Arabic catalog strings carry optional marks ("تتبّع"), but people type
without them, so "تتبع" found no suggestion. Fold the marks out of both
the draft and the candidates before comparing.
Brings the Active Fires intensity work: the FRP filter, the confidence
switch, the ember-to-white-hot ramp and the far-zoom heatmap.
The filter, switch and ramp work on both engines. The heatmap drapes on
Cesium through the layer's Cesium renderer, which also fades the markers
across the hand-off; it goes with that renderer on an engine switch and is
built again on the next Cesium globe. Navara keeps heat cells for the far
bands, now coloured on the FRP ramp.
Main's new layers (drone airspace, OSM places, sea marks, ski pistes) are
Cesium-only for now: they join the layer directory and the Cesium catalog,
are built afresh for each Cesium catalog, and show as unavailable on Navara.
The zoom-out guard is a scene part that attaches to Cesium globes only. The
awareness range pulses and their echoes draw on Cesium through the awareness
display's viewer; Navara keeps the plain range ring. Camera-input wakes go
through the display, and the analyst keeps the session-owned memory with
main's area presentation. A disposed voice runner aborts work in flight, and
an engine switch still keeps the last answer.
Modules main added that read Cesium while they load (zoomOutGuard,
awarenessEcho, the ground-sampling scratch in localGeojsonCore) now build
those values on first use or use src/geo, so a Navara page still starts
without the SDK. The OSM view box reads the engine's view rectangle.
Main grew its own main chunk by 93,873 bytes gzip-9 since the merge base, so
the baseline is raised to the merged size.
- Host test switches AI Provider from chat Settings through the real
provider commands and checks the new mount keeps chat's context, frame,
Back to chat and focus on the provider choice.
- Drop the open and minimized branches of the carried presentation: the
provider choice lives only in Settings, so chat is never open then.
- One showChatSettings helper for the gear and the restore, and one
isStoppable helper for the live-or-error check in chat and the panel close.
Split the Cockpit camera update into the aircraft's own front half and a
shared placeCamera that reads the eye's forward and up offsets, pitch and
ground clearance from a profile. The aircraft profile carries the existing
constants, so the aircraft view is unchanged; other subjects can now supply
their own eye.
Pointing the scaled uniforms at the scaler's own vectors (bc1b56b6) left the atmosphere unscaled: takram's materials read the AtmosphereParameters.DEFAULT vectors themselves, so the frame dimmed (luma 161 against 214) and qa-navara-bloom found nothing to bloom. The scaler writes those vectors in place again, and a page-wide map keeps each one's value from before it was first scaled: a globe built later on the page takes its base from there instead of from the scaled value. Bloom measures as before (214.4 → 240.3), and qa-strict-context-loss passes 3 of 3 under eight busy threads.
The sky, aerial perspective and clouds materials hold the vectors of takram's AtmosphereParameters.DEFAULT, one per page. The look's scaler multiplied them in place and took its base from the first value it saw, so a Navara globe built later on the same page (after the parked one lost its WebGL context) could record an already scaled value as its base and draw the atmosphere scale² bright: a washed-out, nearly uniform frame. The fourth case of qa-strict-context-loss failed this way under CPU load (3 of 3 with eight busy threads; 3 of 3 pass now). Scaled uniforms now point at the scaler's own vectors, which it puts back each frame if a material wrote into them; the page's vectors are never written.
Resolves the DATA_SOURCES.md conflict by keeping main's rewritten Transit
and Radio bullets and restating the count for this branch: eighteen feeds
covering seventeen regions (the two Rio feeds share one region).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Since the weather layer stopped handing Navara the Cesium global, the shared shell read ContextLimits, Math and Rectangle from an undefined SDK and drew nothing on Navara (qa-weather-engine-switch timed out). On Navara the texture limit stays unset, as Cesium's never-created context left it.
Switching AI Provider in Settings opened from chat rebuilt the voice
control, so Settings fell back to voice and the dock went idle, widening
the wings. The switch now carries chat's Settings, or chat itself, to the
new provider as it already carries the input mode; the conversation still
starts fresh and focus stays on the provider choice.
After a voice error, opening chat or closing the voice panel left the dock
red on ERROR, because both stopped only an active session. They now also
end a session left in error, which returns the dock to idle.
The quick commands "Reset to globe view" and "Show earthquakes" reached t()
through a .map(), so the catalog check never saw them and every language
showed them in English. They are now marked with msg() and translated.
Review of the 71 strings from #1507 in all eleven languages:
- zh-Hant is regenerated from zh-CN with OpenCC s2tw, as its guide says;
40 entries had Taiwan vocabulary instead.
- ar "Dock to the right" names the left side, where RTL puts that rail.
- ja "What am I looking at?" no longer asks what the assistant sees.
- Provider Settings, dock, provider, live and map use each glossary's terms.
- de drops the du-form; ja drops spaces between Japanese and Latin text.
- Calques, register and the example commands read naturally.
The long soak's heap slope counted the per-aircraft dead-reckoning history (POSITION_HISTORY_LIMIT fixes for 10 000 aircraft, about 14 MiB over the first 2.5 min of polls) as switch growth: a run with the same workload and no switches grows as much and then levels off. With --assert the soak now waits until the flights feed has been read more times than the history keeps before it switches; over 500 switches the heap then grows 1.07 KiB per switch.
qa-strict-context-loss records, only when no non-blank frame came, how many frames the running and the watched engine drew in 3 s and how much of the last one differs from its first pixel.
index.html no longer loads /cesium/Cesium.js and widgets.css: the Cesium scene loads both before it builds its viewer. A cold Navara start requests nothing of the Cesium library again, and qa-strict-cold-chunks asserts that without the known-gap allowance. The strict harness puts an accessor in place of the Cesium global on every page and fails a scenario that reads it before Cesium.js has run; GEV_QA_CESIUM_TRAP=record lists every such read site instead.
Every Cesium read that a Navara page made was surveyed with the strict acceptance (GEV_QA_CESIUM_TRAP=record). Math on those paths now uses src/geo: screen annotations, weather and wind center readings, cyclone picks, scene data pack polygons and images (new navaraGeometry.js), the awareness subject, the orbit offset of camera verbs, voice frame_overhead, the agent world packet and Directions colours and ground picks. The celestial ring's Sun and Moon come from a bit-identical port of Cesium's Simon1994 model, JulianDate and TEME rotation (src/geo/simon1994.js, julianDate.js, transforms.js); a Cesium globe still applies its ICRF rotation once loaded.
Traffic, Street Level, the Nepal event and locator and geo import build on Cesium's math or entity model on either globe; their enable (or the import) now loads the SDK first through src/app/loadCesium.js, synchronously when it is already there. A production page guards the global so a missed read is reported and starts the load. No module reads Cesium while it loads (scripts/check-cesium-load-time.mjs, run by npm test). Cesium.js is still loaded by index.html here.
A poll that returned the same stale snapshot dropped STALE from the
Contacts cohort until the next answer, because the in-flight loading flag
outranked stale and the source revision ignored the provider name. A
source that has answered now reads its status without the loading flag;
one that has never answered (AIS first connect) still reads as loading,
so it shows no STALE before any snapshot exists.
A new voice fly-to now records the last flight that moved as the one it replaced, landed or not; only a still-flying one passes on its manual-input count. A slow lookup cancelled by a direct flight that landed, followed by a lookup that found nothing, used to drop the queued motion, and a cancelled direct flight in the same place let it fall back to an older landing.
A motion queued behind a voice fly-to acts on the last flight that moved the camera. A later lookup that was not found, threw, or was cancelled once a newer voice fly-to had begun moved nothing, so the motion looks past it rather than refusing or running early and cancelling the flight still in the air. It never acts at a flight older than the one it started waiting on. A lookup cancelled while still the latest voice flight (a search, a share link, Cockpit) still refuses.
The top-down no-op now reports the camera's real pitch.
A motion queued behind a voice fly-to now follows a later fly-to in the same reply whether the earlier one arrived, was cancelled or failed. A geocode that loses navigation reports failed, so the motion used to be dropped; and a flight left flying while the next place is looked up could land and run the motion there, taking navigation and cancelling the newer flight. The deadline, abort and stale-reply checks still end the wait.
pitchDeg -90 or -89 from a top-down view is now unchanged rather than a 1° tilt up, and the tilt tests' pitch loops are bounded.
Asking from the chat composer to open Voice Settings or Voice Preview
closed chat around the focused input, leaving focus on a hidden control.
Focus held in chat now moves to the requested view's tab, as the panel's
own tabs move it; focus anywhere else stays put.
Three 40 ms actions in one round outlast a 50 ms deadline unless each
action re-arms it, so removing that re-arm now fails the test for both
providers.
The conversation is kept to its newest 60 messages, so once full it no
longer grows as a message is sent, and a reader who had scrolled up was
left behind by their own request. The view now spots a send by the newest
user message's id instead of the list's length.
The attach menu promised up to 10 KB per file, but the attachments share
one 10 KB budget, so two 6 KB files were refused. The hint now says the
limit covers them all, in every locale.
With Voice Preview off the voice panel never opens on its own, and chat
Settings hid the voice preferences, so the Preview toggle, inactivity
timeout and model tier could not be reached by pointer. Chat Settings now
show them below the chat options, still in chat's frame and without the
voice tabs.
set_panel_open for voice-settings or voice-preview reported success while
chat's open state kept the voice panel hidden, minimized chat included. A
successful explicit show now closes chat, which keeps the typed session,
its running turn and its messages; reopening chat restores them.
The streamed oversized body was invalid JSON, so the loader returned
nothing whether or not the byte cap applied. Pad a valid catalog with a
usable camera instead, so only the cap can refuse it.
TranStar's Content Usage terms prohibit copying, distribution and
external framing of its images without TxDOT's prior written
permission. The opt-in text only said the pack was off unless
CCTV_HOUSTON_ENABLED=1, without saying what an operator needs before
setting it. Say "enable only if you hold TxDOT's written permission" in
.env.example, the README, DATA_SOURCES.md, CURRENT-STATE and the
changelog, matching the NYC and Florida packs, and fix the column of the
.env.example comment.
The houston-cctv credit now reads "(Houston, when enabled)" and links
the Content Usage terms, like the Florida credit. Where the docs said
frames are "never stored or redistributed", they now say what happens:
frames are relayed on demand, buffered only per request, sent with
no-store and not archived.
TranStar's `dir` field is the side of the freeway a camera sits on, not
the way it faces. TranStar's own map script (translateCameraDirection)
shows it as "Traffic closest to the camera is traveling": South is
eastbound, North westbound, West southbound and anything else
northbound. The live catalog agrees: I-10 cameras are North/South and
I-45 cameras East/West.
Treating `dir` as a high-confidence facing pointed most Houston cameras
across or away from the road they watch, and dropped the "estimated"
marking. Map it to the travel direction instead (S 90, N 270, W 180,
E 0) as a low-confidence prior with the low-confidence pose, and keep
the id-hash fallback for None or unknown values.
German Khokhlov