Introduction

Prool is a library that provides programmatic HTTP testing instances for Ethereum. It is designed to be used in testing environments (e.g. Vitest) where you need to interact with an Ethereum server instance (e.g. Execution Node, 4337 Bundler, Indexer, etc) over HTTP or WebSocket.

Prool contains a set of pre-configured instances that can be used to simulate Ethereum server environments, being:

• Local Execution Nodes: anvil, tempo
• ERC-4337 Bundler Nodes: alto

You can also create your own custom instances by using the Instance.define function.

Table of Contents

• Install
• Getting Started
  • Anvil (Execution Node)
  • Alto (ERC-4337 Bundler Node)
• Reference
  • Server.create
  • Instance.define
  • Pool.create
  • Pool.define
  • Vitest

Install

npm i prool

pnpm add prool

bun i prool

Getting Started

Anvil (Execution Node)

Requirements

curl -L https://foundry.paradigm.xyz | bash   # Install Foundry

Usage

import { Instance, Server } from 'prool'

const server = Server.create({
  instance: Instance.anvil(),
})

await server.start() 
// Instances accessible at:
// "http://localhost:8545/1"
// "http://localhost:8545/2"
// "http://localhost:8545/3"
// "http://localhost:8545/n"

Tempo (Execution Node)

Requirements

curl -L https://tempo.xyz/install | bash   # Install Tempo

Usage

import { Instance, Server } from 'prool'

const server = Server.create({
  instance: Instance.tempo(),
})

await server.start() 
// Instances accessible at:
// "http://localhost:8545/1"
// "http://localhost:8545/2"
// "http://localhost:8545/3"
// "http://localhost:8545/n"

Alto (Bundler Node)

Requirements

• @pimlico/alto: npm i @pimlico/alto

Usage

import { Instance, Server } from 'prool'

const executionServer = Server.create({
  instance: Instance.anvil(),
  port: 8545
})
await executionServer.start() 
// Instances accessible at:
// "http://localhost:8545/1"
// "http://localhost:8545/2"
// "http://localhost:8545/3"
// "http://localhost:8545/n"

const bundlerServer = Server.create({
  instance: (key) => Instance.alto({
    entrypoints: ['0x0000000071727De22E5E9d8BAf0edAc6f37da032'],
    rpcUrl: `http://localhost:8545/${key}`,
    executorPrivateKeys: ['0x...'],
  })
})
await bundlerServer.start()
// Instances accessible at:
// "http://localhost:3000/1" (→ http://localhost:8545/1)
// "http://localhost:3000/2" (→ http://localhost:8545/2)
// "http://localhost:3000/3" (→ http://localhost:8545/3)
// "http://localhost:3000/n" (→ http://localhost:8545/n)

Reference

Server.create

Creates a server for a keyed instance proxy or an exclusive lease pool.

Usage

import { Instance, Server } from 'prool'

const executionServer = Server.create({
  instance: Instance.anvil(),
})
await executionServer.start() 
// Instances accessible at:
// "http://localhost:8545/1"
// "http://localhost:8545/2"
// "http://localhost:8545/3"
// "http://localhost:8545/n"
// "http://localhost:8545/n/start"
// "http://localhost:8545/n/stop"
// "http://localhost:8545/n/restart"
// "http://localhost:8545/healthcheck"

Endpoints:

• /:key: Proxy to instance at key.
• /:key/start: Start instance at key.
• /:key/stop: Stop instance at key.
• /:key/restart: Restart instance at key.
• /healthcheck: Healthcheck endpoint.

A lease pool exposes POST /acquire and POST /release/:token instead:

import { Instance, Pool, Server } from 'prool'

const pool = Pool.create({
  instance: Instance.anvil(),
  limit: 2,
  async reset(instance) {
    // Reset application state before reuse.
  },
})
const server = Server.create({ pool })
await server.start()

API

| Name | Description | Type | | ---------- | -------------------------------------------------------- | --------------------------------------- | | instance | Instance for the server. | Instance \| (key: number) => Instance | | limit | Number of instances that can be instantiated in the pool | number | | pool | Exclusive lease pool. | LeasePool | | host | Host for the server. | string | | port | Port for the server. | number | | returns | Server | Server.Server |

Instance.define

Creates an instance definition, that can be used with Server.create or Pool.define.

Usage

import { Instance } from 'prool'

const foo = Instance.define((parameters: FooParameters) => {
 return {
   name: 'foo',
   host: 'localhost',
   port: 3000,
   async start() {
     // ...
   },
   async stop() {
     // ...
   },
 }
})

API

| Name | Description | Type | | ------- | -------------------- | ------------------ | | fn | Instance definition. | DefineInstanceFn | | returns | Instance. | Instance |

Pool.create

Creates a bounded pool of exclusively leased instances. Acquisitions wait in order when every instance is busy.

If reset fails, the instance is destroyed and the release rejects before the slot becomes available again.

Usage

import { Instance, Pool } from 'prool'

const pool = Pool.create({
  instance: Instance.anvil(),
})
const lease = await pool.acquire()
try {
  await fetch(lease.instance.url)
} finally {
  await lease.release()
}
await pool.close()

API

| Name | Description | Type | | ---------- | ---------------------------------------- | ------------ | | instance | Instance to lease. | Instance | | limit | Maximum concurrent leases. Defaults to half the available logical CPUs. | number | | reset | Resets an instance before its next lease. | (instance: Instance) => Promise<void> \| void | | returns | Exclusive lease pool. | LeasePool |

Pool.define

Defines a pool of instances. Instances can be started, cached, and stopped against an identifier.

Usage

import { Instance, Pool } from 'prool'

const pool = Pool.define({
 instance: Instance.anvil(),
})
const instance_1 = await pool.start(1)
const instance_2 = await pool.start(2)
const instance_3 = await pool.start(3)

API

| Name | Description | Type | | ---------- | -------------------------------------------------------- | ---------- | | instance | Instance for the pool. | Instance | | limit | Number of instances that can be instantiated in the pool | number | | returns | Pool. | Pool |

Vitest

Server.setup starts one lazy keyed proxy for a Vitest project. Each worker can use Server.get to address, reset, or restart its own instance. Pool.setup eagerly starts one direct instance per worker instead.

Config

import { defineConfig } from 'vitest/config'

export default defineConfig({
  test: {
    globalSetup: './test/setup.global.ts',
    setupFiles: './test/setup.ts',
  },
})

Lazy Server

import { Instance } from 'prool'
import { Server } from 'prool/vitest'
import type { TestProject } from 'vitest/node'

declare module 'vitest' {
  export interface ProvidedContext {
    anvil: Server.Context
  }
}

export default Server.setup({
  instance: Instance.anvil(),
  setup(server, project: TestProject) {
    project.provide('anvil', server)
  },
})

import { Server } from 'prool/vitest'
import { inject } from 'vitest'

const anvil = Server.get(inject('anvil'))
await anvil.reset({ signal: AbortSignal.timeout(30_000) })

export const rpcUrl = anvil.url

reset destroys only that worker's instance, and its next request starts a fresh one. restart retains the pooled instance and endpoint.

Eager Pool

import { Instance } from 'prool'
import { Pool } from 'prool/vitest'
import type { TestProject } from 'vitest/node'

declare module 'vitest' {
  export interface ProvidedContext {
    rpcUrls: readonly string[]
  }
}

export default Pool.setup({
  instance: Instance.anvil(),
  setup(instances, project: TestProject) {
    project.provide(
      'rpcUrls',
      instances.map((instance) => instance.url),
    )
  },
})

import { Pool } from 'prool/vitest'
import { inject } from 'vitest'

export const rpcUrl = Pool.get(inject('rpcUrls'))

Authors

• @jxom (jxom.eth, Twitter)
• @tmm (awkweb.eth, Twitter)

License

MIT License