Leaky bucket (feature request) about utils-decorators HOT 6 OPEN

Hi, is this exactly what's throttleAsync is for?

from utils-decorators.

The use case is rather to limit calls to a remote API which allows for, say, 10 calls per second.

With throttleAsync there is no way to set the desired throughput, and setting the limit to 10 would only lead to rejections from the API if one of the calls take less than a second to complete.

from utils-decorators.

Maybe I am missing something. For a given limit, the throttleAsync decorator will invoke the decorated method just for the allowed amount of times. Once one of the promises will be resolved the next call in the queue will be invoked. This basically means that there won't be any rejections. Once the amount of calls will reach to the limit the new calls will be queued.

https://vlio20.github.io/utils-decorators/#throttleAsync

from utils-decorators.

Apologies if that wasn't clear 😅 Here's some pseudocode which might help a bit:

getTwitterData(): Promise<Response> {
  return await fetch("https://api.twitter.com/tweets/20")
  //some api endpoints that allows 10 calls per second
  //this resolves only in a few milliseconds
}

// this ensures that only 10 calls per second are made, should return 200 OK
@leakyBucket({
  timeSpanMs: 1000,
  allowedCalls: 10,
})
getTwitterData(): Promise<Response> {…} 

// this does not prevent more than 10 calls per second, subsequent calls will return 429 Too Many Requests
@throttleAsync(10)
getTwitterData(): Promise<Response> {…} 

from utils-decorators.

Good point, throttleAsync will prevent that the number of concurrent call will acceed the specified number of calls it totally ignores the time perspective.

Great use-case. I would love to add it to the library.

from utils-decorators.

@vpctorr would you like to submit a PR?

from utils-decorators.