Each sprite layer's icon or dot set is now made at a draw order taken
from SPRITE_LAYER_ORDER through the renderer's per-set drawOrder:
flights over military, vessels, directions, transit, bikeshare, FIRMS
and CCTV, whichever layer was turned on first. Picks name the topmost
first. The orders are negative, so selection highlights, route
endpoints, the tracked plane and every other set at the default order
still draw over all sprite layers, as they did before.
The raise-to-top restorer, the sprite registry and the primitive-name
lookups are gone, which takes spriteOrder.js off the viewer-access list.
Before, the restorer raised the layers to the bottom of the scene in the
order they were enabled, so flights usually ended up under the others.
pickGroundPosition reads Cesium's depth and ellipsoid picks, so it moves
unchanged from scenePick into the listed input dispatcher, its only caller.
Weather rendering's default imagery host comes from the listed shell
rendering's globeImageryHost instead of reading the viewer's imagery layers.
The coverage lines are draped line sets on the draw port now, not
ground polyline primitives of the layer's own. A tile draws in sets of
120 line parts, as its primitives were batched; each part is an item of
its own (its id made unique in the tile) that picks its sequence
(`meta.pick`). A filter change redraws the sets in place, so the old
lines show until the new ones are ready. The selected sequence is one
set over the coverage, redrawn in place as tiles land; the tiles hide
their own lines of it, a show change made in place, so it shows
whichever set is drawn over which. The wait for the terrain-height
table goes: the renderer builds draped lines when it is ready. A colour
the layer cannot read is handed to the renderer as given.
Clicks no longer read the renderer's pick. The click dispatcher's event
gains `windowHit(width, height)`, the wider pick read as the canonical
id, the draw-set hit and whether it is the map itself. Street Level
routes its own sets' hits by their pick, anything else (a provider's own
drawing) by its id, as before; the ground under a click comes from the
event, and the ground click is told whether the map was hit. The screen
centre is aimed through the query port, a pick that fails counting as a
miss. Street Level no longer reads the scenePick service.
The Cesium draw port no longer lets a show queued while a draped line
built override one applied once it was ready, before the swap: the
highlight hand-over now depends on that.
lineCoverage leaves the Cesium allow-list; with it every Street Level
file named in this work is off the list.
Differences the contract does not let us avoid:
- The lines draw among the scene's primitives, not its ground
primitives: they now draw over other layers' clamped entities and
ground primitives, and win picks against them.
- The highlight builds asynchronously: a newly selected line is not
drawn for the frames that takes.
- Lines of other sequences in tiles that land after the highlight was
first drawn cross over it.
The photo marker and every provider's image cones are icon sets on the
draw port now, not billboard collections of their own. Each keeps its
pick id as the item id, so clicks route as before; each set is drawn
(empty) as soon as it is made, so the cones stay drawn over the marker
whichever is first given something to show. The horizon cull still runs
in the layer, against the ellipsoid lowered 1,000 m, and sends only the
items whose show changed; a quiet first cull of items about to be drawn
replaces the flags that held its callback back.
marker and sequenceCones leave the Cesium allow-list. The Panoramax line
click reads the camera height from the camera port.
A difference the contract does not let us avoid: the icons turn to their
bearing about local north (the contract's `course`), where the
billboards used the Earth's axis. Looking straight down nothing changes;
in a tilted view a cone now points along its true bearing on the ground
instead of leaning toward the pole.
The Street Level view maths no longer reads Cesium. The camera's place
comes from the camera port's pose; the visible box and the view focus
cast the query port's screen rays and meet them with the ground, the
WGS84 ellipsoid raised to the ground height, in geo; heights come from
the query port; the horizon cull for the marker and the cones tests
against an ellipsoid 1,000 m below WGS84 in geo and re-culls on the
frame port's pre-render. Ground clicks, the screen-centre aim, the
nearest-image point, photo framing, the coverage refresh and the ground
re-measure read the same ports and convert positions with geo.
geo's ellipsoid functions (toLonLatH, rayEllipsoid and those under them)
and horizonFrom take an optional ellipsoid, made with ellipsoidOf or
offsetEllipsoid; anything else, a map index included, means WGS84. Tests
check them against Cesium on raised and lowered ellipsoids.
view, index, cameraFraming, groundClick and selection leave the Cesium
allow-list. The Street Level entries are Cesium-free callers of the
unbound factories: given a viewer of a caller's own, with no renderer
bound, the caller loads ./render/cesium/unbound first. GhostMaps, which
uses the horizon cull and is still on the list, loads the camera factory
itself.
Differences the contract does not let us avoid:
- The ground under the camera no longer reads whether the globe is shown
(the globe port has no visibility getter). Where 3D tiles answer under
the camera they decide, as they did under Google 3D; elsewhere the
terrain answers, then the drawn surface below the camera. Over 3D tiles
shown on a map with the globe (and under other primitives at the
nadir) those now count as ground.
- Without depth-texture support the query port answers no heights, the
terrain included, so the ground then reads as the ellipsoid.
- selection and groundClick still read the click dispatcher's own pick
for now; they move to draw-set hits with the coverage lines.
The street-level provider credits, the map source credit, the shared
inline OpenStreetMap and OpenMapTiles credits and the ALPR provider credit
go through the renderer's credits port instead of Cesium's credit display.
Their markup has one link each and is read as the port's plain data; the
ALPR credit is built as plain data, which the port escapes. src/maps/credits.js
and src/layers/streetLevel/credits.js no longer reach Cesium and leave the
allow-list.
findCredits finds the port as findCamera finds the camera: a Cesium viewer
of a caller's own, with no renderer bound, gets a credits port over its
own credit display, so the public entries keep their attribution. A port
built that way is released, not disposed, when a renderer binds later, and
a credit let go after its viewer is destroyed leaves the display alone.
The per-layer data credits and the transit and event credits stay on
Cesium's display: their markup carries several links each, which the
contract's credit cannot hold.
The embed view, the celestial-ring focus, the ground-guard lift, the
Bhote Koshi approach and orbit, launch selection and replay flights, and
the cockpit's view and takeover now drive the renderer's camera port.
The contract gains a sineInOut easing, pinned by a conformance trace.
A viewer with no renderer bound, such as one a caller of a public
export brings, now gets a camera port too, as it already got an input
port: cameraOf builds one over the viewer's own Cesium camera and keeps
it for the viewer's life, until a renderer is bound. The Cesium adapter
provides the factory as it loads; code that reaches Cesium takes cameraOf
from the adapter so the factory comes with it, and findCamera answers
null where a missing camera is not an error (HUD, share link, world
focus, Recent Imagery's ZOOM IN). A test walks every package export and
fails one that can reach the bridge without loading the factory.
A destroyed viewer gets no camera and lets go of a built one, and the
application renderer stays bound until its viewer is destroyed.
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.
About 20 sites called scene.sampleHeight, globe.getHeight,
scene.clampToHeight or the MostDetailed variants directly, each with its own
idea of a miss (undefined, NaN, a throw). They now ask heightsFor(scene) in
src/queries/heights.js:
- now(point, { against, exclude }) answers from what is loaded, or null;
against is 'surface' (what is drawn) or 'terrain' (loaded terrain only).
- clampNow(position) moves a position onto the drawn surface, or null.
- batch() and clampBatch() wait for the most detailed tiles: null per
missed point, a rejection when the whole query fails.
- supported() reports { surface, clamp } without allocating.
Cesium still answers now() synchronously, so nothing changes on screen and
nothing is slower: in the live app 1,000 calls cost the same through the
service as direct (terrain 0.5-0.7 ms, surface 2-6 ms a call, run-to-run
noise larger than the difference). Every site already treated a miss as
"hold the last value or fall back"; a renderer that can only answer from a
cache or asynchronously can now plug in at one place.
- A Bhote Koshi evidence point whose terrain sample misses stays at its
fallback elevation instead of reading the miss as sea level.
- The mesh floor sampler skips a scene without depth-texture support
instead of spending its budget on probes that throw.
- The world annotation probe's trace reads hit/missed.
- A source scan forbids direct height calls outside the service.
The service is exported as ./queries/heights and documented in
docs/APPLICATION.md.
Every layer and tool bound its own ScreenSpaceEventHandler and picked the
same click again: with the usual layers on, one click ran about 26
scene.pick and 2 drillPick passes (120-215 ms in NYC with 18 layers). Pointer
input now comes in through src/input/dispatcher.js, one handler per scene.
A click is picked lazily and at most once per kind of query, and every
subscriber reads the same ClickEvent (pick, hit, id, ownerId, gesture,
drill, ground). The same click now costs 3 picks and 2 drills, 17-70 ms.
- Ambient onClick subscribers run only while no tool holds the pointer,
checked once per event; tools subscribe with their lease. Moves and
releases always reach ambient subscribers, so hover and drags end cleanly.
- Camera input and the viewer's stock select/track clicks are ref-counted
holds (suspendCamera, suppressStockClicks). The dispatcher owns the
controller's enableInputs, so a Cesium flight that ends no longer turns
input back on under a drag or the cockpit.
- The draw tool, imagery box, CCTV gizmo, cockpit, directions and every
selecting layer use it; bindTrackingClickGesture is gone (the gesture is
on the event). A source scan forbids new handlers, enableInputs writes and
stock-click removal outside the dispatcher.
- The sensor-style pick correction transit had now applies to every layer.
- Radio's 8-pick near-miss ring is one 17x17 windowed drill.
- A missed pick passed to pickContact is no longer picked again, and a
touch tap's contact search shares the click's drill.
- A gizmo drag whose release Cesium never reported ends on the next press
instead of leaving the camera held.
The dispatcher is exported as ./input with the pointer-lease functions, and
documented in docs/APPLICATION.md.
Findings from the astra, sol and fable reviews of the Arabic screenshots:
- Wording: forecast valid time is «عند», not an expiry; acres are آكر, not
feddans; live layers are «تتبّع … المباشر» (not "direct flights"); the
straight-ahead lane, turboprop, call sign, reconcile, proximity, retrying,
count agreement (colon form, fixed counts, ي for compact day ages) and
smaller fixes. Guide and glossary updated.
- Right to left: the scene editor's JSON fields are LTR; canvas map cards
draw with LTR geometry and an Arabic face; whiteboard callouts anchor
inside their cards; HDG/FOV keep their values; the cyclone timestamp no
longer splits its date; bare-number counters read 01 / 25, translated
ones say «من».
- Built-in shot titles of the Nepal shot pack's bootstrap shots localize
(the applied pack vouches for them), in every language.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* feat(street-level): export the KartaView and Panoramax providers
A host that composes Street Level with its own provider list can already
import the Mapillary and Google Street View providers; KartaView and Panoramax
were reachable only through the default list, which needs a Mapillary source.
Exporting them lets a host offer the keyless providers without Mapillary.
* chore: changelog entry for the Street Level provider exports
* feat(street-level): send Mapillary Graph lookups to a source endpoint
A Mapillary source can take a graphEndpoint, a Graph API base that adds
the token itself. Its sequence lookups go there through the source's
transport, and the photo viewer uses MapillaryJS's stock GraphDataProvider
at that endpoint, so a host that proxies Mapillary needs no client token in
the page. Without it, behaviour is unchanged: direct Graph API calls with
the client token.
An ESLint no-unused-vars scan over src, server, scripts, tools, build
and pinokio, repeated until it is stable, reported these.
Runtime modules keep every call and drop only the unread binding:
- src/ui/cockpitInput.js: the unread result of enter()/exit().
- src/layers/launches/rendering.js: the unread entity returned by
entities.add().
- src/layers/radio/rendering.js: the unread map key in the horizon scan,
which now iterates the map's values.
QA scripts and tools drop an unused helper, two unused in-page module
imports, unused counters and destructured fields, and an unread wait
result. Tests drop source files read into variables that no assertion
uses, redundant StyleManager imports (the shell-source helper already
imports it), and unused destructured fields.
The one remaining report is the `_` loop variable in src/tools/schema.js,
which counts iterations and is intentional.
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.
The slim-down in #974 dropped --sl-text-small, which .sl-since-note still
used, so with KartaView lit "KartaView: by upload date" rendered on its
own line larger than the value. The note now sits beside the value in the
same mono type and letter spacing, a size smaller (0.5rem), as
"(by upload date)". One source is lit at a time, so the provider name is
dropped. es/ru catalogs, tests, README, CURRENT-STATE and the registry
JSDoc follow.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Google 3D: the OPEN button re-measures a ground whose tiles are still
loading (a few times, a second apart) and on a map change, not only
after the next camera move.
- Mapillary: two mounts on one host during the first download share one
build instead of leaking a second WebGL viewer.
- Retired floating-panel keys (v8 panelPos, panelFloatHintShown) are
dropped from localStorage; dead inline-position resets, the drag JSDoc
and the dead qa-ui-disposal observer branch are gone.
- Docs: z ladder, panel chrome, position storage and the fixture gate's
actual coverage; changelog fragment for #974 (old .gev-cache/mapillary
can be deleted).
- ru/es: the layer is named as everywhere else in the UI.
- Fixtures and test fakes: unused exports removed; a fake credit removal
of an absent credit no longer drops another one; the tile test reset
frees upstream slots.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- The selected sequence is drawn past the per-tile caps again, and its
highlight takes every loaded part, drawn or not: in a dense tile a
clicked Mapillary or Panoramax line kept its cones but lost its line.
- photoViewer tries :fullscreen and :popover-open one at a time, each in
its own try: Safari before 16.4 parses neither, and the fallback threw
on every flat KartaView or Panoramax photo.
- A lit Street View off Google 3D says it needs the Google 3D map again,
instead of "every provider is switched off".
- Chips carry their provider colour again (--chip-color, as on main).
Each comes with the test that reproduced it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Main gained three imagery providers (Google Street View, KartaView,
Panoramax), shared line-coverage, cone and photo-viewer modules, one source
lit at a time and the translated interface. Their shared modules still
carried what this branch removes, so the merge ports them onto it:
- lineCoverage.js is this branch's coverage.js made provider-neutral
(draped lines near the ground only, the selection drawn over them, key
and rate-limit blocking); sequenceCones.js is sequences.js the same way.
Mapillary, KartaView and Panoramax all build on them; Mapillary moves
back to providers/mapillary/ beside the others.
- No ground casting, mesh sampling, surface modes, FOLLOW, legend, prewarm,
orbit overview points (Panoramax's density dots included), stale-tile
swaps or water-filled point budget come back.
- Kept from main because the new providers need them: requiresMapStack and
attachMapStackController (Street View on Google 3D only), ground clicks,
openNearest with an optional provider nearestImage, openImage {frame},
reportError, the view-cone marker (now an SVG glyph), filters /
coverageLabel / sinceBasis, OPEN STREET VIEW, keyAlternatives and i18n.
- photoViewer.js treats the viewer as expanded when its wrap is full screen
or the iPhone popover, as EXPAND now is.
- The fixture browser gate keeps its 8 steps and adds one each for
KartaView and Panoramax; catalogs, docs, boundaries and format scope
follow.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A superseded open wait cleared the shared cancelShown slot that belonged to the newer wait, leaving its auth listener and timer alive after close; open() now settles the previous wait first. A lookup whose viewer unmounted while the Street View service loaded still sent a billed getPanorama; describe() now rechecks the lookup after acquisition.
- i18n scanner: lastError sink, no false sinks on defaults and ternaries
- shared src/i18n/marker.js for modules that only mark strings
- wind field messages move to wind/fields.js; CITY_POIS imports from cityPois.js
- drop review IDs, change history and stale references from comments and docs
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Conflicts:
- KartaView viewer moved into the shared photoViewer.js on main; its
labels now take the provider name ("{name} street-level photo") and
are translated there.
- Weather: main's cloud filter strength slider, with its labels in t().
- Street Level caption: translated creator line plus main's licence.
Panoramax status and error texts are marked like KartaView's, and
"Panoramax answered HTTP {status}" gets a template in the panel and the
layer row. Catalogs: es/ru 2553 messages each, complete.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Fixes the 51 P0–P2 findings from the full review of the i18n branch and
the issues found while checking every new string on screen in ru and es.
- Translate errors and statuses that still reached ru/es in English:
feed and tile errors, provider statuses, the ALPR chip, the traffic
mode chip, voice plan step labels, fire feed values, receiver bands,
scene save reasons and the Cesium attribution link.
- Show preset city names in the UI language (src/ui/placeNames.js);
data and everything sent to the model stay English.
- Share layer names between the layer panel, HUD and voice card
(src/ui/layerNames.js).
- Keep long ru/es strings whole: voice card error lines, analyst
titles, the voice tier status, the Nepal evidence button, the ALPR
retry chip. English layout is unchanged.
- Ask before a language switch drops unsaved work.
- Catalog fixes: glossary terms (пуск, пресет), gender agreement and
consistent "peticiones" in es, fewer calques in both languages.
- Fix English text that was itself wrong (e.g. "Exit cockpit to fly to
a weather", raw tool ids in voice plan steps).
Catalogs: es/ru 2535 messages each, complete.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Image loading (photoViewer.js)
- One AbortController per open: a new open, close or unmount stops the
previous photo's first and sharper downloads (an <img> by src = '').
A collapsed viewer never falls back to the original; Panoramax's flat
fallback is the thumbnail. A superseded open resolves quietly.
- Flat HD tracked as "loading" and "shown" apart: a discarded response no
longer blocks later upgrades, and resizes while it loads ask once.
- No upgrade of the photo still shown while another opens.
- Size limits before decoding: imageSize.js reads JPEG/PNG/WebP/GIF
headers; more than 64 MP or 16,384 px a side is refused undecoded, an
<img> that decodes larger is refused and let go, flat HD needs a known
width of at most 8,192 px, and panoramas always decode with a resize.
Panoramax provider
- Retry-After parsed as seconds or an HTTP date; the browser cannot read
it here, so busy answers back off 5 s doubling to 300 s, reset once
answered.
- Headers timed, then the body by an idle timeout (15 s of silence), so a
slow link still gets its megabyte tiles.
- Cones centre only on the picture shown (onShowing), never on a late
lookup; lookups nobody waits for are aborted (shared, refcounted).
- onReopen: a GPU reset redraws in the viewer, not through the core.
- Cones draw the home z15 tile at once and add neighbours as they arrive;
a neighbour that failed is not counted as drawn.
- Coverage tiles decode in ~8 ms slices (a 2 MB z10 tile over Paris was
one ~85-200 ms task); sorted by measured length without a Map.
- Box search asks for 20, not 100 (the API returns them nearest first).
- looksLikeEmail checks every '@'; creatorOf de-duplicates with a Set.
lineCoverage.js
- A redraw while a swap still builds keeps what is on screen (the swap's
old lines and its ready new ones) and drops only the unready ones.
Tests for each (mutation-checked), two vacuous tests made real, the
cones' pictures lane and readable()'s busy/timeout paths covered; the
fixture's box search is nearest first. Docs: README filters and share
links, CHANGELOG HD wording, CURRENT-STATE.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From the maintainer review of #1134 (approve with follow-ups):
- lineCoverage: a share is judged against the cap the tile was last drawn
with, not against later shares that redrew nothing, so small increases
add up (a retained tile stayed at 100 points after its share grew to
116). A rebuild's measuring pass no longer counts as a draw.
- photoViewer: a picture refused for its size (bytes, streamed bytes or
decoded side) is marked `tooLarge`, and the viewer never loads that file
again by any path; the flat fallback had retried an oversized SD file
through an <img>, which has no size limit.
- Regressions for both: cumulative share growth below the threshold, and a
refused SD panorama that is not fetched again as a flat photo.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Conflicts resolved by keeping both sides: the Events group gains
gnss-integrity with a translated label, the cockpit aircraft line shows
main's transponder emergency through a localized formatter (cockpitMath.js
stays i18n-free, like the Context scope line), and the boundary lists keep
both the i18n modules and main's new data modules.
Main's new UI is translated into es and ru: the GNSS Integrity row, legend
and notes; geo file import (button, hints, list actions, every import
error, with numbers passed as placeholders); transponder status names.
Also formats kartaview/viewer.js so format:check passes.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
CHANGELOG: keep both new [Unreleased] entries (Panoramax, and the
emergency-transponder highlight from #856).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Adds a case that fails before de-duplication (a 24-slot overview held repeats
and reported itself truncated) and says why the date-line halves overlap.
This branch moved twice at once: upstream merged KartaView (#1121, a
third keyless street-level provider sharing a new lineCoverage.js), and
the remote i18n-spanish gained the whole-interface translation work
(1993 catalog entries, translateKnown, data-i18n-runtime, the unmarked-
string detector in npm run i18n) built on the Street View merge.
Integration keeps the remote's mechanisms and re-adds KartaView on top:
- presentMeta: remote's translateHint for coverage hints, plus the
KartaView count line as its own literal tn pair (Russian plurals need
the noun inside the template; keyed off coverageLabel.noun with the
sequences pair as the default).
- uiState: upstream's highlightsSequence legend gate (KartaView's lines
do not highlight) with t('Selected').
- layer-panels.html: remote's data-i18n/data-i18n-runtime hooks plus the
sl-since-note wiring KartaView's upload-date SINCE note needs.
- kartaview viewer: the photo-navigation aria-label and img alt go
through t(), as the remote's unmarked-string detector requires.
- catalogs: remote's 1987-entry base (its refined keys supersede this
branch's earlier Street View wording) plus the KartaView entries:
coverage hints, the upload-date note, the streets count line (with
Russian one/few/many/other forms) and the viewer strings.
npm run i18n: complete, no unmarked strings. npm test: 6520/6522 (the
one failure is the pre-existing environmental OPENAI_API_KEY leak in
devFreshDotenv). Build passes.
Upstream's KartaView PR (#1121) extracted the Mapillary coverage engine
into a shared lineCoverage.js and made the panel wording
provider-specific (coverage noun/action, SINCE upload-date note,
highlightsSequence legend gate), again in the regions this branch
localizes.
- mapillary/coverage.js: taken as upstream's thin re-export of the new
lineCoverage.js; the generic too-high hint stays msg()-marked, now in
lineCoverage.js where the default lives.
- kartaview/policy.js: coverage hints (tooHigh, noCoverage) marked as
msg() data so they translate like every other coverage hint; thrown
error messages stay English, matching the Mapillary precedent.
- presentMeta: KartaView counts streets, and Russian plurals need the
noun inside the template, so the count line gets its own literal tn
pair keyed off coverageLabel.noun, falling back to the sequences pair;
upstream's noun/action data contract is untouched.
- uiState legend: upstream highlightsSequence gate with t('Selected');
sinceNote localized with a {names} placeholder.
- layer-panels.html: sl-since-note wiring kept, data-i18n hooks kept.
- CHANGELOG: both sides kept. es/ru: 4 new keys each.
npm run i18n complete (953 messages each); npm test 6504/6506 (the one
failure is the pre-existing environmental OPENAI_API_KEY leak in
devFreshDotenv, unrelated); build passes.
The round-3 verification found the shared point budget rebuilding a dense
tile by removing its lines before the new ones were built, so central
Paris vanished for 1–2 s after every pan:
- Tiles redrawn for a new share, or for a selected line the cap had left
out, go through the swap path: the old lines stay until the new ones
are ready.
- A share must move by 10% to redraw a tile (a 1.6% change rebuilt it).
- The selected line, admitted past the cap, is not counted against it,
so its tile no longer redraws on every settle.
- A selection the imagery filter hides is not redrawn for.
- Shares are recomputed when a refresh ends loading without a tile
finishing.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
KartaView's later viewer changes are carried into the shared
photoViewer.js: the camera bearing (`bearingOf`), redrawing the same
photo after a GPU reset through `onReopen`, no pose while a lost context
recovers, the 2,560 px flat version only once the viewer outgrows
1,280 px, `refreshPose`, and one sharper-panorama load at a time with a
second look once it lands. Its tile cache's `keyOf` and `clear` move to
the shared tileCache.js. KartaView gets `onEscape` beside `onReopen`.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From the round-2 verification (approve with follow-ups), all follow-ups:
- The point budget is shared across the tiles in view once loading
settles (water-filling): a sparse tile keeps what it has, a dense city
centre takes the rest instead of drawing thinner than its suburbs. The
selected sequence is always drawn, and selecting one a cap had left out
draws it.
- The cones' picture-list cache keeps only finished lists: a re-centre
while the cones loaded inherited the aborted load's tiles and drew
none in them.
- Douglas–Peucker measures to the segment, so dead ends and U-turns keep
their far end.
- Switching Panoramax off lets its decoded tiles and lists go.
- Picture downloads are counted while they stream (a host sending no
length cannot fill memory before the cap).
- Comments on the via caption and the simplification tolerance match the
code.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Esc moved to the viewer's box (`.sl-viewer-wrap`) in the last commit, but
without one it was a second keydown listener on the root; Esc is now taken
by the root's own handler there. format-scope.json listed
panorama.test.mjs twice, which made the formatter refuse the scope; the
files it skipped are formatted.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From a five-lens IC8 review (correctness, security, tests with mutation
testing, performance, UX):
- P0: over Paris at 12–60 km the unsimplified z10–z12 lines sent ~2M
points to Cesium's ground-polyline builder (gigabytes of geometry; the
tab hung). decode.js now simplifies lines (Douglas–Peucker, 4 tile
units: 6–10% of the points), and lineCoverage gains an opt-in point
budget (`maxPoints`, 10,000 a tile for Panoramax).
- Decoded coverage tiles are kept (KartaView's tile cache moves to the
shared tileCache.js); a sequence's z15 picture lists are kept and load
on their own connection; cones know their tiles from the start of a
load, so a picture opening meanwhile re-centres them and ‹ › no longer
restarts the load.
- The HD panorama loads only for the expanded viewer (a docked one on a
dense screen asked for 3 MB originals) and is aborted when another
photo opens; picture files over 48 MB or 16,384 px are refused; a
bitmap no renderer took after a context loss is closed.
- A nearest lookup survives one of its two searches failing; only a
timeout is retried; offline reads as unreachable; Retry-After is
capped at five minutes; the search box stays on the map at ±180°;
cones near both edges of a narrow (high-latitude) tile read both.
- Every producer is credited (app and photographer); with none to name,
the caption says "via <instance>" rather than crediting the host.
- Esc anywhere in the docked viewer (toolbar, caption) clears the
sequence before the panel takes it. A filter that hides every density
dot says so; wording aligned ("coverage", "photo").
- Tests for every surviving mutant (Esc consumed, keepOrder and dot-size
wiring, licence in the core, limiter aborts, radius floor, bbox corner,
first-producer-only, byte and size caps), a limiter test, and a gate
fixture that answers box searches and leaves out pictures facing away;
the gate now clicks over the other sequence's pictures, so only the
line click's own-sequence lookup can pass.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
KartaView's server filters by upload date only, so SINCE goes by it
throughout for KartaView: the lines (as before), the photo a click opens
and the cones, the cut-off taken as its UTC day as KartaView's startDate
(passesKartaViewFilter). Providers may declare sinceBasis: 'upload'
(validated); while one is lit the panel shows "KartaView: by upload date"
under the SINCE slider, read out with it. The gate checks the note, and
step 13's picks get 60 s on a loaded machine.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
KartaView filters coverage on its server only in its PNG tiles (its JSON
tiles come back empty with a filter), so a filtered tile's PNG is a mask
over the cached JSON segments: a segment stays when at least half the
points along it lie on drawn pixels. A filter change reloads the lines
(lineCoverage.reload keeps the drawn ones until the new ones are in);
masks are cached per filter and tile, and a filtered tile KartaView is
still drawing (504 or a timeout) is asked again, four tries in all.
The nearest lookup asks the photo API for the projection and date, and
the cones follow the filter like Mapillary's. The gate fixture draws
filtered PNG tiles and adds an older flat-only street; a new gate step
checks 360 and LAST YEAR thin KartaView's lines and ALL brings them back.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From a UX QA round in the browser against the live API:
- A line click could time out while coverage loaded: the meta-catalogue
speaks HTTP/1.1 only (six connections), and six tiles plus four
searches queued the lookup behind megabyte tiles. Tiles now take
three connections and searches two; a search that timed out is asked
once more.
- Between 30 and ~100 km Paris read as empty: the density grid is ~10 km
a cell. Lines now draw to 60 km (z10), as Mapillary's do; above, dots
grow with the pictures they stand for.
- Esc in the docked viewer collapsed the whole panel instead of clearing
the sequence the panel says it clears (shared photo viewer: KartaView
too). The viewer now offers Esc to the provider first; expanded, the
dialog keeps it.
- A picture whose producer is an email address is credited to the
instance holding it (its `via` host), so Licence Ouverte pictures name
their source.
- Empty states say why: "no Panoramax photos here match the filters"
when the filter hid every line, and nothing to zoom into over empty
sea. Wording follows KartaView's ("photos", "click a line for its
nearest photo").
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From an adversarial functional QA round against the live API:
- The nearest lookup missed pictures looking away from the click
(place_position leaves them out at any tolerance): it now runs a
bounding-box search beside it and ranks both by distance. A line
clicked from high up searches its sequence within height / 100
(50–400 m).
- The per-tile line cap kept the newest sequences, not the longest
(lineCoverage re-sorted them): providers can now keep their decoded
order (`keepOrder`).
- A long producer name froze the tab (quadratic email regex): linear
check. A malformed link escape no longer fails the lookup; null or
string coordinates no longer land at 0,0.
- A line with no camera type or day stays drawn under the filters.
- A sky click on a line changes nothing (as KartaView), instead of
selecting a sequence with no cones.
- An unreadable tile reads as such; a failed neighbour z15 tile only
thins the cones; unknown or refused ids say the picture is gone.
- Shared viewer: the sharp-panorama guard is per open (a late download
of the previous photo no longer lets a second one start); an HD-only
360° picture is shrunk or skipped like any original; panorama.js read
the bitmap size after closing it, so 2:1 power-of-two panoramas
clamped instead of wrapping.
- Tests for each, and for the gaps mutation testing found (preferred
point, cone order, no reload within drawn tiles, newest-first gate,
stale bitmap release, flat upgrade retry).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Panoramax, the open federated street-level commons, registers beside
Mapillary, Google Street View and KartaView. Its meta-catalogue
(api.panoramax.xyz) is keyless and CORS-open, so it runs in the browser:
- coverage from its vector tiles: sequence lines at z11–14 below 30 km
(longest first under the per-tile cap), and above that the grid layer's
density cells as binned points, not counted as sequences;
- a click on a line opens that sequence's picture nearest the click
(STAC search, any direction), without moving the camera, then any
sequence's if it has none within 50 m; the 360° and since filters go to
the lines and to the API;
- cones from the z15 tiles' pictures layer around the open picture
(~110 bytes a picture, against ~7.5 kB for a STAC item), re-centred when
a picture opens outside them; ‹ › follow the picture's prev/next links;
- the caption names the author (not an account named by an email
address) and the picture's licence (StreetPose `license`), and links to
the picture in the Panoramax viewer; the globe credits Panoramax.
KartaView's viewer becomes the shared photoViewer.js (flat <img>, 360° on
panorama.js, ‹ ›, keyboard, pinch, context loss), driven by a per-provider
PhotoSources table; the request limiter moves to limiter.js. Panoramax shows
the 2,048 px SD derivative and fetches the HD original only when the viewer
is wide enough, once, shrunk while it decodes (never above 8,192 px).
Orange #ff6f00 (Panoramax's own), off by default, share option
0.m.0_0.r.panoramax-1. The MCP panel allows the meta-catalogue and the
biggest instances. Tests: policy, decode, source, viewer, provider, and a
Panoramax fixture in the hermetic browser gate.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Viewer: resizes while a sharper photo loads start no second load (the
last size is looked at once it lands); decoded bitmaps nobody will draw
(an overtaken preview or upgrade) are closed.
- Cones: a photo opened while the drive's first page loads is the one
they centre on.
- A sequence page that lands while a lost WebGL context recovers keeps
its cones: no pose is emitted without a renderer, and a viewer error
never rejects page data.
- Source: an answer turned away unread (HTTP error, 429, length past the
cap) aborts its request and cancels its body, freeing the connection.
- Gate: findPickPoint no longer picks each centre five times at radius 0,
aims step 13's cone pick at the drawn billboards, and caps each
evaluation at 4 s, resuming from a cursor on the next poll, so a slow
machine cannot spend a whole wait in one evaluation.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Marker and direction (reported on the PR):
- KartaView has its own marker, a purple view cone (Street View's is
blue), not GEV's default pin that Mapillary uses.
- The camera direction (marker, cones, FOLLOW, a panorama's centre) is
the direction of travel to the next photo a few metres on. KartaView's
heading is the device compass: whole drives read 0 (no compass) and
others sit a steady ~110° off. Until the drive loads the compass stands
in (0 = unknown), and the pose is sent again once the course is known.
Remaining P4s from the review of #1121:
- Segment chaining is linear (a per-point cursor) and capped at 20,000
segments a tile and 5,000 points a line: a hostile tile of 60k segments
sharing one point took 1.5 s, now tens of ms.
- Decoded tiles are dropped when KartaView is switched off.
- No network / a refused request says so and is not retried as a timeout;
an unreadable answer says so.
- A docked flat photo skips the 2,560 px download until the viewer grows.
- The panel counts the drive's photos (countActivePhotos), not the thinned
page of cones (2,250, not 217).
- After a GPU reset the viewer redraws the photo itself, so a lookup the
user started is not cancelled; a failure is still reported.
- The gate's keyless page lights KartaView and checks it draws with no key;
the fixture answers the sequence count; the line-click step checks the
camera looks along the drive when the compass says 200°.
- README/CHANGELOG show the share-link form the app writes.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From the IC8 review of #1121 (2 P2, 7 P3; each with a test that fails on
the old code):
P2
- A WebGL context lost while a newer panorama loaded left a black viewer
and a 'successful' open: the panorama now builds on the new canvas (or
falls back to its front view).
- The streaming body read every real answer takes was tested with one
chunk only: answers split mid-character are reassembled; the shared test
fake now streams, so every source test runs the production path.
P3
- Lighting another source by its chip (params) now stops a running lookup
and clears its spinner and error, as setProviderEnabled did; a restore
that changes nothing leaves them alone.
- Panoramas decode off the main thread (ImageBitmap, RGBA upload: ~90 ms →
~15 ms at 4,096 px on a phone-class CPU) and are 2,048 px docked, 4,096
once the viewer grows; a closed viewer keeps a 1×1 texture.
- The lit KARTAVIEW chip label is #8f8aff (~6:1; #635bff was 3.7:1).
- The viewer's keyboard focus ring is drawn above the photo.
- Tests: a second context loss and a drag on the new canvas; the look
direction kept across a loss; a page's 30 s first try.
- The browser gate loses a real WebGL context in the 360° step and checks
the 429 lookup was asked again (the fixture counter is reset first).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From the second IC8 pass (launch ready; P3s closed here):
- A lost WebGL context is redrawn on a new canvas (the old one only gives
back its lost context), through the core, and only if no newer photo is
opening; verified live with WEBGL_lose_context, twice.
- The 16 MB answer cap counts bytes while reading: KartaView sends
chunked answers with no Content-Length, so the header check never fired.
- A lookup gets 8 s, then 15 s once more: KartaView sometimes sits on a
lookup for 11 s and answers the same one in 3 s.
Left as follow-ups: cancelling abandoned sequence pages; finding a page
by binary search on the index when thousands of photos are missing.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A mutation review (91, then 130 mutants) found tests that could not fail
(quick presses with synchronous fakes, a no-flight check with a provider
that never posed), and untested paths. New and rewritten tests kill 123 of
130 mutants; the 7 survivors are equivalent or harmless redraws.
- viewer: held-back loads for quick presses, stale sharp/flat loads,
close mid-load, ends checks for an older photo, WebGL retried per mount,
focus handed back, look direction kept on ›.
- provider: id-ordered pages with out-of-order indexes, page and probe
failures retried, overtaken lookups, cones re-centred on ‹ ›, leftover
imagery filters hide no line or cone, coverage hints, tile cache.
- source: 429 backoff and Retry-After under mock timers, hang timeouts,
the nearest query's parameters. policy: storage-host injection, FOV 360,
longitudes past ±180. panorama: a stub WebGL context (angles, lost
context, re-measuring). lineCoverage.test.mjs: tooHigh hint, passes().
- core: openImage({frame:false}) and openNearest leave the camera
(with a posing provider and a default-open control), KartaView's legend
and wording end to end; viewerHost supportsRenderModes.
- Fixed tick-count waits replaced by an until() poll.
- Browser gate fixture: odd photos archived (409 at storage2, served by
the proxy), the first lookup rate-limited (429, Retry-After: 0); the
line-click step asserts the camera stays and › skips the archived photo.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Screens that only open with provider keys showed a few gaps:
- Google Street View's own controls (Map Data, Terms, Report a problem)
and its addresses were always English: the Maps script now loads with
the UI language. English adds no parameter, as before.
- The selected-vessel HUD line kept KT and DEG after translated labels.
- Layer counts used a decimal point (12.0K); other languages use their
own separator (12,0K).
- A collapsed rail title was cut with an ellipsis in Russian; tighter
tracking fits it whole. English is unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>