The sprite and card were hidden from the bridge but the selected-vessel
trail was not: at sea it ran astern from under the eye, and on a raised
river it lay buried under the water. It is now hidden while the bridge
holds its ship, keeping its history, and shown again on release.
The ground clamp read the floor at the ship's last AIS fix while the camera
rode that fix dead-reckoned up to 180 s ahead, so a quay or bridge deck at
the old position could lift the eye. Vessels now read the floor at the
dead-reckoned target, as the water read already does; aircraft are
unchanged.
The radar centred on the ship's last AIS fix while the camera rode that fix
dead-reckoned up to 180 s ahead: at 20 kn about a nautical mile apart, so a
ship already passed still showed ahead. Own ship now sits where the camera
rides and each contact is dead-reckoned to the same moment; the query reaches
past the range for ships that have since moved into it.
Ships with no MMSI each get a key of their own instead of all sharing an
empty one on the scope.
A full-detail water read that sampled nothing (tiles not loaded) fell back
to the geoid, and while a full read ran, a loaded-tile read with one coarse
low probe did the same: on the Seine either dropped the eye ~27 m under the
water until the next read. An empty read is now no answer, and once a
full-detail read has answered only full-detail reads change the height.
The read sequence numbers go: only one full-detail read runs at a time,
so a late read can never be older than one already applied.
Voice entry tracked an aircraft first, which cleared the vessel selection
and ended the bridge, and then failed because Cockpit was still active:
no bridge, no cockpit, no selected ship, and voice said entry was
unavailable. It now refuses before it touches tracking and says to exit
the bridge first.
With a vessel selected and no aircraft tracked, Cockpit took C and stopped
it even when the bridge gate was shut, so the key did nothing and CCTV no
longer toggled. C is now taken only when the bridge can be entered.
The arrow keys also stay with comboboxes, spinbuttons, trees, grids,
menubars, scrollbars and text boxes on a bridge, which move with them.
Layer display tests no longer re-check Cesium primitives in every layer:
they run on the fake engine, and one Cesium translation check per ground
kind lives in the Cesium engine's ground tests. Their harnesses are shared,
and the fake-engine, Cesium-engine and flights tests that repeated the
contract suite are gone, 340 tests in all.
The press-travel measurement is shared by the selection dispatcher and
CCTV, which loses its ScreenSpaceEventHandler shim and world-click module.
ALPR, installations and bikeshare share one selection keeper and selected
marker write. Layers with their own click rules keep their handlers.
Navara's pulse is a class like the ground pass, the ground and imagery
passes share the surface uniforms, and the colour, log-depth and occlusion
GLSL lives in one chunk module, each caller keeping its depth bias. The
ground suite runs inside the contract suite (--only=ground:), and fixture
and script names say what they test.
Behaviour is unchanged on both engines.
Every layer draws on both engines, so the per-engine availability
machinery always answered yes. The renderer manifest is now the id to
display map plus the wind flow backend; the catalog filter, the voice and
QA-probe refusals and the engine-kind checks go with it. The Navara
notice lists the same layers in the same words.
Paths that read the Cesium viewer behind an engine that is always there
are gone: the Recent Imagery viewer fallbacks, the shell's imagery-host
service (one shared host module remains), the engineless map controller,
viewer gates in voice actions, world packets and camera verbs, the
orientation controls' tracked-entity fallback, and the Bhote Koshi and
FIRMS Cesium-shaped paths. setCameraInputs, which nothing called, leaves
the contract; suspensions remain.
Smaller duplicates go too: one tracker-release helper for navigation and
location, the no-op frame latch, missing-engine guards, one name for the
injected display, flat bikeshare and installations displays, one overlay
host and binder, one name per world-overlay method, and stub-viewer
guards in the Cesium engine. Layers that still read the viewer keep the
shared init(viewer, context) entry.
Behaviour is unchanged on both engines. Viewer lines fall from 849 to
750 and engine-name literals from 32 to 27.
Remove code that nothing in the app, the runners or CI uses:
- the one-off migration compare, look and bench scripts and the vector
ground prototype they measured, with the bench tables in the docs;
- the unused Navara-named shared render modules (polygons, ground picks,
mover bank, ground ring);
- the Cesium-shaped aircraft and vessel test renderers, and the trail,
engineFromViewer and horizon-occluder helpers only they kept alive.
Their tests now run on the production displays over the fake engine;
- Navara's unread diagnostics, fallback ports, dead exports and
parameters, stale effect-chain anchors and the unread capability flags.
The QA scripts share one keyless guard and readiness wait, the session
harness reuses the strict harness's browser flags and block list, and the
harness honours QA_ANGLE and QA_RESERVED_PORTS. The bikeshare fixture
keeps only what its scenario reads. The docs name the engine picker where
it is (VISUAL PRESETS) and drop dated run results.
No behaviour changes on either engine. Navara layer renderer lines fall
from 189 to 0 and viewer lines from 891 to 849.
Five places still read the Cesium viewer and so did nothing, or the wrong
thing, on Navara. Each now goes through the globe contract, and Cesium
behaves as before.
- The Navara ground floor stands aside while the camera inputs are
suspended, not only while they are switched off. The cockpit suspends
them, so its eye 12 m up no longer jumps to the floor's 30 m between
updates. A follow's own suspension keeps the floor on, as before.
- Follow state comes from engine.camera.following() and followChanged
instead of viewer.trackedEntity. Enabling CCTV during a follow no longer
flies away and ends it, a new follow stamps navigation, Director cards
wait while a subject is followed, and an explicit navigation stops the
follow on both engines.
- Navara registers every data credit, as Cesium does. The allowlist left
out the layers that moved to Navara later (CCTV agencies, street-level
imagery, GIBS, routing, bike feeds and others).
- Annotation links and embed mode attach on every engine; their camera
flights use engine.camera.flyTo. Only Cesium adds the timer that keeps
an inline panel drawing.
- The FPS readout counts engine.frames.postRender, so it shows on Navara.
Each fix has a test that fails without it.
The licence page refuses automated requests; its text was read in a browser.
Credit now uses the licence's own attribution statement.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Prague, Warsaw, Kraków (buses and trams), Poznań, Gdańsk, Rzeszów, Vilnius,
Bucharest, Cluj-Napoca, Iași, Rome, Turin, Venice (buses and waterbuses),
Bordeaux, Montpellier, Rouen, Toulon, Le Havre, Tours, Málaga, Lisbon's city
Carris, Sendai, Johor Bahru, Kota Bharu, Adelaide, Atlanta, Denver, Edmonton,
Halifax and Victoria.
Each entry records its licence, or that none is stated and the Metro
Transit/OVapi basis it ships on. Montpellier and Toulon publish km/h and are
converted. Route ids are read as the rider-facing line; Prague, Bordeaux and
Le Havre trams and Venice waterbuses take their own mode. Adelaide needs
`Accept: */*` to get binary protobuf. Every feed was fetched through the dev
proxy on 2026-11-08.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Keyless GTFS-Realtime feed from Carris Metropolitana, documented in the
operator's public API repository. No licence is stated for the realtime
endpoint (the static GTFS is CC BY 4.0), so it is credited as a courtesy on
the same basis as the Metro Transit and OVapi feeds. The feed publishes km/h,
so a new `speedUnit` registry field lets the proxy convert to m/s. Route ids
wrap the GTFS route id in an area prefix and are read as the line number.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Describe the radial disc envelope, corridor backing and reveal control,
rectangular input pick windows and optional horizon ellipsoid. Record the
launch-pad and Transit restoration and the Street Level and CCTV callers.
Explain synchronous Transit and reveal construction and surface readiness
for classified ground materials after camera changes.
Cesium remains the appearance and interaction reference; this documentation
changes no behaviour. All eleven coupling counts remain unchanged.
Draw launch-pad zones as radial ground discs with the original fill, rim,
transitions, emission and depth bias. Drape Transit history and its backing
on ground corridors. Keep the current age bands and once-per-second updates;
advance the active head with a reveal fraction and rebuild only at segment
crossings. Its wider dark edge and colour core blend once, as on Cesium.
Navara composes these looks after tone mapping.
Build reveal corridors and dynamic Transit history synchronously once shared
terrain bounds are loaded. Keep requesting frames until the visible globe
tiles and committed primitives are ready: classification needs current
surface depth after a camera move. Retain the reference radial material,
its alpha envelope and the head's material-only reveal updates.
Remove the empty Navara ground effect from the render chain when the last
shown set is hidden or destroyed. It otherwise remains as the insertion
anchor for later pulse effects after its packed textures have been disposed.
Cover reveal, destruction and replacement with fresh-state comparisons for
corridors, polygons, discs and cells. Cesium drawing is unchanged by this fix.
Keep the radial pixel check at a 500 metre radius and 4 kilometre view,
sample the brightest of six inner points around the native fill gap, and
sample across the rim width. Restore its view during arrival flights and
run it last among ground checks. Both suite launchers wait for the linked
view to arrive before checking. Keep these checks with the radial look.
Pass rectangular pick windows and drill limits through input subscriptions.
Street Level uses its 9 pixel window and CCTV uses its 14 pixel window and
six-result drill. Down picks stay synchronous, moves coalesce per frame and
Navara clicks keep the GPU answer from their gesture. Add a projector
ellipsoid option and use WGS84 radii reduced by 1,000 metres for Street Level.
These changes restore the original Cesium pad fill and ground trail rather
than the temporary ring and raised lines. Navara follows the same appearance
and horizon policy, and now honours explicit CPU pick windows. Existing
callers without these options retain their defaults. The four debts share
the contract, suite and engine implementations, so keep them together.
No coupling count grew or fell: renderer lines remain 0 and 189, external
Cesium imports 3, external viewer lines 891, engine-kind checks 1, engine
resolution calls 0, extension reads 3, ground passes 1, SDK loads 3, screen
handler files 4 and engine-name lines 32.
The contract and ground suites now wait until the camera has reached the
view in their link before checking, because the app's restore flight could
land in the middle of a check; that was also the cause of the intermittent
camera-events failure. The radial check uses a launch pad's size and samples
the fill at several inner points: Cesium's own old pad leaves the same gap in
a small ground disc's fill. The pulse-band check takes its base frame once
the surface has settled.
Checked in the browser, keyless: the contract suite passes on both engines
(84 and 85 checks), the ground suite passes on both, CCTV v2 passes 58
checks, and the space, transit, Street Level and CCTV engine switches pass.
The transit and Street Level scripts fail as they do before this change.
With no aircraft tracked, Cockpit can ride the selected AIS vessel. Entry
needs only Live AIS Vessels and a selected vessel; C, double-click and the
entry buttons (reading BRIDGE with a ship icon) enter it, while voice stays
aircraft-only.
The eye sits on the AIS position at a representative bridge height for the
vessel's type family, looks along true heading and moves along course over
ground. The last fix is dead-reckoned for up to 180 s between the layer's
refreshes, and on rivers and lakes the waterline is read from the rendered
surface around the ship rather than the geoid.
The bridge HUD shows speed and course over ground, a straight heading
tape, window mullions, the AIS destination as a reported voyage and a
heading-up radar of nearby ships (3 to 24 NM). The arrow keys look around,
Home or 0 looks ahead again. The vessel layer hides the ridden ship's
sprite and card, ignores scene clicks while held, and returns the camera
to the vessel's map view on exit.
Describe the segment batch lifecycle, per-item picks and styles, and the
single-primitive engine implementations. Document the restored Wildlife
fill and outline and replace the stale no-fill description.
The documentation follows Cesium's stem behavior and original draped
privacy appearance. No coupling count fell or grew; all eleven metrics
remain unchanged.
Draw distinct privacy cells in one ground set with taxon-coloured fills
at 25 percent opacity and two-pixel outlines at 80 percent opacity.
Restore the original Cesium ground appearance on both engines, replacing
the shared display's raised, depth-free outline. Ground classification
handles terrain, tiles and horizon occlusion; picks pass through.
Retain bounds-and-accent deduplication, unchanged-geometry reuse, selection,
wide-view removal and detach cleanup. Check the fill and outline on fake
and headless Cesium engines, and keep engine-switch ownership coverage.
Update the Wildlife switch check to read ground cells.
No coupling count fell or grew: all eleven metrics are unchanged, including
891 viewer lines and three external Cesium import files.
Checked in the browser, keyless: the Wildlife engine switch draws the cell
set on both engines; its keyboard step still meets the known Tab issue.
Add copied, committed segment batches to the contract, Cesium, Navara and
the fake engine. Keep each segment's colour, width and pick identity while
drawing one primitive or overlay mesh per display. Preserve Cesium's stem
sizes, ground sampling, visible cohorts, update cadence and click targets.
Navara draws the batch after tone mapping through the existing overlay.
Cover disconnected geometry, picks, hidden replacement, empty batches and
cleanup in the contract suite and engine tests. Keep the infrastructure
and cable display assertions on fake and headless Cesium engines, and
read batch counts and pick identities in the browser checks.
No coupling count fell or grew: all eleven metrics are unchanged, including
891 viewer lines, three external Cesium import files and 189 layer Navara
render lines.
Checked in the browser, keyless: the contract suite passes on both engines,
with the new segment picking and pixel checks; the infrastructure and
submarine-cable engine switches, the cable overlay and infrastructure LOD
scripts pass.
Exercise the real shell primitives and materials over the headless engine.
Cover raised rectangle geometry, alpha and depth policy, camera lift, shell
ordering, upload readiness, hidden frames, detail replacement and moved
windows, render idling and resource cleanup. Drive the public imagery API
through detail bounds and staged source settlement as well. Supply only the
geometry and texture completion signals that require a renderer.
Preserve Cesium's reference behavior without changing runtime code. Navara
and the fake engine are unchanged. Include the new test in the format scope.
All eleven coupling counts remain unchanged; none fell or grew. Renderer
lines remain 0 and 189, external Cesium import files 3, viewer lines 891,
Navara kind checks 1, engine-sniff calls 0, extension reads 3, ground passes
1, SDK loads 3, screen-handler files 4 and engine-name lines 32.
Move engine request ordering and the single rollback decision into the scene
host. Run handoff, renderer release, candidate acquisition, attachment,
restoration and settling in one transaction. Pass that transaction to nested
parts so unrelated calls wait without putting nested work behind itself.
Keep the engine-switch entry point as a delegating adapter.
Route direct shell attachment and detachment through the same host and remove
the shell queue. Serialize late part registration with attachment and release.
Keep failed-attachment cleanup and renderer release inside the transaction,
including failed restoration. Replacement applications share the transition
owner with fresh parts, and retired applications release their part registry.
A shell used alone creates its own scene host.
Preserve Cesium's picker and history ordering, single recovery, retained layer
session, cockpit restoration, hidden-tab settling and parked-renderer policy.
Navara keeps the same behavior. No user-visible change is intended on either
engine. Keep the existing behavioral assertions and cover concurrent operations,
failed cleanup followed by recovery, retired transaction handles, replacement
startup failure and parked expiry during a switch.
All eleven coupling counts remain unchanged; none fell or grew. Renderer lines
remain 0 and 189, external Cesium import files 3, viewer lines 891, Navara kind
checks 1, engine-sniff calls 0, extension reads 3, ground passes 1, SDK loads 3,
screen-handler files 4 and engine-name lines 32.
Checked on the keyless production build: the engine, Director and layer
handoff switches, app-once, and the rapid, hidden-tab, context-loss,
parked-release, failure-matrix and soak scenarios pass.
Main now ships an Iceland pack that loads the full live Vegagerðin list
(about 500 views) with its CC BY 4.0 credit. The World catalog's 166
Vegagerðin rows point at the same frames under different ids, so every one
of those cameras would be drawn twice. Drop them from the World export and
its loader policy; World is now 2,230 rows and the default catalog limit
7,350 (4,000 shared slots plus 3,350 reserved).
Register Data attribution entries for Ottawa, Québec, TripCheck and MLIT,
as every other CCTV pack has. MLIT's terms ask for the source in the form
"出典：国土交通省 川の防災情報" with the page link, and ODOT asks for
"Camera courtesy of ODOT"; TripCheck partner cameras keep their owner's
name in the label.
The Houston guard test from main checked the old 4,000 default ceiling;
check the 4,000 shared slots instead, like the other pack tests.
Give each scene attachment its own tracked readout, follow subscription,
selected context, active card identity and label-subject cache. Bind layer
context publishers to that attachment and preserve the page selection slot
and existing selection and awareness event payloads. Mirror a clear to the
page only when it clears the page selection. Capture display context
services so retired callbacks cannot publish into a replacement attachment.
Route aircraft, satellite, Local ADS-B and ALPR label refreshes and annotation
card lookup to the attached readout. Remove Local ADS-B contexts before
dropping its outgoing renderer, and mirror context pruning to the page.
Carry the retained page selection through the layer session when an attachment
is released. Adopt it only if the page still selects the same record, seed the
incoming readout on initialization, and replace its retained carrier when the
new display registers its marker. Keep the original page selection timestamp
and avoid a duplicate page selection event. A clear between attachments wins,
and page teardown still clears the selection. Seed the incoming readout from
that retained record without validating the outgoing display's visibility:
FIRMS waits for camera restoration before registering its new display. Keep a
matching selected fire in the refresh set before pruning top-N contexts, so
weaker selected detections retain the page selection too.
Preserve Cesium's label text, selected-card policy, visual anchors and
follow-camera anti-jitter. Navara retains the same shared behavior. There is
no intended user-visible change on either engine. Tests cover simultaneous
follows, identical selection IDs, independent refresh and clear, stale
notifications, destruction, parked reattachment and real layer publication
on the fake and headless Cesium engines. The production aircraft tests also
cover grounded fleet and tracked models, the display floor on both sides of
the zoom boundary, follow-label refresh and cached position reads. The fire
regression exercises the application scene handoff, delayed viewport resume,
top-N pruning, unchanged selection timestamp and deliberate clear on both
engines. These fixes preserve the reference Cesium selection behavior.
All eleven coupling counts remain unchanged; none fell or grew. The counts
remain 0 and 189 renderer lines, 3 external Cesium import files, 891 viewer
lines, 1 Navara kind check, 0 engine-sniff calls, 3 extension reads, 1 ground
pass, 3 SDK loads, 4 screen-handler files and 32 engine-name lines.
Checked in the browser, keyless: the overlay baselines on both engines, the
ALPR and FIRMS fire-camera switches, and the layer handoff, Local ADS-B,
Contacts, space, infrastructure, CCTV, urban, transit, directions and
annotation engine switches pass. The tracking regression passes every check
that passes before this change.
Move the world overlay into scene.services.overlay. Each attachment owns its
sources, collision domains, paint lanes, canvases, observers, timers and
accessible action subscribers. Keep both canvases inside the engine container,
with detection beside the card root so its screen blend still reaches the
scene. Release shared font measurements only after the last host detaches.
Resolve retained layer publications against their attached engine, clear the
outgoing source before releasing its renderer, and reconnect the shell's action
subscription on attachment. Keep destroyed hosts inert and give parked engines
fresh services. Bind annotation rectangles and the existing tracked bridge to
their engine's overlay. The tracked bridge itself remains page-owned.
Preserve Cesium's projection, card placement, arbitration, fades, hit tests,
blend order and render demand. Navara keeps the same shared behaviour. No
user-visible change is intended. Exercise simultaneous hosts, either teardown,
stale handles, reattachment and a retained selected card on the fake and
headless Cesium engines. Update QA reads and package ownership for the service.
No coupling metric falls or grows: renderer lines remain 0 Cesium and 189
Navara; external Cesium import files 3; viewer lines 891; Navara kind checks 1;
engine sniff calls 0; extension reads 3; ground passes 1; SDK loads 3;
ScreenSpaceEventHandler files 4; engine-name lines 32.
Each host now takes the page's overlay ids from the host it replaces and gives
them back to a remaining host when destroyed, and the chrome's static overlay
placeholders are gone. The new hosts had left the ids on those empty
placeholders, so the voice pointer policy, the id selectors and the QA probes
that read #world-overlay-canvas missed the drawn canvas.
Checked in the browser, keyless: the overlay baselines on both engines, CCTV
v2 (58 checks), military names, and the engine, voice, ground-layer and Ghost
Maps switches pass. The transit script fails the same 25 checks as before.
- Label repeat-collapse uses letter lookarounds instead of ASCII-only \b, so
'Þverá Árnessýslu' is no longer joined and 'séð séð' collapses.
- 'til austurs og vesturs' and 'í norður og suður' name two facings and keep
the low-confidence pose.
- Fallback name uses the frame stem when the station number is missing.
- Tests: label and heading regressions, dedupe keeps the last row, cap clamp,
high-confidence pose, frame redirect pin; file added to format scope.
- Credit names the dataset, links the terms and states no endorsement.
The event's before and after images went to each engine's display as URLs,
so every engine switch fetched and decoded both again: the strict Nepal event
scenario saw three reads of each image where Cesium's kept providers made one.
The event now decodes the pair once with its own loader and hands the images
to each display, which takes a URL or a decoded image. The pair is dropped
with the rest of the event state on disable, and a failed load is retried on
the next attachment.
The wind engine-switch scenario waited for a diagnostics field the wind layer
lost when it moved onto engine.imagery; it now waits for imageryActive.
Checked in the browser, keyless: the Nepal event, wind and submarine-cable
engine switches pass.
Give each scene its own render governor, owner holds and recent idle requests.
Make the Cesium render backend own its request mode directly, so the governor
drives engine.render on both engines without a dependency back from Cesium.
Route layer requests through their attached engine and capture the attachment's
governor for camera motions, orbit, annotations and shell controls. Release
retained-layer holds before clearing the outgoing engine, and destroy the
governor after scene consumers have detached.
Preserve Cesium's owner-keyed holds, pre-install buffering, last-release settling
frame and one-shot requests during continuous rendering. Navara keeps the same
render policy. There is no intended visible change on either engine. Simultaneous
scenes no longer share holds, and an old captured service cannot release a
replacement attachment's hold. Cover both rules on fake and headless Cesium
engines, including an old orbit ending after reattachment.
Viewer-reference lines fall from 910 to 891, Navara-kind checks from 2 to 1,
and engine-name lines from 33 to 32. All other coupling counts stay unchanged.
Checked in the browser, keyless: qa-perf passes on both engines with the
same idle, active and release counts as before (24 checks on Cesium, 22 on
Navara).
Create pick registries and selection dispatchers for each scene attachment,
publish them through scene.services, and release their predicates and input
subscriptions before parking the renderer. Resolve layer ownership with the
attached engine so identical layer IDs in simultaneous scenes cannot overwrite
or remove one another. Re-register retained ALPR and cyclone picks on attachment,
and resolve voice pointer picks against the engine that supplied the pick.
Preserve Cesium's selection, sibling ownership, inert trails, touch expansion,
drag and long-press rules on both engines. Extract the unchanged gesture
predicates so portable picking does not acquire a native input dependency.
Exercise scene isolation, teardown and reattachment on fake and headless Cesium
engines, retaining the existing layer assertions.
No coupling count falls or grows. Cesium and Navara renderer lines remain 0 and
189; external Cesium imports 3; viewer lines 910; Navara kind checks 2; engine
sniff calls 0; extension reads 3; ground passes 1; SDK loads 3; screen-handler
files 4; engine-name lines 33.
Restrict runtime SDK imports to their engine directories, including package
subpaths, re-exports and dynamic imports. Keep three measured, reasoned
exceptions for the Bhote Koshi property-record models and reject stale
exceptions. Move the Cesium annotation renderers and both wind adapters to
their engine folders, preserving their implementations and lazy loading.
Load all 35 manifest display entries on a fake engine with an unknown kind
and no native extensions. Draw small fixtures and verify resource and
subscription cleanup before destroying the engine. Name the three entries
without display factories as skips. Update the architecture, infrastructure
lifecycle and engine integration documentation to describe the code today.
Cesium behavior is unchanged, and Navara keeps its existing behavior. The
weather comparison harness supplies the Navara flow-mesh factory and accepts
the previous backend location for comparison runs. No user-visible strings
change.
Cesium layer-renderer lines fall from 14 to 0, Navara layer-renderer lines
from 314 to 189, outside-engine Cesium import files from 5 to 3, viewer lines
from 933 to 910, extension reads from 4 to 3, and engine-name lines from 36
to 33. No coupling count grows. Targeted checks pass: 845 tests, three named
rehearsal skips. Import directions, package boundaries, formatting and syntax
checks pass.
Checked in the browser, keyless: the contract suite passes on Navara and on
Cesium apart from the camera-events check that also fails intermittently
before; Navara annotations pass across engine switches; the ground-look
comparison is pixel-identical. qa-draw-tool's hidden-line check and the wind
engine switch fail the same way before this change.
Sameh Khamis