Delaying requests instead of failing

[ClementJ18]

Hello,
I have this minimal example

from pyrate_limiter import Duration, Limiter, Rate

GLOBAL_THROTTLE = Duration.MINUTE * 5
GLOBAL_LIMITER = Limiter(
    [
        # request stuff slowly, like a human
        Rate(1, Duration.SECOND * 1),
        # take breaks when requesting stuff, like a human
        Rate(40, GLOBAL_THROTTLE),
    ],
    max_delay=GLOBAL_THROTTLE + 500,
)

global_limiter_decorator = GLOBAL_LIMITER.as_decorator()


def mapping(*args, **kwargs):
    return ("moddb", 1)


def ratelimit(func):
    return global_limiter_decorator(mapping)(func)

@ratelimit
def ratelimited_function(x):
    print(x)

def test_ratelimit():
    for x in range(41):
        ratelimited_function(x)

if __name__ == "__main__":
    test_ratelimit()

What I expect to happen is that a number will be printed every second until it hits 39 (which would be the 40th call). On the 41st call the library will check to see how much time is required to sleep, which should be 5 minutes. The library then sees that the max delay is set to 5 minutes + 0.5 seconds which is more. Therefore, the library sleeps for 5 minutes and then retries the 41st call which print successfully.

What I instead see is that on the 41st call the following error happens

                Re-acquiring with delay expected to be successful,
                if it failed then either clock or bucket is probably unstable

Traceback (most recent call last):
  File ".\test.py", line 33, in <module>
    test_ratelimit()
  File ".\test.py", line 30, in test_ratelimit
    ratelimited_function(x)
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 329, in wrapper
    accquire_ok = self.try_acquire(name, weight)
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 300, in try_acquire
    result = self.handle_bucket_put(bucket, item)
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 265, in handle_bucket_put
    return _handle_result(acquire)  # type: ignore
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 245, in _handle_result
    return self.delay_or_raise(bucket, item)
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 234, in delay_or_raise
    return _handle_reacquire(re_acquire)
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 174, in _handle_reacquire
    self._raise_bucket_full_if_necessary(bucket, item)
  File "C:\Users\Clement\Documents\GitHub\moddb\env\lib\site-packages\pyrate_limiter\limiter.py", line 134, in _raise_bucket_full_if_necessary
    raise BucketFullException(item, bucket.failing_rate)
pyrate_limiter.exceptions.BucketFullException: Bucket for item=moddb with Rate limit=40/5.0m is already full

Could someone help me understand where in my assumptions I am going wrong.

[ClementJ18]

In the end I decided to move to my own implementation.