Live transit

A live window into Oslo.

Pick a vehicle in each tab. Keep your own view, share the live feed.

Preparing the live view…

Connections are counted from real socket events. Each tab loads its starting fleet with an HTTP query; live updates share the WebSocket.

Try it across tabs

Choose Open a second tab. Both boards receive the live feed; the connection display counts the actual upstream sockets. Choose a different vehicle in each tab: the position plot follows your selection, and Apollo caches stay local.

Once both tabs are receiving data, choose Compare per-tab connections. The boards keep working while each tab opens its own connection. Switch back to sharing to see them join one connection again.

Keep the tabs in the same browser profile. If SharedWorker is unavailable, the example reports local execution instead of claiming the connection is shared.

How it works

Entur publishes vehicle changes through GraphQL over WebSocket. The Spinetab Apollo link shares matching subscription operations; Apollo still owns each tab’s cache and React renders its own board. The HTTP link handles the current-state query.

The core integration is a split link:

import { graphqlWs } from "spinetab/graphql-ws";
import { spinetabSplit } from "spinetab/apollo";

const endpoint = graphqlWs(
  "wss://api.entur.io/realtime/v2/vehicles/subscriptions",
  {
    anonymous: true,
    connectionParams: {
      headers: { "ET-Client-Name": "your-company-your-app" },
    },
  },
);

const link = spinetabSplit(client, endpoint, http, {
  reconcile: ({ signal }) => refresh(signal),
});

Here client is the Spinetab browser client, http is Apollo’s HTTP link and refresh refreshes and merges the current fleet after uncertain delivery. Vehicle batches contain changes, so this example merges reports by vehicle ID and timestamp. A reconnect alone cannot restore changes missed while offline. The complete source includes the query, snapshot merge, cleanup and comparison instrumentation.

For your own app, start with the Astro setup and Apollo recipe. Use your own truthful Entur application identifier and retain the data attribution.

About the feed

The board shows recent bus reports across Oslo and its surroundings, ordered by line and vehicle. It is a vehicle board, not departures from a particular stop. Scroll the list to explore the fleet; it keeps up to 80 recent reports. Coverage depends on the operators publishing data to Entur, so this is not a complete inventory of every bus in Oslo. Destinations, delays and stop status come from the same live feed. Delays are reported in seconds; missing timings are labelled unavailable. Stop names are not supplied by this feed. Each tab starts with a current-state query, then merges live changes; sharing the feed does not replicate Apollo caches. Times are the provider’s reported times in Oslo; update frequency varies by vehicle. An empty or quiet board does not mean the connection failed. The coordinate plot centres on the selected vehicle, with north up and a fixed scale. Nearby markers show other reports from the board; vehicles outside the view are clipped. Coordinates and bearings are reports, not predictions or arrival estimates.

Data is provided by Entur under the Norwegian Licence for Open Government Data (NLOD). This example filters the feed and presents selected fields. Read Entur’s terms of service before reusing it.