Shouldn't we do the same for RecoverableExceptionInterface also?

Shouldn't we do the same for RecoverableExceptionInterface also?

Indeed, I forgot about that one.

I don't think we need to do the same for RecoverableExceptionInterface, because when an exception implements that interface, it signals that the operation should be retried. With this new behavior, we're explicitly avoiding retries for exceptions that implement UnrecoverableExceptionInterface, so the distinction remains meaningful.

@santysisi RecoverableExceptionInterface enforces the retry regardless of what retry strategy says – f.e. max_retries may be reached, but if recoverable exception was thrown, the message must still be retried.

RecoverableExceptionInterface can also override the retry delay set by retry strategy – so ignoring it here would mean always using retry delay from the retry strategy, which would again be unintended/unexpected behavior.

Got it, thanks for the explanation! I'll make that change. 🙌