Analytics Engine

Write event data points with env.NAME.writeDataPoint and read the latest back.

Configuration

"analytics_engine_datasets": ["EVENTS"]

A dataset takes a bare name only. It has no id and needs no provisioning: it exists from its first writeDataPoint.

Usage

export default {
  async fetch(request, env, ctx) {
    // The standard Workers call: returns nothing, and is not awaited.
    env.EVENTS.writeDataPoint({
      indexes: ["signup"],
      blobs: [request.headers.get("cf-ipcountry") ?? "unknown"],
      doubles: [1],
    });
    return new Response("recorded\n", { status: 202 });
  },
} satisfies ExportedHandler<Env>;

writeDataPoint is the Workers Analytics Engine call, so this code runs on Workers unchanged. Reading data back is where the two differ. On Workers you query a dataset with the separate SQL API; Sproutboat adds a query() method to the binding, so a handler can see what it wrote with no second API:

// Sproutboat only: this call does not exist on Workers.
const { count, rows } = await env.EVENTS.query({ limit: 50 });
// rows: [{ timestamp, indexes, blobs, doubles }], newest first

Methods