- src/geo drops the ported debug argument checks and the tests that
bitExact already covers; package boundaries accept a directory entry,
so src/geo/ is listed once per group.
- Cesium no longer carries movers and lines nobody draws with: they are
optional engine capabilities. The fake engine and the Cesium camera use
the shared cameraMath; unused contract tables, clock.ticked, the
'finest' surface detail and Navara click statistics are gone.
- ALPR keeps one overlay core and one presentation model for both
engines, bikeshare one station bookkeeping, the Navara ground effects
one definition helper; image atlases, view centres, mover banks and
colours are shared.
- The legacy switch route, the clock offset and rate controls and other
unused exports are removed.
- The acceptance runs the Navara draw, geo import and scope checks;
scenarios use the harness switch and page helpers; the main chunk
check keeps the size budget only and runs once in CI.
The page creates the voice once; each engine attachment only gives it that scene's action runner, pointer canvas and annotation events. Calls made during a switch wait for the incoming engine and are refused with ENGINE_SWITCHING when none attaches in time; results name the engine that answered, and the session hears engine-changed after a switch. The analyst's follow-up memory and layer enable times outlive one engine's runner. scripts/qa-voice-engine-switch.mjs drives a fake microphone and Realtime transport through the switches.
Preserve the camera, map source and layer selection across engine sessions. Restore the previous engine if startup fails and retain local history navigation.
Knip, configured with the application, package exports, scripts, QA
harnesses, workers, tools and tests as entry points, reports exports
that no module imports. Each was checked by hand against string-based
and namespace imports.
- Symbols that are still used inside their own module lose only the
`export` keyword (176 constants and helpers).
- Unused re-exports are removed from barrel modules (flight, military,
vessel, CCTV, directions, transit history, local ADS-B and street-level
policy modules).
- Declarations that nothing uses at all are deleted, among them
AIS_WATCHDOG_STATUSES, applyFilter, isRadioCountryCode, compactText,
DENSITY_STOPS, the unused CCTV endpoint constants, and the SDR band
constants.
- Default exports that nothing imports are removed. The ones in
src/data/earthquakes.js, src/data/telegeographySubmarineCables.js,
src/data/wind.js and src/data/bhoteKoshiLocator.js each constructed a
layer instance at import time that nothing used; those instances are
no longer created.
Every export of a package.json entry stays, as does TOMTOM_CREDIT,
which the traffic layer reads from the credits namespace it is given.
in push-to-talk, answer a Gemini tool call that arrives with no input, reply
or tool result outstanding with a failure instead of running it, and refuse
every other call in the reply it starts, including one after unprompted
speech; the reply stays tracked so the next input retires it and its late
boundary cannot close the user's turn
in push-to-talk, a caption that arrives after its turn ended no longer opens
a reply, so the next request is not retired with its tool calls discarded
open mic cannot tell either case from a new reply and stays unchanged
check packet ownership after the narration callback before queuing tools,
and raise the per-connection call history to 4096 IDs
move the pointer turn event, context item and announcement into
pointerContext so the OpenAI and Gemini controllers share them, and drop the
Gemini token route's copy of the shared exact-Origin comparison
The voice runner drives the globe engine (an `engine` option, by default
the one registered for the viewer) instead of the Cesium viewer:
track_entity, stop_tracking (camera.following().stop()), frame_overhead
(camera.flyToSphere with the pose heading), fly_to_location, zoom_to_globe
(follow released through camera.following()), adjust_camera_zoom,
next_iss_pass / next_satellite_pass, select_nearest_aircraft and the
analyst view context, get_current_view_state, get_entity_context
(projector.worldToWindow), the view target (picker.pickPosition, then
camera.pickEllipsoid, then picker.pickGlobe) and its moveEnd prewarm, and
the viewport samples. getBasemapLabelContext takes the engine or a viewer
with a registered engine (the HUD).
Without a viewer the tools that still need one (VIEWER_ONLY_ACTIONS:
control_scene, annotate_map, clear_annotations, fly_route, move_camera,
control_cockpit, set_context_mode, control_radio, control_cctv) answer a
structured refusal (code UNAVAILABLE_ON_ENGINE) instead of throwing.
Point-and-ask picks with engine.picker and keys the camera pose from
engine.camera.pose(); the viewport capture reads engine.host.canvas and
waits for the engine's postRender (both engines keep their drawing
buffer: Cesium's contextOptions, Navara's own WebGL2 canvas, S4).
Tests: gevActionsEngine.test.mjs runs each moved action on the fake
engine. gevActions.test.mjs keeps its expectations; its runner helper now
registers an engine for the test viewer (the Cesium engine over it, or a
fake engine at its positionCartographic) and the two harness cameras gain
roll and frustum for camera.pose().
- Voice layer manifest (src/voice/layerManifest.js) generates the layer enums,
aliases and analyst field hints; a test fails when a catalog layer has no entry.
- analyst_query validates fields, operators, units, values and scopes, counts up
to 250,000 loaded records per layer, reports complete/partial/unanswered, tells
LAYER_OFF / NOT_READY / FEED_UNAVAILABLE apart from zero, refuses mismatched
follow-ups and never lets a superseded query replace follow-up memory. Seven
more layers expose analyst records.
- Tool results carry a code-built say line and a display card; the voice card in
the mic control shows captions, plan steps and the result
(src/voice/resultDisplay.js is the one adapter). Slow tools narrate their
current step out of band; holding Space while the assistant talks cuts it off.
Turn spans go to the debug log; caption transcription is metered; the log
omits what was said unless GEV_VOICE_LOG_CONTENT=1.
- Point and ask: the Space-down pointer snapshot resolves "this", "that one" and
"around here" through 'pointer' targets, and "the second one" through
referent:n from the last numbered list.
- Prompt states each policy once (instructions + tools 48.5k -> 46.7k chars).
- Voice benchmark harness (scripts/qa-voice-bench.mjs, scripts/voice-bench/)
shares its phrase table and grading with scripts/qa-voice-routing.mjs.