graphql-ws / client / ClientOptions
client.ClientOptions
Configuration used for the GraphQL over WebSocket client.
- connectionParams
- generateID
- isFatalConnectionProblem
- jsonMessageReplacer
- jsonMessageReviver
- keepAlive
- lazy
- on
- onNonLazyError
- retryAttempts
- retryWait
- url
- webSocketImpl
• Optional connectionParams: Record<string, unknown> | () => Record<string, unknown> | Promise<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.
• 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
▸ (): string
string
• Optional isFatalConnectionProblem: (errOrCloseEvent: unknown) => 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
▸ (errOrCloseEvent): boolean
| Name | Type |
|---|---|
errOrCloseEvent |
unknown |
boolean
• 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.
• 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.
• Optional keepAlive: 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
• Optional lazy: boolean
Controls when should the connection be established.
false: Establish a connection immediately. UseonNonLazyErrorto handle errors.true: Establish a connection on first subscribe and close on last unsubscribe. Use the subscription sink'serrorto handle errors.
default true
• 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.
• Optional onNonLazyError: (errorOrCloseEvent: unknown) => 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 inretryAttempts). - An
Error: some internal issue has occured, all internal errors are fatal by nature.
default console.error
▸ (errorOrCloseEvent): void
| Name | Type |
|---|---|
errorOrCloseEvent |
unknown |
void
• 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:
1002: Protocol Error1011: Internal Error4400: Bad Request4401: Unauthorizedtried subscribing before connect ack4409: Subscriber for <id> already existsdistinction is very important4429: Too many initialisation requests
These events are reported immediately and the client will not reconnect.
default 5
• Optional retryWait: (retries: number) => 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
▸ (retries): Promise<void>
| Name | Type |
|---|---|
retries |
number |
Promise<void>
• 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.
• 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.