fenecdb for the browser

The engine as WebAssembly, and the client that drives it: @fenecdb/web on npm, and the fenec-web bundle of each release.

| File | What it is | | --- | --- | | fenec.js | the client: the module's glue, persistence and the sync layer, a dependency-free ES module that imports and re-exports the two below | | builder.js, http.js | the query builder and the HTTP client | | client.js | the two of them, and nothing that could pull in the engine, sync or storage: @fenecdb/web/client, 6 KB brotli in an app's bundle | | fenec.d.ts, client.d.ts | their types | | schema.js, schema.d.ts | a schema declared in code, Drizzle's way: @fenecdb/web/schema | | fenec.wasm | the engine with every index and the schema check: 183 KB brotli | | collate/ | the collation data the module fetches beside it, a chunk a group of scripts |

import { Fenec } from './fenec.js';
const db = await Fenec.open('./fenec.wasm');

A page whose queries run on a server never fetches the module -- only Fenec.open does -- and can import the client entry, which holds nothing of the engine, the sync layer or storage (8 KB brotli bundled through @fenecdb/web, 6 through it):

import { connect } from '@fenecdb/web/client';   // or './client.js', beside builder.js and http.js
const db = connect('https://db.example.com', { token });
const rows = await db.from('docs').where('year', '>=', 2024).limit(10).rows();
const stop = db.live(db.from('docs').where('done', false), render);   // again after a write it may read

From npm the files are the same. A page serves the module and collate/ as static files and hands Fenec.open the module's URL -- with Vite, say:

import { Fenec } from '@fenecdb/web';
import wasm from '@fenecdb/web/fenec.wasm?url';
const db = await Fenec.open(wasm, { collation: '/collate/' });   // collate/ copied into public/

Node reads both out of the package:

import { readFile } from 'node:fs/promises';
import { Fenec } from '@fenecdb/web';
const file = (path) => readFile(new URL(import.meta.resolve(`@fenecdb/web/${path}`)));
const db = await Fenec.open(await file('fenec.wasm'), {
  collation: (name) => file(`collate/${name}.bin`),
});

Which package. A page whose queries run on a server imports @fenecdb/web/client and loads no module; a database in the page, or a synced replica, loads fenec.wasm, with every index and the schema check. A page that needs fewer indexes can build its own module with them alone -- make wasm FEATURES="text sorted", FEATURES=none for none, and SCHEMA=0 leaves the schema check out (7.5 KB brotli) -- and the file is the same either way: a store one build wrote opens in the other. Without the graph a near measures every vector, as exact does; without the text index a match throws a FenecError naming it.

Serve .wasm as application/wasm, compressed once at build time. Full reference: https://fenecdb.com/docs/javascript