The satellites layer now sorts every object into Stations, Imaging,
Telecom, Starlink, Probes or Debris, each with a chip and a live count on
the layer row. Stations and Imaging show by default; the Starlink chip is
the former DENSE catalog switch. Hidden classes keep their points, so a
toggle needs no reload, and they leave the detection overlay and the voice
position lists.
A class comes, strongest evidence first, from CelesTrak SATCAT (object type
and status, served by a new /api/celestrak-satcat/<group> proxy route and
cached for a day), the bundled UCS Satellite Database pack (operator, owner
country and purpose), name patterns for later launches, and the CelesTrak
group. The core catalog also reads the resource, weather, planet, sar and
science groups. Payloads with no evidence of life launched 20 or more years
ago are filed as debris. The UCS pack, its build script and its data credit
are added.
An application built with fewer layers no longer restores, stores or shares a
layer it does not have when a link or saved state from a build that has it
names that layer. LayerStateCoordinator drops unregistered enabled-layer ids
before normalizing, so dependent options (a tracked selection) go with them.
An application with every layer registered is unchanged.
Both reviews of the timeout found that Cesium cannot cancel a full-detail
read: each one given up kept its nine picks traversing every frame, so
stalled tiles piled up a new set every 30 s, while slow but valid answers
were dropped. Raise-only loaded reads then locked in a wrong-high height.
Back to one full-detail read at a time whose answer always applies, and
loaded reads that may adjust a river height but never take it to the sea;
a coarse probe can again lower it partway until the full read lands.
Keeps the test for another ship selected inside the double-click memory,
and corrects CURRENT-STATE: a failed read keeps the height it had.
Before the first full-detail read lands, a loaded-tile read could still
lower a river height partway (27 m to a 15 m bank). Loaded-tile reads now
only raise it; lowering is the full-detail read's to say.
That makes a full read that never lands (stalled tile requests) hold a
wrong height for the whole bridge, so one still running after 30 s is given
up for a fresh one; its late answer is dropped.
Also tests the other half of the double-click memory fix: another ship
selected within the memory is not entered through the first ship's spot.
Before the first full-detail water read lands (it can take many seconds
from cold tiles), a refresh's loaded-tile read with one low probe still
dropped a river height to the geoid. Before that landing, loaded-tile reads
may now find raised water but only a full-detail read lowers it.
CURRENT-STATE also says what the earlier fixes changed: C stays the CCTV
shortcut when the bridge cannot be entered, voice entry is refused on a
bridge, and the ridden ship's trail is hidden with its sprite.
The vessel layer draws other ships at their last AIS fix and does not
advance them, so dead-reckoning radar contacts moved the mismatch from own
ship to everyone else: a 15 kn ferry reported 90 s ago showed ~700 m ahead
of the ship out of the window. Own ship stays centred where the camera
rides; contacts go back to their last fixes, and the query drops the extra
reach and candidates (which could also cut the nearest ship in a crowd).
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.
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]>
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.
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.
One JSON list from gagnaveita.vegagerdin.is (about 500 camera views at roughly
170 road stations). Frames are JPEG stills in a single directory on
www.vegagerdin.is; only links inside that directory are registered, and the
media proxy pins the same directory.
The list decides ids, coordinates and which frame path is proxied, so it is
fetched over https with redirects refused and the body read capped at 2 MiB;
every rejection path releases the response body. Camera ids come from the frame
file name, which is unique per view in the live feed, and repeated frames
collapse before the cap.
The feed's view description (`Skyring`) is the label. An explicit Icelandic
facing ("til vesturs", "í suður", "séð austur") sets a high-confidence heading
(264 of 499 views); a view that states none, or two different ones, keeps the
id-hash low-confidence pose. A bare road direction ("Miklabraut austur") is not
read as a facing.
IRCA publishes road camera images under CC BY 4.0 and asks that the licensor is
acknowledged without implying endorsement; the attribution is in the Data
attribution popover and DATA_SOURCES.md, and every camera carries the licence
string. Default cap is 250 with anchors at Reykjavík and four regional centres so
a cap keeps every region. CCTV_ICELAND_ENABLED=0 turns the pack off.
Verified against the live feed through fetchCctvImageFromUpstream: 499 of 499
frames returned a JPEG, all distinct, all under 20 minutes old.
createTrafficSource({ roadModes }) lets a host offer a subset of the TomTom,
OSM and Hybrid road modes (all three by default). Without OSM the layer never
draws simulated roads, and a single mode shows no road chip.
While flow is live the Street Traffic row offers Lines (every road with flow
as a steady green/amber/red line, from the heat-line renderer) and Dots (the
moving dots), saved in layer state and share links. Defaults: Lines off,
Dots on.
A throw from the crop capture now has a test that it closes its decision with an error outcome. Direct calls to the crop and image senders pin the channel-closed and pointer-scale-not-useful skips. The docs say a send decision requests an image, and when a pointer snapshot counts as already requested.
Review fixes on top of #1578:
- Exiting Space Missions cancels the layer's own camera flights again (entry,
roster preview, mission focus); only a flight navigation started after
releaseCameraOwnership() runs on.
- A failed first index load says so: "Space Missions unavailable: the
mission index could not load. Try again later." on both the Context and
Data Layers routes, instead of a generic transition toast. Translated.
- Regression tests for both; docs and changelog describe the real feedback.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Third review pass on #835. No runtime change.
- Controller test: viewport decisions reach the controller's debug log.
- Policy test: a second request for a snapshot still being captured is
logged pointer-already-requested and is not captured twice.
- CURRENT-STATE: pointer-not-fresh also covers a missing snapshot,
already-requested covers an earlier attempt that threw, and the image
step runs when any tool output of the response was sent.
- Changelog: describe what is recorded rather than a fixed miscount.
Second review pass on #835's diagnostics. Behaviour is unchanged.
- Report viewport_context.result as soon as the send returns, so a throw
in the retained-frame bookkeeping cannot log a real send as "error".
- List the decision reasons and when each field is present in
CURRENT-STATE, note that the rate limit can split a pair, and say which
codes are defensive.
- Changelog: the decision is logged when an entity-context answer reaches
the image step, not for every answer.
Review fixes on top of #1601:
- Only redirect a provider the params switch on to a keyless source, so a
filter-only setParams (360°, SINCE) no longer moves an explicitly chosen
keyless Mapillary to KartaView.
- Keyless QA gate: assert the new KartaView default and that the Mapillary
chip explains its key instead of switching (it used to expect KEY REQUIRED).
- Tests: Graph-proxy keyReady, filter-only updates; shared test helper.
- README and CURRENT-STATE describe the keyless default.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Review follow-up for #835. What is captured, sent and kept is unchanged.
- Log nothing for tool results other than get_entity_context, so other
tools stop spending the shared debug-log budget.
- Give every decision and outcome the same contextId and image kind, so
overlapping requests pair up.
- Log why a crop decision did not send (stale, off ground, off screen,
expired, pose changed, already requested) instead of leaving a lone
"pointer-ground-crop" that reads as sent.
- Split capture-invalidated into session-reset, superseded, channel-lost
and frame-stale.
- Close the pair with "error" when a step throws, then rethrow.
- Document the events in CURRENT-STATE and reword the changelog fragment.
- Keep the open card across a widened reload only when its index still
holds the same report; the server may answer a pinned version with newer
bytes.
- Narrowing to a window the loaded data already covers clears a failed
wider reload, so the row no longer stays in error.
- Label the every-age option "All" (contextual) so the six-option switch
fits the row in every locale; tooltip and info line keep "All markers".
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Review fixes on top of #1528:
- Narrowing the age window only removes markers; a product loaded with a
narrower window is reloaded when the window widens again. Fixes markers
missing after 7d -> 1d -> 7d and avoids reparsing every KMZ on narrowing.
- A wider window over the same edition keeps the open report card.
- Date-only event dates cover their whole UTC day, so 1d still shows
yesterday's reports.
- The age options are one labelled "Event age" radio switch (row
`segmented` control, as FIRMS confidence) instead of toggle chips mixed
with the product chips.
- Un-nest the per-product age test; QA compares data-source counts with the
live defaults instead of a hard-coded 2.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Publish classificationName next to the raw NHC code; unknown codes stay
unexpanded.
- Publish staleCacheFetchedAt only while the proxy serves its cache.
- Touch the shared store on clear only after this layer published, so a
feed failure no longer evicts another layer's dangling selection.
- Drop the unreachable no-storm branch and the unused return value.
- Tests: split the heading fixture into due north and stationary, cover
the unavailable and failure paths with the real store, and check that
clear/disable leave another layer's selection alone; fold the fake-store
test into the real-store one.
Keyless GTFS-Realtime feed published by ZDiTM Szczecin under CC0 1.0.
Speeds are m/s (checked against vehicle movement), the file refreshes about
every 20 s, and the feed carries no route ids, so vehicles take the default
bus mode.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
Denis Groshev