Is there a way to specify task concurrency programatically only within a single flow? Something like this

@task(max_concurrency=100)  # concurrency limit defined in task decorator
def fn(i: int)
    print(i)

@flow
def main():
    map_range = list(range(1000))
    fn.map(map_range)

or this

@task
def fn(i: int)
    print(i)

@flow
def main():
    map_range = list(range(1000))
    fn.map(map_range, max_concurrency=100)  # concurrency limit defined at task submission

In either case, we'd only want 100 instances of

fn

running at a given time. I know we can set concurrency limits in the cloud and task tags but this would be much nicer for us if possible!

hi @Erik Amundson - have you tried the context manager for concurrency?

Oh that looks perfect, thanks!