Skip to content

Latest commit

 

History

History
309 lines (201 loc) · 8.97 KB

client.clientoptions.md

File metadata and controls

309 lines (201 loc) · 8.97 KB
Raw

graphql-ws / client / ClientOptions

Interface: ClientOptions

client.ClientOptions

Configuration used for the GraphQL over WebSocket client.

Table of contents

Properties

Methods

Properties

connectionParams

• Optional connectionParams: Record<string, unknown> | () => undefined | Record<string, unknown> | Promise<undefined | Record<string, unknown>>

Optional parameters, passed through the payload field with the ConnectionInit message, that the client specifies when establishing a connection with the server. You can use this for securely passing arguments for authentication.

If you decide to return a promise, keep in mind that the server might kick you off if it takes too long to resolve! Check the connectionInitWaitTimeout on the server for more info.

Throwing an error from within this function will close the socket with the Error message in the close event reason.

disablePong

• Optional disablePong: boolean

Disable sending the PongMessage automatically.

Useful for when integrating your own custom client pinger that performs custom actions before responding to a ping, or to pass along the optional pong message payload. Please check the readme recipes for a concrete example.

jsonMessageReplacer

• Optional jsonMessageReplacer: JSONMessageReplacer

An optional override for the JSON.stringify function used to serialize outgoing messages from this client. Useful for serializing custom datatypes out to the client.

jsonMessageReviver

• Optional jsonMessageReviver: JSONMessageReviver

An optional override for the JSON.parse function used to hydrate incoming messages to this client. Useful for parsing custom datatypes out of the incoming JSON.

keepAlive

• Optional keepAlive: number

The timout between dispatched keep-alive messages, naimly server pings. Internally dispatches the PingMessage type to the server and expects a PongMessage in response. This helps with making sure that the connection with the server is alive and working.

Timeout countdown starts from the moment the socket was opened and subsequently after every received PongMessage.

Note that NOTHING will happen automatically with the client if the server never responds to a PingMessage with a PongMessage. If you want the connection to close, you should implement your own logic on top of the client. A simple example looks like this:

import { createClient } from 'graphql-ws';

let activeSocket, timedOut;
createClient({
  url: 'ws://i.time.out:4000/after-5/seconds',
  keepAlive: 10_000, // ping server every 10 seconds
  on: {
    connected: (socket) => (activeSocket = socket),
    ping: (received) => {
      if (!received) // sent
        timedOut = setTimeout(() => {
          if (activeSocket.readyState === WebSocket.OPEN)
            activeSocket.close(4408, 'Request Timeout');
        }, 5_000); // wait 5 seconds for the pong and then close the connection
    },
    pong: (received) => {
      if (received) clearTimeout(timedOut); // pong is received, clear connection close timeout
    },
  },
});

default 0

lazy

• Optional lazy: boolean

Controls when should the connection be established.

  • false: Establish a connection immediately. Use onNonLazyError to handle errors.
  • true: Establish a connection on first subscribe and close on last unsubscribe. Use the subscription sink's error to handle errors.

default true

lazyCloseTimeout

• Optional lazyCloseTimeout: number

How long should the client wait before closing the socket after the last oparation has completed. This is meant to be used in combination with lazy. You might want to have a calmdown time before actually closing the connection. Kinda' like a lazy close "debounce".

default 0 // close immediately

on

• Optional on: Partial<Object>

Register listeners before initialising the client. This way you can ensure to catch all client relevant emitted events.

The listeners passed in will always be the first ones to get the emitted event before other registered listeners.

retryAttempts

• Optional retryAttempts: number

How many times should the client try to reconnect on abnormal socket closure before it errors out?

The library classifies the following close events as fatal:

  • 1011: Internal Error
  • 4400: Bad Request
  • 4401: Unauthorized tried subscribing before connect ack
  • 4406: Subprotocol not acceptable
  • 4409: Subscriber for <id> already exists distinction is very important
  • 4429: Too many initialisation requests

These events are reported immediately and the client will not reconnect.

default 5

url

• url: string | () => string | Promise<string>

URL of the GraphQL over WebSocket Protocol compliant server to connect.

If the option is a function, it will be called on every WebSocket connection attempt. Returning a promise is supported too and the connecting phase will stall until it resolves with the URL.

A good use-case for having a function is when using the URL for authentication, where subsequent reconnects (due to auth) may have a refreshed identity token in the URL.

webSocketImpl

• Optional webSocketImpl: unknown

A custom WebSocket implementation to use instead of the one provided by the global scope. Mostly useful for when using the client outside of the browser environment.

Methods

generateID

▸ Optional generateID(): string

A custom ID generator for identifying subscriptions.

The default generates a v4 UUID to be used as the ID using Math as the random number generator. Supply your own generator in case you need more uniqueness.

Reference: https://gist.github.com/jed/982883

Returns

string

isFatalConnectionProblem

▸ Optional isFatalConnectionProblem(errOrCloseEvent): boolean

Check if the close event or connection error is fatal. If you return true, the client will fail immediately without additional retries; however, if you return false, the client will keep retrying until the retryAttempts have been exceeded.

The argument is either a WebSocket CloseEvent or an error thrown during the connection phase.

Beware, the library classifies a few close events as fatal regardless of what is returned. They are listed in the documentation of the retryAttempts option.

default Non close events

Parameters

Name Type
errOrCloseEvent unknown

Returns

boolean

onNonLazyError

▸ Optional onNonLazyError(errorOrCloseEvent): void

Used ONLY when the client is in non-lazy mode (lazy = false). When using this mode, the errors might have no sinks to report to; however, to avoid swallowing errors, consider using onNonLazyError, which will be called when either:

  • An unrecoverable error/close event occurs
  • Silent retry attempts have been exceeded

After a client has errored out, it will NOT perform any automatic actions.

The argument can be a websocket CloseEvent or an Error. To avoid bundling DOM types, you should derive and assert the correct type. When receiving:

  • A CloseEvent: retry attempts have been exceeded or the specific close event is labeled as fatal (read more in retryAttempts).
  • An Error: some internal issue has occured, all internal errors are fatal by nature.

default console.error

Parameters

Name Type
errorOrCloseEvent unknown

Returns

void

retryWait

▸ Optional retryWait(retries): Promise<void>

Control the wait time between retries. You may implement your own strategy by timing the resolution of the returned promise with the retries count. retries argument counts actual connection attempts, so it will begin with 0 after the first retryable disconnect.

default Randomised exponential backoff

Parameters

Name Type
retries number

Returns

Promise<void>