alonmiz / poll-until-promise Goto Github PK

Hi,

Would you accept a PR for a new option like maxBackoffInterval? For example, if you specified interval = 1, backoffFactor = 2, and maxBackoffInterval = 10, you'll get retries at 1, 2, 4, 8, 10, 10, 10, 10, 10, ... and so on until success or timeout reached.

Backoff retry is obviously great but growing without bound is probably not wanted in practice.

Thanks!

For example, if you use backoffFactor = 2 and interval = 1000, the first retry will take place after 2000ms. I would expect the first retry to take place after 1000ms and the second retry to take place after 2000ms.

If that is intentional, a usage note in the docs would be nice.

const { waitFor } = require('poll-until-promise');
function A() {
return waitFor(() => {
  if (Math.random() >= 0.5) {
    throw new Error('try again');
  } else {
    console.log('all good')
  }
})
}

async function B() {
 await A()
}

this will not give stacktrace of B

Hi @AlonMiz,

Curious if you'd be interested in a PR for the following:

• maxAttempts -- Sometimes, the semantics of "attempt this operation X times" are preferable over "attempt this operation until X time has passed". You could get close to this behavior by setting timeout = interval * numDesiredAttempts, but it's not exact/guaranteed and it also gets more complicated if you're using a backoffFactor or similar. Note that this option is additive and wouldn't replace the timeout; if maxAttempts is passed, then polling stops when either the timeout is reached OR attempts are exhausted.

• fail-fast when stopOnFailure===false -- The stopOnFailure setting is all-or-nothing; that is, either ANY error will stop the loop, or no errors will. It would be nice to have some dedicated error that could be conditioned to stop the loop even when stopOnFailure is false.

// Usage in execution function:
throw new AbortError(myError); // <--- stops polling immediately, myError is reported