The data refresh is committed and its provenance (extract, timestamp, MD5,
filter) stays in the dataset README. The script needed osmium-tool and could
only re-verify while Geofabrik still listed that extract, so it goes with its
test.
Wildlife Sightings (#1591) published share token 9, so News takes the
next free token, 00. The token tests read the ledger before both where
they were written against it, and the README counts both layers.
A slow re-read of a refused key read as unavailable until it landed; it
now stays refused. The first-read wait is 10 s, for a margin under the
browser's 15 s. A failed Conflicts read no longer stops Local's refresh,
which the server reads on its own. Tests cover GDELT's wait, the refused
key and a caller abort past the timeout, and fail rather than hang; the
docs say what a section serves meanwhile.
A pooled paint item reused by another card keeps a fresh frame stamp, so
rows for cards that stopped painting survived every prune and the index
grew for the whole session. The prune now also drops a row whose item
paints a different entry id or source, and a test pins the bound.
A country's first World News API read, or Liveuamap's, could take up to
20 s while the browser waits 15, so the panel showed the browser's own
"signal timed out". A read with nothing to serve yet is now waited for
at most 12 s, as GDELT's already was: past that the section answers
unavailable while the read goes on, and the next refresh is served from
it. A client timeout reads "News feed unavailable".
Main's wildlife-sightings layer took share-link token 9, so shipwrecks
takes the next free token, 00. Locale, docs, format-scope and voice
manifest conflicts keep both sides. Ledger and registry tests now model
the ledger before shipwrecks took 00, and layer counts include both
new layers.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
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]>
A failed flow tile now carries the server's reason. A hosted deployment that limits each account answers USAGE_LIMIT_REACHED or RATE_LIMITED, which the traffic panel labels as the monthly allowance or a rate limit; a 503 with another reason reads as the traffic service being unavailable, not a missing key. The local proxy's no_key and daily budget labels are unchanged.
flyToBounds takes an easing, as flyTo does: in the contract, the Cesium
port and the reference recorder, pinned by a new conformance trace.
The flights of radio, awareness, CCTV, submarine cables, cyclones, the
launch preview, installations, ALPR, street-level framing, recent
imagery, weather and world focus now go through the renderer's camera
port. Street-level framing cancels its own flight handle rather than
whatever flight is current. worldFocus.js no longer reaches Cesium and
leaves the allow-list.
Layer pose reads stay for each layer's own slice. The local GeoJSON
flight stays on Cesium: gods-eye-view/infrastructure is used with plain
Cesium viewers that have no renderer bound.
The Cesium camera port moves into render/cesium/camera.js, with the
listener list and position conversion shared from common.js. It now
covers flights (flyTo, flyToBounds, cancelFlight, flying), lookAt and
release, command hooks, and a flightEnd event for the flights it starts.
A flight's end that Cesium raises inside one of the port's own calls is
delivered once Cesium's call returns, so its callback sees the flight
let go of. A flight Cesium declines to start, mid scene-mode morph,
ends as cancelled instead of never settling.
Flight traces run in Node on the recorder's headless camera with a
virtual clock; the browser harness keeps the instant traces.
The startup fly-in, landmark and viewport flights, share-link restore
and the shell's camera state now go through the port. camera.js and
sharelink.js leave the Cesium allow-list.
The first camera-family slice. The camera port's events move into a
shared module (render/camera/events.js) that the stub and the Cesium
port both use: listener lists, delivery with errors reported, and the
contract's per-listener `changed` measure.
The Cesium port gains:
- camera.on: `changed` measured per listener on every tick, where Cesium
measures its single threshold (the counts match Cesium's own at each
threshold); moveStart and moveEnd relayed from Cesium's camera;
- camera.viewRectangle, in degrees (now in CESIUM_CHECKS);
- lon and lat in pose, and NaN where Cesium has no cartographic position.
Code that still holds the viewer reaches the renderer through
rendererOf(viewer); the application renderer binds it on create and
unbinds on destroy.
The HUD reads its telemetry from the camera port and no longer imports
Cesium (off the allow-list). The share link writes the hash from the
pose. Its restore path stays on the viewer's camera for a later slice.
Behaviour changes:
- The share link listens to `changed` at its own 0.5 threshold, not at
whatever global threshold layers had set, and also to moveEnd, so the
hash always catches up with where the camera comes to rest.
- The HUD's and share link's camera callbacks run from the port's single
relay, ahead of other moveStart/moveEnd listeners, and a throw in one
is reported to the renderer's error listeners instead of escaping from
Cesium's render.
The Cesium renderer now creates the input port and exposes it as
renderer.input. The input dispatcher becomes an adapter over the port:
one port listener per event type, fanned out to its subscribers with the
ownership check read once per event, keeping the legacy event shapes.
- Ctrl+wheel zoom is switched on through renderer.input.ctrlWheelZoom.
- A dispatcher made for a viewer with no port creates its own port with
takeStock: false, which borrows the stock select and double-click
follow only while they are suppressed.
- The stock actions see the press as it began: a click whose press
started while select or follow was held off never runs them.
- A scene accepts one input port; cesiumInputOf(scene) returns it.
- Releasing one mouse button while another is held no longer gives the
held button a click. Cesium still ends that press.
src/render/cesium/input.js feeds the shared router from one
ScreenSpaceEventHandler: every button's press, release and click, double
clicks, moves and pinches, under any one modifier. The DOM event being
dispatched gives the full modifiers and pointer type; an action Cesium
raises on a timer takes its own modifier. Wheel comes from the canvas in
lines and is never cancelled. Clicks are picked once, lazily, through the
query port. Cesium ends every held button on any release; those presses
end there, but only the button released clicks.
The viewer's stock select and follow move off its handler and run first,
as they did, but only for a press no lease took and while not suppressed.
Camera holds go through the scene's shared holds, ctrlWheelZoom through
ctrlWheelZoom.js. The router gains a before-hook and dispose, which stops
delivery mid-event.
The conformance renderer exposes the port and the harness feeds it;
input.lease, input.lease.grab, input.lease.press, input.clickPick and
input.behaviours join the Cesium checks.
installTrackpadPinchZoom and its pinch scaling move unchanged from
src/app/viewer.js to src/render/cesium/ctrlWheelZoom.js, where the input
port's ctrlWheelZoom will turn it on and off. The application scene
imports it from there; the application/viewer export no longer carries
this undocumented helper.
The take-over of the camera controller's enableInputs moves, unchanged,
to src/render/cesium/cameraInput.js. acquireCameraInput(scene) gives every
user one set of counted holds per scene, so the renderer's input and
camera ports can hold camera input alongside the dispatcher; the
controller gets its own property back when the last user releases. The
dispatcher acquires its share once its handler is bound. The input guard
names the one module allowed to write the flag, and the new render files
join the format scope.
Frame-phase and draw-set ready listeners now run in subscription order,
skip one removed during delivery, defer one added during it, and hand a
throw to the renderer's error listeners (or reportError) instead of into
Cesium's render loop. The renderer gains on('error' | 'contextLost' |
'contextRestored'); Cesium cannot restore a lost context, so it never sends
contextRestored. Dispose clears every list. renderer.listeners joins the
Cesium checks.
The stub's lease and press routing moves to src/render/input/router.js,
unchanged, so a real renderer can feed it its own engine's events. The
router also keeps the behaviour holds (telling the renderer when one turns
off or on, unless it lacks it) and ctrlWheelZoom, and dispatch says who an
event went to.
The governor no longer writes the Cesium scene's render mode itself: each
owner holds the renderer's frame port once, requests go to it, and the
port is the one place the loop's mode is set. The port now also keeps
the loop from redrawing for scene time alone, which the governor did.
The application installs the governor as soon as its renderer exists and
holds the loop continuous ('boot') until its tools are up, where the
governor used to be installed, so boot draws as before. The hidden-tab
suspension goes through the port too, and the application's renderer now
has the whole frame port.
The frame port now does everything the contract asks of it on Cesium:
preUpdate says whether the tick draws (Cesium's own render test, read
live until the tick is over), holds are counted per reason, and suspend,
target fps, renderNow and diagnostics are real. Releasing the last hold
asks for one more frame.
whenSettled also waits until Cesium stops redrawing on its own. Each
request or worker task Cesium completes queues a redraw (a terrain
picker build after the camera moved, say), which no loading flag shows,
so it now waits for quiet frames and then a short stretch in which the
loop draws nothing by itself. It is bounded by its timeout, names a
stopped loop, and ends on destroy.
The app renderer still leaves out the members that steer the loop:
the render governor and the page's visibility handling own those
settings until they move onto the port.
The new frame.prediction check compares the preUpdate answer with what
each tick drew.
From 400 km, a tilt to -5° that a zoom out x2.5 caught 70 frames in eased down from the camera's old pitch at the new range. That put the camera below the place's horizon: the first frame's crosshair hit the ground 186 km from London, and London stayed hidden for about 11 frames. The re-solve also left the pivot pitch at 0°, so a zoom in before the next frame computed an infinite clearance floor, was refused, and stayed at 1,000 km.
The re-solved pivot pitch now stops at the place's horizon limit, or where the pivot already was. A camera that the zoom leaves past that limit steps down to it with the zoom and holds it, and the zoom still reports that angle. A zoom in straight after brings the range back to 400 km.
The voice runner subscribes to camera handoffs as it is created. The Cyber sonar test builds its runner on a StyleManager made from the prototype, which has no navigation owner, so the facade's delegation threw and both Node jobs failed. The facade now returns nothing there, like its other optional owners; the real shell builds its navigation owner before voice creates a runner.
One error was chosen before each failure's delay was known. An antimeridian
box 400 hid the other box's 503 Retry-After; a grid tile 400 with Retry-After
set a cooldown through the snapshot metadata; a short Retry-After on one tile
undercut a sibling 5xx's default backoff; and a huge Retry-After made the
deadline Infinity.
Every failed request now keeps its status and Retry-After. One function gives
each failure's delay: 4xx other than 429 none, 429 and 5xx their Retry-After
or the default backoff. The layer waits for the longest, and shows the
failure that set it. Retry-After is capped at one day, the proxy's own
daily-budget deadline, and non-finite values are ignored. Each publish of
one acquisition derives the default backoff from the backoff before it
started, so repeated tile publishes no longer double it.