I have the following code:
notificationsWsSubject.pipe(
filter((socket): socket is Socket => !!socket),
switchMap(socket => fromEvent<Socket.DisconnectReason>(socket, 'disconnect')),
tap(() => wsConnectedSubject.next(false)),
filter(reason => (['ping timeout', 'transport close', 'transport error'] as Socket.DisconnectReason[]).includes(reason)),
switchMap(() => signedInObservable),
switchMap(user => forkJoin([of(user), from(getNotificationsWebsocketTicket())])),
).subscribe(values => {
// Connect with websocket
}, error => {
// Throw error to user
})
The general flow:
disconnect event in socket.io-clientsignedInObservablegetNotificationsWebsocketTicket() (and creating an Observable from it using from()subscribe(value => ... )My problem is that I would like to retry the from(getNotificationsWebsocketTicket()) in case it fails, with a delay of 5 seconds between each failure.
Only after 3 retries I want the entire main observable to fail.
something like:
from(getNotificationsWebsocketTicket()).pipe(delayedRetry(3, 5000))
Is that possible?
Yeah, you can do this:
from(getNotificationsWebsocketTicket()).pipe(
retryWhen(e$ => e$.pipe(
take(3),
delay(5000)
))
);
retryWhen can be a bit of an interesting operator to get your head around. e$ is an observable that is managed by the retryWhen operator. Anytime the source errors, retryWhen emits that error to e$.
What happens next is up to the observable your lambda returns:
Errors are errors,
Completes are completes,
BUT emissions (next) are a cue to retry