Not one of HSR's 80 route ids equals the number on the bus. Hamilton was
registered without a routeLabel, so every vehicle in the city read as
"Route 5783" where a rider would say 5 DELAWARE — 228 of them in an
afternoon sample. The mechanism to fix it already existed and simply was
never pointed at this feed.
Two details the table cannot skip:
Hamilton runs two generations of route id at once. `5688` and `5780` are
both the 2 BARTON, and both arrive in the live feed, so both are listed.
That is why 80 ids collapse to 46 distinct routes.
The seasonal and event shuttles publish a name where the local routes
publish a number — TC230, PEACH, ROCKTON. Those are already what a rider
would say, so they survive as published rather than being forced into a
number they do not have. Numbers are unpadded to match the rest of the
registry: routes.txt says `02`, a rider says 2.
Ids and names are from the City's own published GTFS routes.txt, which
sits under the same open-data licence as the realtime feed. Checked
against the live feed: all 30 route ids in service translate, none fall
through.
YRT encodes a branch as a two-digit suffix on the base route id. The Viva
table already depended on that scheme — `60102` is Viva Blue's B branch —
but local routes were passed through whole, on the reasoning that a YRT
route id is already the number on the bus. For a plain route it is. For a
branch it is not: a rider watching route 83's B-branch bus saw "8301".
A 15:41 EDT sample of the live feed found four branch ids in service.
Three are published in York Region's GTFS `routes.txt` and are named here
from it, unpadded to match the local routes beside them (the file says
`083A`; the registry has always preferred `105` to `005`):
10702 -> 107B KEELE
8301 -> 83A TRENCH
9002 -> 90B LESLIE
The fourth, `9101`, runs in the realtime feed but appears nowhere in
`routes.txt`. The pattern all but spells "91A" and that is precisely why
it is absent: the registry translates published names, it does not infer
them. It stays opaque until York Region publishes it.
Renamed YRT_VIVA_ROUTE_NAMES to YRT_ROUTE_NAMES, since the table is no
longer only about Viva.
Burlington's route ids are internal keys with a trap in them: `351` is
route 1 while `3510` is route 10, so no prefix trim is correct and the
table is a real join against the City's own published GTFS. Its licence is
the same 2011 municipal template as Mississauga — revocable, and the terms
must travel with the data, so licenseUrl is the terms document itself.
Plain http:// has no listener at all, so https is mandatory, not preferred.
Barrie and YRT ship on. Both operators invite acceptance of their licence
through a form, and both licences also state that use is itself acceptance
("By using Barrie Transit Data, you agree to be bound by these terms";
"Your Use of Information indicates Your acceptance"). This deployment
relies on that clause, and both entries record it so the reasoning does not
have to be re-derived.
Nine Ontario feeds now, all verified live against their upstreams.
York Region's feed is first-party infrastructure under an open licence that
permits commercial use and does not even require attribution. It is the one
feed that fills the gap north of Toronto, and it is switched off anyway.
The operator asks that its licence be accepted through a form collecting a
name, company, phone and email. No credential is issued and nothing gates
the endpoint — the licence body says use is itself acceptance — but a form
asking for contact details is close enough to the registration this registry
refuses that a person should decide it, not a poller. Barrie is off for the
same reason. Both are one line from on.
Viva route ids get labels: YRT publishes `601` where every station sign says
Viva Blue. The numeric local routes already read correctly and carry no
entry, so the table is six lines, not 125.
No vehicle in this feed carries a bearing; heading comes from observed
movement, as with Durham.
Guelph's feed publishes the internal database key of a route (`2991`)
where the rider sees the number on the bus (`1`), which would have put
"Route 2991" on the label a person reads. So a feed may now carry its own
`routeLabel`, the same way it may carry its own `modeWords`. Guelph's table
is its 33 routes, taken from the City's own published GTFS; an id missing
from it renders unchanged, because a raw id you can compare against the bus
beats a blank. Every other feed is identity and carries no table.
Kingston is catalogued by the City against this exact URL. Its licence PDF
link is pinned to the canonical location rather than the legacy path that
currently redirects there — Durham's entry records what happens when such a
redirect is eventually dropped.
Barrie ships REGISTERED AND OFF. The City offers its licence for acceptance
behind a click-through gate, and nobody on this project has accepted it.
Whether to accept is a person's decision, not an assumption a poller should
make, so the feed is visible, documented, and one line from enabled.
Three keyless GTFS-Realtime feeds around Toronto, each verified live: all
vehicles carry real coordinates, and each is polled only near its own
region rather than across the Golden Horseshoe.
The licences differ in ways the registry now records rather than flattens:
Hamilton PRESCRIBES its attribution wording, so the credit is that exact
sentence; Durham makes credit optional and its catalogued licence PDF is
dead after a site relaunch, so licenseUrl points at the live page where the
v.1.0 text was actually read; Mississauga's licence is revocable and asks
that the terms travel with the data, so licenseUrl is the terms themselves.
All three are bus-only and set no route hint, so their modes default rather
than resolve. Durham publishes no bearing on any vehicle — heading comes
from observed movement, as with any feed that omits it.
GO Transit / Metrolinx stays out: its feed requires an API key.
A Toronto 501 and a Helsinki 1010 are both the `tram` mode. Only one of
them is called a tram by the people who ride it, and the detection label
is documented as "what a person standing at the stop would call it".
So the mode stays shared — same silhouette, same speed limits — and a
registry entry may carry `modeWords` naming the modes it runs. Everything
without an override falls through to the shared table, so this renames one
word for one operator and nothing else.
Threading the feed to the detection metric needed a resolver in queries.js,
since the cached-contact path has only the entry. It answers from live feeds
first, so a feed disabled mid-session stops supplying vocabulary.
TTC BusTime publishes keyless GTFS-Realtime VehiclePositions with real
coordinates for the surface network: ~1,180 buses and streetcars with
bearing and per-vehicle timestamps. Registering it is one registry row.
The subway is deliberately absent. BusTime covers surface vehicles only
and the subway publishes no positions, so Lines 1/2/4 are left out rather
than inferred from schedules — the same rule that keeps NYCT out.
Route numbers establish the mode: the three-digit 500-series is streetcar,
as are the four Blue Night routes on streetcar trackage. A leading 5 alone
would misclassify bus routes 50-59, so the rule is anchored and tested.
Licensing is the portal-wide Open Government Licence - Toronto; the CKAN
record for the BusTime NVAS dataset leaves its own licence field blank,
and the terms note says so rather than overstating the grant.
Refs #70
Supersedes #218 by @spcpza, ported onto the current component layout with fixes from review.
Verified before merge: format, package boundaries, unit suite, build, and the browser tracking suite; a browser harness for the change where one applies; adversarial review applied.
Contributor: @spcpza. His original commit is the base of this branch, and his own re-port onto main (2026-09-15, `5b69011`) was consulted while finishing the height work. That port reads every vehicle's surface from the app's shared ground-floor system (`cachedGroundFloor` / `warmGroundFloor` / `GROUND_FLOOR_LIFT_M`) rather than sampling the scene per layer, keeps a vehicle hidden until its cell answers, warms a bounded batch per poll on a timer instead of per frame, and renders at the ellipsoid above the altitude gate. That is the sounder primitive — one floor system for the app, with the rooftop-versus-street question already settled against a datum-correct DEM prior — and this branch's height work was rebuilt on it, replacing a bespoke cell cache, ground prior and rooftop rule written here. Credit for that direction is his.
Future work (not in this PR), both written up in docs/CURRENT-STATE.md:
- Readable route names. The realtime feeds carry only `route_id`; `route_short_name` / `route_long_name` live in each operator's static GTFS `routes.txt`, and headsigns in `trips.txt`. Doing it properly wants a server-side metadata loader beside the transit proxy with fixed registered static sources, bounded indexes built once and persisted, a slow conditional refresh, and a versioned route dictionary the browser caches. Shipped here: ids are never rewritten, and a cut one carries an ellipsis instead of reading as a typo.
- Below-ground rendering. Subway vehicles are drawn on the surface directly above themselves, because the feeds carry no depth and a train nobody can see is not worth drawing; the card says so plainly. Real treatment wants OSM tunnel geometry per city and a ghosted sprite tethered to the surface. The owner recorded it and left it.
Round four and after (2026-09-15). The owner tested round three under FLIR and NVG, and then asked for the design to be taken further: elegant in performance, beautiful on camera, trails with a cache for public transport, Cesium-native, and a core other moving layers can adopt. Codex Astra reviewed the plan (it confirmed the speed fault: a glide started from the drawn position at receipt time but spanned the gap between report times, so with MBTA reporting every ~10 s against a 15 s poll every second glide ran at 1.5x; it rejected a queued-segment patch), then wrote the design, then implemented it in commit-sized passes with adversarial review between them and real-browser harness runs by the integrator.
What this PR now contains beyond the contributor's original port:
- Delayed playback in a contact-agnostic core (`src/data/contactPlayback.js`): each vehicle keeps a typed, bounded fix history and a display clock that runs at one second per second through it, a feed-derived lag behind real time, holding at its newest fix rather than extrapolating and never sprinting to catch up. Marker, course, motion word, card and trail all read the same displayed sample. The clock has an explicit time-offset and seek API so a rewind is a later feature, not a rewrite; flights and vessels are untouched.
- A selected-vehicle trail (`src/data/contactTrailRenderer.js`, `src/layers/transit/trails.js`): the retained history drawn with batched Cesium primitives, clipped at display time so the head is exactly the sprite, faded by age, in the mode colour (white under sensor presets), on the same surface path as the marker.
- Bounded proxy-side history (`server/providers/transitHistory.js`): fifteen minutes and at most 128 fixes per vehicle with hard byte ceilings, MBTA only for now (opt-in per feed), served by a small endpoint that never fetches upstream, so a selected trail survives a browser reload while the server keeps running.
- Scene work bounded to what is visible and moving, an occluder, cached DETECT candidates, holds on the render loop only while something visible moves.
- Sensor presets: white-hot haloed sprites under NVG/FLIR/noir (black-hot inverts), saturated mode colours under CRT, per-mode bracket tiers in every theme, the selected card keeping its mode colour.
- A discriminating browser harness (`scripts/qa-transit.mjs`) that samples postRender frames, reads the real card anchor, and measures fleet budgets at 800 and 3,000 vehicles.
Not built, recorded as follow-ups: curated route display names (lowest priority per the owner), fleet-wide tails, route snapping, a rewind UI, adoption of the playback core by flights and vessels.
Verified at merge (2026-09-15 morning): unit suite 3,896 passing, build clean, format and boundary gates clean, tracking regression 109/109, browser harness keyless 200 pass / 0 fail (13 sensor checks unexercised because the live view offered fewer than six isolated sprites; every judged sprite passed), keyed on Google 3D tiles 198 pass with the per-frame allocation gate as the only failure. Follow-ups recorded: the keyed build's playback sampler still allocates per frame (CPU and frame time stay within budget; the 3,000-vehicle metro view sits at a 27.6 ms frame interval against a 25 ms budget), the shared detection overlay's per-paint allocation, friendlier route names, fleet-wide tails, a rewind UI, and adopting the playback core for vessels and flights.
Co-authored-by: spcpza <[email protected]>
Cliff Flood