a share switches on
for the session: a share's entries still join the library, but only a
local restore and the viewer's own choices change what is remembered.
2026-11-12 15:58:22 -08:00
Sameh Khamis 53f0a32a1b Keep a Data Library entry whose layer fails to tear down
Remove dropped an entry from the selection and reported success even
when the manager could not disable or destroy its layer, leaving a
registered, possibly visible layer that Browse, share links and saved
state no longer knew. Remove now keeps the entry and reports failure,
so its row and Remove all still reach it and a later Remove can retry.
2026-11-12 15:57:02 -08:00
Sameh Khamis f33a6ada52 Refuse a Data Library entry past the 64-entry cap before it registers
Add registered and switched on an entry before the selection was cut to
its first 64, so a 65th was dropped from the selection but stayed on:
Browse showed it as not added, share links left it out and Remove all
missed it. Start-up registered a share's entries the same way past the
cap. Add now refuses an entry a full library cannot keep, and start-up
registers only what the selection keeps.
2026-11-12 15:53:29 -08:00
Sameh Khamis e25cce874d Read an ArcGIS load stopped at maxFeatures as truncated
Paging stopped once maxFeatures were read before it noted that the
upstream had more, so a page holding exactly the cap with
exceededTransferLimit set read as the whole dataset. FEMA USA
Structures (2,000 features, one page) never showed PARTIAL in a dense
view. Stopping at the cap or the last page with more upstream now marks
the load truncated.
2026-11-12 15:52:31 -08:00
Sameh Khamis 134239fa5a Read a view over an incomplete GeoParquet index as PARTIAL
The browser's partition index kept whether it was complete only to
schedule a rebuild and handed the partitions on without it, so a view
counted unread partitions from an index that might be missing some and
could read as complete, even empty, over an area it had never indexed.
A view over an incomplete index now reads PARTIAL until a rebuild
completes.
2026-11-12 15:51:52 -08:00
Sameh Khamis e05a9566e4 Retry a GeoParquet footer whose body fails mid-stream
Only a failed fetch() marked a footer read as worth retrying. A body cut
off after its headers arrived (a reset, the proxy's 20-second timeout)
was taken for an unparseable file: the partition was left out and the
index still read as complete, so the server cached it without that
partition for the entry's cacheSeconds (24 hours for EPC). Such a read
is now retried like any fetch failure; an oversized body is still final.
2026-11-12 15:50:46 -08:00
Sameh Khamis 7d91a1875b Cancel a view-bounded load when the camera returns to the drawn view
Moving from view A to B started B's request; moving back to A before it
finished returned early because A was already drawn, so B's request ran
on and its features replaced A's while the camera was at A. Returning to
the drawn view now cancels the request in between.
2026-11-12 15:49:56 -08:00
Sameh Khamis 0f13c0a1c2 Let a newer share switch off a library layer still switching on
A share restore skipped every entry whose settled state already matched
the share, but a layer still switching on for an earlier share reads as
off until it settles. Opening a second share in that window left the
first share's layer to come on anyway. The restore now compares with the
manager's effective visibility (the newest request), so the newer share
supersedes the earlier one.
2026-11-12 15:49:11 -08:00
Sameh Khamis d58ba23fba Refresh live Data Library datasets once older than cacheSeconds
A dataset layer loaded its features once and kept them until it was
removed: the parsed set survives disable/enable, so NWS alerts, GDACS and
EONET showed what they held when first switched on for the whole session.

A library dataset now asks the manager for a refresh every cacheSeconds.
Once its data is that old it loads again, and swaps the new set in only
when it is ready, so the old features stay drawn while the request runs
and stay if it fails. An enable after cacheSeconds also loads afresh.
Disable and destroy cancel a refresh in flight.
2026-11-12 15:48:14 -08:00
Sameh Khamis e9085e8df3 feat(render): the Cesium renderer passes every conformance check a page can run
- Cesium's own tweens (its flights) step on the renderer's clock, so the
  flight traces compare frame by frame; dispose gives the scene its own
  clock back.
- moveStart and moveEnd are measured each tick on that clock, by the
  shared move detection, a change of the field of view or aspect
  included; Cesium's own events timed them by the system clock.
- A follow of a drawn item tracks where the draw port draws it.
- Pick rays are summed from the eye's axes: Cesium's lose about 1e-9 of
  direction at a near plane of 0.1 m. Ellipsoid picks and the view
  rectangle use them; a canvas with no size has none, as in Cesium.
- The harness always runs the virtual clock and claims 120 checks: 119
  pass on the page and query.heights skips. The five pointer traces
  need a camera controller there, and are recorded from Cesium's own
  controller in Node.
- The package boundaries list the cesium plugin export and the modules
  its groups now import.
2026-11-12 14:59:45 -08:00
Sameh Khamis 720932e1f9 test(render): the Cesium renderer claims five more conformance checks: lit shapes, a cancelled flight, flight callbacks, onCommand and camera.rect 2026-11-12 13:27:43 -08:00
Sameh Khamis 9fb39eb6da test(render): a GPU icon set's draw commands: translucent sets draw one blended pass in the opaque pass, LESS_OR_EQUAL with depth written; others an opaque and a translucent pass 2026-11-12 12:44:19 -08:00