The same view. A new place to run it.

Your browser.
Your live database.

CouchDB-inspired map/reduce, running right here. An AssemblyScript view maps your documents and reduces each group in WebAssembly. No server required.

Write a view once. Run it here or in Python.

The metrics module accepts documents with a string metric and a finite numeric value. Its map emits the metric as a key. Its reducer maintains count, sum, minimum, and maximum; finalization computes the mean. Other document shapes are ignored by the map.

Numbers use double precision in both environments. Keep large identifiers as strings. Strings must be valid Unicode; each module request is limited to 256 KiB of encoded JSON.

import { BrowserDatabase } from "./portable/client.js";

const db = await BrowserDatabase.connect();
await db.add({ metric: "latency_ms", value: 42 });
const rows = await db.rows();
console.log(rows[0].result);
db.close();

from foldlitedb import Database, View

view = View.bundled("metrics")
with Database("metrics.fold", view) as db:
    db.add({"metric": "latency_ms", "value": 42})
    row = list(db.rows())[0]
    print(row.result)

Your documents are the source of truth

Try the live stream, add a few measurements, then select Rebuild & verify. A fresh in-memory database replays every retained document through the same module and compares its complete view states and results. Existing views keep their cursors.

Exported JSON includes document identities, sources, positions, and payloads. Importing it again is an identical replay: each retained document counts once. Changing an existing identity or position fails the whole batch. Use those same records to rebuild the native database.