Canonical Code Style Twitter Follow

Wraps postgres API in a pg compatible API.

import postgres from 'postgres';
import { createPostgresBridge } from 'postgres-bridge';
const PostgresBridge = createPostgresBridge(postgres);
// pg.Pool Configuration
const configuration = {
 host: 'localhost',
 user: 'database-user',
 max: 20,
 idleTimeoutMillis: 30000,
 connectionTimeoutMillis: 2000,
};
const pool = new PostgresBridge(configuration);
const connection = await pool.connect();
await pg.query('SELECT 1ドル::text as name', ['foo']);

postgres is leaner/faster implementation of PostgreSQL protocol in Node.js than pg. However, postgres API is very different from the more broadly adopted pg client.

This package allows to adopt postgres without going through a painful migration. In particular, this compatibility layer has been designed to allow adoption of postgres using Slonik PostgreSQL client.

postgres-bridge was primarily developed to enable postgres use with Slonik PostgreSQL client. However, the scope has since been expanded to support several projects. It is now used in production by a handful real-world applications.

Known incompatibilities:

• Implicit query pooling is not implemented, i.e. You must use pool.connect()
• connection.processID not implemented
• pool._pulseQueue not implemented
• callback (CPS) interface is not implemented (use promises instead)
• providing configuration using environment variables is not implemented
• bring your own promise is not implemented

Please submit PR if you require additional compatibility.

A basic benchmark shows no overhead as a result of using postgres-bridge:

pg query: 880ms
postgres query: 867ms
postgres-bridge query: 871ms

While these benchmarks do not show meaningful difference between pg and postgres, in production, in a large codebase, we noticed average response time improve by 30%. It means that in real-world scenarios, postgres overhead is significantly lesser than that of pg.

Running postgres-bridge tests requires having a local PostgreSQL instance.

The easiest way to setup a temporary instance for testing is using Docker, e.g.

docker run --rm -it -e POSTGRES_HOST_AUTH_METHOD=trust -p 5432:5432 postgres